CN117033882A - Vehicle data processing method, device, computer equipment and storage medium - Google Patents
Vehicle data processing method, device, computer equipment and storage medium Download PDFInfo
- Publication number
- CN117033882A CN117033882A CN202211039706.XA CN202211039706A CN117033882A CN 117033882 A CN117033882 A CN 117033882A CN 202211039706 A CN202211039706 A CN 202211039706A CN 117033882 A CN117033882 A CN 117033882A
- Authority
- CN
- China
- Prior art keywords
- sample
- vehicle
- data
- weight parameter
- backward
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000003672 processing method Methods 0.000 title claims abstract description 25
- 238000000034 method Methods 0.000 claims abstract description 40
- 238000012545 processing Methods 0.000 claims description 153
- 238000000611 regression analysis Methods 0.000 claims description 60
- 238000012549 training Methods 0.000 claims description 41
- 238000004590 computer program Methods 0.000 claims description 30
- 238000000605 extraction Methods 0.000 claims description 23
- 239000011159 matrix material Substances 0.000 description 36
- 230000006870 function Effects 0.000 description 17
- 238000004364 calculation method Methods 0.000 description 12
- 238000004891 communication Methods 0.000 description 9
- 238000010276 construction Methods 0.000 description 8
- 238000010586 diagram Methods 0.000 description 8
- 230000008569 process Effects 0.000 description 6
- 238000006073 displacement reaction Methods 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 4
- 238000011478 gradient descent method Methods 0.000 description 4
- 206010039203 Road traffic accident Diseases 0.000 description 3
- 238000013473 artificial intelligence Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000013500 data storage Methods 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 238000012804 iterative process Methods 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 230000003068 static effect Effects 0.000 description 2
- OKTJSMMVPCPJKN-UHFFFAOYSA-N Carbon Chemical compound [C] OKTJSMMVPCPJKN-UHFFFAOYSA-N 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 230000001413 cellular effect Effects 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 238000007418 data mining Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 229910021389 graphene Inorganic materials 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 239000004973 liquid crystal related substance Substances 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Computational Mathematics (AREA)
- Pure & Applied Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Computing Systems (AREA)
- Databases & Information Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Algebra (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Health & Medical Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Operations Research (AREA)
- Probability & Statistics with Applications (AREA)
- Medical Informatics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Traffic Control Systems (AREA)
Abstract
The vehicle data processing method based on the traffic field comprises the following steps: acquiring first sample traveling data of a first sample vehicle and second sample traveling data of a second sample vehicle which has a traveling sequence with the first sample vehicle on a sample road; acquiring sample pavement data of each sample sensor related to a first sample vehicle; determining a first association relationship according to the first sample driving data and the second sample driving data; determining a second association relation according to the pavement data of each sample; based on a first initial weight parameter of a second sample vehicle and a second initial weight parameter of each sample sensor, combining the first and second sample driving data, the first and second association relations and each sample road surface data, and performing cross iterative weight parameter adjustment to obtain a first target weight parameter and each second target weight parameter; the first target weight parameter and the second target weight parameter are used to determine a distance between vehicles in which a driving order exists. The method can improve the accuracy of inter-vehicle distance prediction.
Description
Technical Field
The present application relates to the field of computer technology, and in particular, to a vehicle data processing method, apparatus, computer device, storage medium, and computer program product.
Background
The vehicle-road cooperation is a safe, efficient and environment-friendly road traffic system which is formed by adopting advanced wireless communication, new generation internet and other technologies, implementing vehicle-vehicle and vehicle-road dynamic real-time information interaction in an omnibearing manner, developing vehicle active safety control and road cooperation management on the basis of full-time idle dynamic traffic information acquisition and fusion, fully realizing effective cooperation of human-vehicle roads, ensuring traffic safety and improving traffic efficiency.
In a conventional vehicle-road cooperation scheme, the distance between vehicles is generally predicted by feature data collected by the vehicles. However, there may be a repeated portion of the feature data collected by different vehicles, and the same feature data is repeatedly superimposed during the distance prediction process, resulting in inaccurate predicted distances between vehicles.
Disclosure of Invention
In view of the foregoing, it is desirable to provide a vehicle data processing method, apparatus, computer device, computer-readable storage medium, and computer program product that can improve accuracy.
The application provides a vehicle data processing method, which comprises the following steps:
acquiring first sample driving data of a first sample vehicle and second sample driving data of a second sample vehicle; the first sample vehicle and the second sample vehicle have a driving sequence on a sample road;
acquiring sample pavement data acquired by a plurality of sample sensors related to the first sample vehicle for the sample road respectively;
determining a first association relationship between the first sample vehicle and the second sample vehicle according to the first sample driving data and the second sample driving data; determining a second association relationship among the plurality of sample sensors according to the plurality of sample pavement data;
acquiring a first initial weight parameter of the second sample vehicle and a second initial weight parameter of each sample sensor;
based on the first initial weight parameter and each second initial weight parameter, carrying out cross iteration weight parameter adjustment by combining the first sample driving data, the second sample driving data, each sample pavement data, the first association relationship and the second association relationship to obtain a first target weight parameter corresponding to the second sample vehicle and a second target weight parameter corresponding to each sample sensor; the first target weight parameter and the second target weight parameter are used to determine a distance between at least two vehicles for which a driving sequence exists.
In one embodiment, the performing, based on each of the second initial weight parameter, the forward initial predicted distance, and the backward initial predicted distance, an iterative process of the weight parameter in combination with the forward expected distance, the backward expected distance, each of the sample road surface data, and the second association relationship, to obtain a second target weight parameter corresponding to each of the sample sensors includes:
determining a global expected distance according to the forward expected distance and the backward expected distance; determining a global initial predicted distance according to the forward initial predicted distance and the backward initial predicted distance; and carrying out iterative processing on the weight parameters based on each second initial weight parameter and the global initial prediction distance and combining the global expected distance, each sample pavement data and the second association relation to obtain a second target weight parameter corresponding to each sample sensor.
The application also provides a vehicle data processing device, which comprises:
the sample driving data acquisition module is used for acquiring first sample driving data of the first sample vehicle and second sample driving data of the second sample vehicle; the first sample vehicle and the second sample vehicle have a driving sequence on a sample road;
A sample road surface data acquisition module, configured to acquire sample road surface data acquired by a plurality of sample sensors related to the first sample vehicle for the sample road respectively;
the association relation determining module is used for determining a first association relation between the first sample vehicle and the second sample vehicle according to the first sample driving data and the second sample driving data; determining a second association relationship among the plurality of sample sensors according to the plurality of sample pavement data;
a weight parameter acquisition module for acquiring a first initial weight parameter of the second sample vehicle and a second initial weight parameter of each sample sensor;
the weight parameter adjustment module is used for carrying out cross iteration weight parameter adjustment by combining the first sample driving data, the second sample driving data, the sample pavement data, the first association relationship and the second association relationship based on the first initial weight parameter and each second initial weight parameter to obtain a first target weight parameter corresponding to the second sample vehicle and a second target weight parameter corresponding to each sample sensor; the first target weight parameter and the second target weight parameter are used to determine a distance between at least two vehicles for which a driving sequence exists.
In one embodiment, the association determining module is further configured to perform feature extraction of a plurality of first dimensions based on the first sample driving data, so as to obtain first sample features of the first sample driving data in each first dimension; the plurality of first dimensions includes a vehicle travel dimension and a vehicle attribute dimension; performing feature extraction of the plurality of first dimensions based on the second sample travel data to obtain second sample features of the second sample travel data in each first dimension; and determining a first association relationship between the first sample vehicle and the second sample vehicle according to each first sample characteristic and each second sample characteristic.
In one embodiment, the association determining module is further configured to determine a sub-association between each of the first sample features and each of the second sample features; and determining a first association relationship between the first sample vehicle and the second sample vehicle according to each sub-association relationship.
In one embodiment, the association determining module is further configured to perform feature extraction of a plurality of second dimensions based on each sample pavement data, so as to obtain sample pavement features corresponding to each sample pavement data in each second dimension; the plurality of second dimensions includes a road surface travel dimension and a road surface attribute dimension; and determining a second association relationship among the plurality of sample sensors according to a plurality of sample pavement characteristics corresponding to each sample pavement data.
In one embodiment, the weight parameter adjustment module is further configured to determine an initial predicted distance between the first sample vehicle and the second sample vehicle based on the first initial weight parameter and each of the second initial weight parameters, in combination with the first sample travel data, the second sample travel data, and each of the sample road surface data; acquiring an expected distance between the first sample vehicle and the second sample vehicle, and performing iterative processing of weight parameters based on the first initial weight parameter and the initial predicted distance and combining the expected distance, the first sample driving data, the second sample driving data and the first association relation to obtain a first target weight parameter corresponding to the second sample vehicle; and carrying out iterative processing on the weight parameters based on each second initial weight parameter and the initial predicted distance and combining the expected distance, each sample pavement data and the second association relation to obtain a second target weight parameter corresponding to each sample sensor.
In one embodiment, the second sample vehicle comprises a forward sample vehicle and a backward sample vehicle of the first sample vehicle; the forward sample vehicle is located before the first sample vehicle in its travel position on the sample road, and the backward sample vehicle is located after the first sample vehicle in its travel position on the sample road; the second sample travel data includes forward sample travel data of the forward sample vehicle and backward sample travel data of the backward sample vehicle; the initial prediction distance comprises a forward initial prediction distance and a backward initial prediction distance between the forward sample vehicle and the backward sample vehicle and between the backward sample vehicle and the first sample vehicle respectively; the expected distance comprises a forward expected distance and a backward expected distance between the forward sample vehicle and the backward sample vehicle and the first sample vehicle respectively; the first association relationship comprises a forward association relationship and a backward association relationship between the forward sample vehicle and the backward sample vehicle and the first sample vehicle respectively; the first initial weight parameters include forward initial weight parameters of the forward sample vehicle and backward initial weight parameters of the backward sample vehicle; the first target weight parameters include forward target weight parameters of the forward sample vehicle and backward target weight parameters of the backward sample vehicle.
In one embodiment, the weight parameter adjustment module is further configured to determine a forward initial predicted distance between the first sample vehicle and the forward sample vehicle based on the forward initial weight parameter and each of the second initial weight parameters, in combination with the first sample travel data, the forward sample travel data, and each of the sample road surface data; and determining a backward initial prediction distance between the first sample vehicle and the backward sample vehicle by combining the first sample driving data, the backward sample driving data and the sample pavement data based on the backward initial weight parameters and each second initial weight parameter.
In one embodiment, the weight parameter adjustment module is further configured to perform iterative processing of weight parameters based on the forward initial weight parameter and the forward initial predicted distance, in combination with the forward expected distance, the first sample driving data, the forward sample driving data, and the forward association relationship, to obtain a forward target weight parameter corresponding to the forward sample vehicle; and carrying out iterative processing on the weight parameters based on the backward initial weight parameters and the backward initial predicted distance and combining the backward expected distance, the first sample driving data, the backward sample driving data and the backward association relation to obtain backward target weight parameters corresponding to the backward sample vehicle.
In one embodiment, the weight parameter adjustment module is further configured to perform, based on each of the second initial weight parameter, the forward initial predicted distance, and the backward initial predicted distance, and in combination with the forward expected distance, the backward expected distance, each of the sample pavement data, and the second association relationship, iterative processing of the weight parameter, to obtain a second target weight parameter corresponding to each of the sample sensors.
In one embodiment, the weight parameter adjustment module is further configured to determine a global expected distance according to the forward expected distance and the backward expected distance; determining a global initial predicted distance according to the forward initial predicted distance and the backward initial predicted distance; and carrying out iterative processing on the weight parameters based on each second initial weight parameter and the global initial prediction distance and combining the global expected distance, each sample pavement data and the second association relation to obtain a second target weight parameter corresponding to each sample sensor.
The application also provides a computer device comprising a memory and a processor, the memory storing a computer program, the processor implementing the following steps when executing the computer program:
Acquiring first sample driving data of a first sample vehicle and second sample driving data of a second sample vehicle; the first sample vehicle and the second sample vehicle have a driving sequence on a sample road; acquiring sample pavement data acquired by a plurality of sample sensors related to the first sample vehicle for the sample road respectively; determining a first association relationship between the first sample vehicle and the second sample vehicle according to the first sample driving data and the second sample driving data; determining a second association relationship among the plurality of sample sensors according to the plurality of sample pavement data; acquiring a first initial weight parameter of the second sample vehicle and a second initial weight parameter of each sample sensor; based on the first initial weight parameter and each second initial weight parameter, carrying out cross iteration weight parameter adjustment by combining the first sample driving data, the second sample driving data, each sample pavement data, the first association relationship and the second association relationship to obtain a first target weight parameter corresponding to the second sample vehicle and a second target weight parameter corresponding to each sample sensor; the first target weight parameter and the second target weight parameter are used to determine a distance between at least two vehicles for which a driving sequence exists.
The present application also provides a computer readable storage medium having stored thereon a computer program which when executed by a processor performs the steps of:
acquiring first sample driving data of a first sample vehicle and second sample driving data of a second sample vehicle; the first sample vehicle and the second sample vehicle have a driving sequence on a sample road; acquiring sample pavement data acquired by a plurality of sample sensors related to the first sample vehicle for the sample road respectively; determining a first association relationship between the first sample vehicle and the second sample vehicle according to the first sample driving data and the second sample driving data; determining a second association relationship among the plurality of sample sensors according to the plurality of sample pavement data; acquiring a first initial weight parameter of the second sample vehicle and a second initial weight parameter of each sample sensor; based on the first initial weight parameter and each second initial weight parameter, carrying out cross iteration weight parameter adjustment by combining the first sample driving data, the second sample driving data, each sample pavement data, the first association relationship and the second association relationship to obtain a first target weight parameter corresponding to the second sample vehicle and a second target weight parameter corresponding to each sample sensor; the first target weight parameter and the second target weight parameter are used to determine a distance between at least two vehicles for which a driving sequence exists.
The application also provides a computer program product comprising a computer program which, when executed by a processor, performs the steps of:
acquiring first sample driving data of a first sample vehicle and second sample driving data of a second sample vehicle; the first sample vehicle and the second sample vehicle have a driving sequence on a sample road; acquiring sample pavement data acquired by a plurality of sample sensors related to the first sample vehicle for the sample road respectively; determining a first association relationship between the first sample vehicle and the second sample vehicle according to the first sample driving data and the second sample driving data; determining a second association relationship among the plurality of sample sensors according to the plurality of sample pavement data; acquiring a first initial weight parameter of the second sample vehicle and a second initial weight parameter of each sample sensor; based on the first initial weight parameter and each second initial weight parameter, carrying out cross iteration weight parameter adjustment by combining the first sample driving data, the second sample driving data, each sample pavement data, the first association relationship and the second association relationship to obtain a first target weight parameter corresponding to the second sample vehicle and a second target weight parameter corresponding to each sample sensor; the first target weight parameter and the second target weight parameter are used to determine a distance between at least two vehicles for which a driving sequence exists.
The above-described vehicle data processing method, apparatus, computer device, storage medium, and computer program product determine a correlation between travel data acquired by a first sample vehicle and a second sample vehicle by acquiring first sample travel data of the first sample vehicle and second sample travel data of the second sample vehicle, the first sample vehicle and the second sample vehicle having a travel order on a sample road, to determine a first correlation between the first sample vehicle and the second sample vehicle based on the first sample travel data and the second sample travel data. Sample road surface data acquired by a plurality of sample sensors related to the first sample vehicle respectively for the sample road are acquired, so that a second association relationship among the plurality of sample sensors is determined according to the plurality of sample road surface data, and accordingly correlation among the plurality of sample sensors on the road surface data is determined.
Based on the first initial weight parameter of the second sample vehicle and the second initial weight parameter of each sample sensor, the first sample driving data, the second sample driving data, the pavement data of each sample, the first association relationship and the second association relationship are combined to carry out cross iterative weight parameter adjustment, and the weight calculation can be carried out by combining the correlation between the driving data of each vehicle and the correlation between the pavement data acquired by each sensor to obtain the first target weight parameter of the second sample vehicle and the second target weight parameter of each sample sensor, so that the problem of inaccurate weight parameters caused by repeated superposition calculation of the same data is avoided, and the accuracy of weight parameter calculation is effectively improved. The distance between at least two vehicles with the running sequence is determined by using the first target weight parameter and the second target weight parameter, so that the accuracy of the distance prediction between the vehicles can be effectively improved.
The application also provides a vehicle data processing method, which comprises the following steps:
acquiring first driving data of a first vehicle and second driving data of a second vehicle; the first vehicle and the second vehicle have a driving sequence on a target road;
acquiring target road surface data acquired by a plurality of sensors related to the first vehicle aiming at the target road respectively;
according to the driving sequence, acquiring a first target weight parameter obtained by pre-training corresponding to the second vehicle and a second target weight parameter obtained by pre-training corresponding to each sensor;
performing regression analysis processing based on the first driving data, the second driving data and the first target weight parameter to obtain first regression data;
carrying out regression analysis processing based on the pavement data and each second target weight parameter to obtain second regression data;
and determining the vehicle distance between the first vehicle and the second vehicle according to the first regression data and the second regression data.
In one embodiment, the second vehicle includes a forward vehicle and a backward vehicle of the first vehicle, the forward vehicle being before a travel position on a target road is at a travel position of the first vehicle, the backward vehicle being after the travel position on the target road is at a travel position of the first vehicle; the second traveling data includes forward traveling data of the forward vehicle and backward traveling data of the backward vehicle, the first target weight parameter includes a forward target weight parameter of the forward vehicle and a backward target weight parameter of the backward vehicle, the first regression data includes forward regression data of the forward vehicle and backward regression data of the backward vehicle, and the vehicle distance includes a forward distance between the first vehicle and the forward vehicle and a backward distance between the first vehicle and the backward vehicle.
In one embodiment, the performing regression analysis based on the first driving data, the second driving data, and the first target weight parameter to obtain first regression data includes:
performing regression analysis processing according to the first driving data, the forward driving data and the forward target weight parameter to obtain forward regression data; performing regression analysis processing according to the first driving data, the backward driving data and the backward target weight parameter to obtain backward regression data;
the determining the vehicle distance between the first vehicle and the second vehicle according to the first regression data and the second regression data comprises the following steps:
determining a forward distance between the first vehicle and the forward vehicle according to the forward regression data and the second regression data; and determining the backward distance between the first vehicle and the backward vehicle according to the backward regression data and the second regression data.
The application also provides a vehicle data processing device, which comprises:
the driving data acquisition module is used for acquiring first driving data of the first vehicle and second driving data of the second vehicle; the first vehicle and the second vehicle have a driving sequence on a target road;
The road surface data acquisition module is used for acquiring target road surface data acquired by a plurality of sensors related to the first vehicle aiming at the target road respectively;
the target weight acquisition module is used for acquiring a first target weight parameter obtained by pre-training corresponding to the second vehicle and a second target weight parameter obtained by pre-training corresponding to each sensor according to the driving sequence;
the processing module is used for carrying out regression analysis processing based on the first driving data, the second driving data and the first target weight parameter to obtain first regression data; carrying out regression analysis processing based on the pavement data and each second target weight parameter to obtain second regression data;
and the distance determining module is used for determining the vehicle distance between the first vehicle and the second vehicle according to the first regression data and the second regression data.
In one embodiment, the second vehicle includes a forward vehicle and a backward vehicle of the first vehicle, the forward vehicle being before a travel position on a target road is at a travel position of the first vehicle, the backward vehicle being after the travel position on the target road is at a travel position of the first vehicle; the second traveling data includes forward traveling data of the forward vehicle and backward traveling data of the backward vehicle, the first target weight parameter includes a forward target weight parameter of the forward vehicle and a backward target weight parameter of the backward vehicle, the first regression data includes forward regression data of the forward vehicle and backward regression data of the backward vehicle, and the vehicle distance includes a forward distance between the first vehicle and the forward vehicle and a backward distance between the first vehicle and the backward vehicle.
In one embodiment, the processing module is further configured to perform regression analysis processing according to the first driving data, the forward driving data, and the forward target weight parameter, to obtain forward regression data; performing regression analysis processing according to the first driving data, the backward driving data and the backward target weight parameter to obtain backward regression data;
the distance determining module is further configured to determine a forward distance between the first vehicle and the forward vehicle according to the forward regression data and the second regression data; and determining the backward distance between the first vehicle and the backward vehicle according to the backward regression data and the second regression data.
The application also provides a computer device comprising a memory and a processor, the memory storing a computer program, the processor implementing the following steps when executing the computer program:
acquiring first driving data of a first vehicle and second driving data of a second vehicle; the first vehicle and the second vehicle have a driving sequence on a target road; acquiring target road surface data acquired by a plurality of sensors related to the first vehicle aiming at the target road respectively; according to the driving sequence, acquiring a first target weight parameter obtained by pre-training corresponding to the second vehicle and a second target weight parameter obtained by pre-training corresponding to each sensor; performing regression analysis processing based on the first driving data, the second driving data and the first target weight parameter to obtain first regression data; carrying out regression analysis processing based on the pavement data and the second target weight parameter to obtain second regression data; and determining the vehicle distance between the first vehicle and the second vehicle according to the first regression data and the second regression data.
The present application also provides a computer readable storage medium having stored thereon a computer program which when executed by a processor performs the steps of:
acquiring first driving data of a first vehicle and second driving data of a second vehicle; the first vehicle and the second vehicle have a driving sequence on a target road; acquiring target road surface data acquired by a plurality of sensors related to the first vehicle aiming at the target road respectively; according to the driving sequence, acquiring a first target weight parameter obtained by pre-training corresponding to the second vehicle and a second target weight parameter obtained by pre-training corresponding to each sensor; performing regression analysis processing based on the first driving data, the second driving data and the first target weight parameter to obtain first regression data; carrying out regression analysis processing based on the pavement data and the second target weight parameter to obtain second regression data; and determining the vehicle distance between the first vehicle and the second vehicle according to the first regression data and the second regression data.
The application also provides a computer program product comprising a computer program which, when executed by a processor, performs the steps of:
acquiring first driving data of a first vehicle and second driving data of a second vehicle; the first vehicle and the second vehicle have a driving sequence on a target road; acquiring target road surface data acquired by a plurality of sensors related to the first vehicle aiming at the target road respectively; according to the driving sequence, acquiring a first target weight parameter obtained by pre-training corresponding to the second vehicle and a second target weight parameter obtained by pre-training corresponding to each sensor; performing regression analysis processing based on the first driving data, the second driving data and the first target weight parameter to obtain first regression data; carrying out regression analysis processing based on the pavement data and the second target weight parameter to obtain second regression data; and determining the vehicle distance between the first vehicle and the second vehicle according to the first regression data and the second regression data.
The vehicle data processing method, the device, the computer equipment, the storage medium and the computer program product acquire second running data of a first vehicle and a second vehicle which have running sequences on a target road, and first target weight parameters obtained by pre-training corresponding to the second vehicle, and perform regression analysis processing based on the first running data, the second running data and the first target weight parameters, so that the first target weight parameters are used as regression coefficients to carry out regression on the first running data and the second running data, and correlation between the vehicle distance to be solved and the running data of each vehicle, namely, first regression data is obtained. Acquiring target road surface data which are acquired by a plurality of sensors related to a first vehicle and respectively aiming at a target road, and respectively corresponding second target weight parameters which are obtained by pretraining each sensor, and carrying out regression analysis processing based on the road surface data and the second target weight parameters, so that the second target weight parameters are used as regression coefficients to carry out regression on the road surface data of the sensors, and obtaining the correlation between the vehicle distance to be solved and the data acquired by the sensors, namely second regression data. According to the correlation between the vehicle distance to be solved and the running data of each vehicle and the correlation between the vehicle distance to be solved and the data acquired by each sensor, the vehicle distance between the first vehicle and the second vehicle can be calculated more accurately.
Drawings
FIG. 1 is a diagram of an application environment for a vehicle data processing method in one embodiment;
FIG. 2 is a flow chart of a method of vehicle data processing in one embodiment;
FIG. 3 is a schematic diagram of an interface of a sample vehicle and a sample sensor in a sample roadway in one embodiment;
FIG. 4 is a schematic diagram of constructing a second correlation covariance matrix based on respective sample road surface data in one embodiment;
FIG. 5 is a flow chart illustrating adjustment of weighting parameters for cross-iteration in one embodiment;
FIG. 6 is a flow chart of a method of processing vehicle data in another embodiment;
FIG. 7 is a schematic diagram of an interface for determining forward and backward distances in one embodiment;
FIG. 8 is a flow chart of a method of processing vehicle data in another embodiment;
FIG. 9 is a block diagram of a vehicle data processing device in one embodiment;
FIG. 10 is a block diagram showing a configuration of a vehicle data processing apparatus in another embodiment;
FIG. 11 is an internal block diagram of a computer device in one embodiment.
Detailed Description
The present application will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present application more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the application.
The embodiment of the application can be applied to various scenes, including but not limited to cloud technology, artificial intelligence, intelligent traffic, auxiliary driving, automatic driving, data mining and the like. For example, intelligent transportation systems (IntelligentTraffic System, ITS) applied to the intelligent transportation field are also called intelligent transportation systems (Intellig entTra ns portation Sys tem), and intelligent vehicle-road coordination systems (IntelligentVehicle Infrastructure Cooperative Systems, IVICS), simply referred to as vehicle-road coordination systems.
The scheme provided by the embodiment of the application relates to a vehicle data processing method in the traffic field, and specifically is explained through the following embodiments.
The vehicle data processing method provided by the embodiment of the application can be applied to an application environment shown in fig. 1. Wherein a terminal 102 deployed on a first sample vehicle communicates with a server 104 over a network. The data storage system may store data that the server 104 needs to process. The data storage system may be integrated on the server 104 or may be located on the cloud or other servers. The terminal 102 and the server 104 may each individually perform the vehicle data processing method provided in the embodiment of the present application. The terminal 102 and the server 104 may also cooperate to perform the vehicle data processing method provided in the embodiments of the present application. When the terminal 102 and the server 104 cooperate to perform the vehicle data processing method provided in the embodiment of the present application, the terminal 102 acquires first sample travel data of the first sample vehicle and second sample travel data of the second sample vehicle; the first sample vehicle and the second sample vehicle have a driving sequence on the sample road; the terminal 102 acquires sample road surface data acquired by a plurality of sample sensors associated with the first sample vehicle for the sample road, respectively. The terminal 102 transmits the first sample travel data, the second sample travel data, and the sample road surface data to the server 104. The server 104 determines a first association relationship between the first sample vehicle and the second sample vehicle from the first sample travel data and the second sample travel data, and determines a second association relationship between the plurality of sample sensors from the plurality of sample road surface data. The server 104 obtains a first initial weight parameter of a second sample vehicle and a second initial weight parameter of each sample sensor, and performs cross iteration weight parameter adjustment by combining first sample driving data, second sample driving data, various sample pavement data, a first association relationship and a second association relationship based on the first initial weight parameter and each second initial weight parameter to obtain a first target weight parameter corresponding to the second sample vehicle and a second target weight parameter corresponding to each sample sensor; the first target weight parameter and the second target weight parameter are used to determine a distance between at least two vehicles for which a driving sequence exists. The terminal 102 may be, but is not limited to, various personal computers, notebook computers, smart phones, tablet computers, intelligent voice interaction devices, intelligent home appliances, car terminals, aircrafts, portable wearable devices, etc. The vehicle terminal is also called as a vehicle-mounted terminal, and refers to short for vehicle-mounted information entertainment products installed in a vehicle, and the vehicle can realize information communication between people and vehicles and between vehicles and the outside (vehicle and vehicle) functionally. The server 104 may be an independent physical server, a server cluster or a distributed system formed by a plurality of physical servers, or a cloud server providing cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDNs, basic cloud computing services such as big data and artificial intelligence platforms, and the like. The terminal 102 and the server 104 may be directly or indirectly connected through wired or wireless communication, and the present application is not limited herein.
It should be noted that the numbers of "plural" and the like mentioned in the respective embodiments of the present application each refer to the number of "at least two".
In one embodiment, as shown in fig. 2, there is provided a vehicle data processing method, in which a description is given by applying the method to a computer device (the computer device may be a terminal or a server in fig. 1), including the steps of:
step S202, acquiring first sample driving data of a first sample vehicle and second sample driving data of a second sample vehicle; the first sample vehicle and the second sample vehicle have a driving order on the sample road.
Wherein, the first sample vehicle and the second sample vehicle refer to vehicles needing distance prediction. The second sample vehicle includes at least one of a forward sample vehicle or a backward sample vehicle of the first sample vehicle.
The first sample travel data refers to travel data associated with the first sample vehicle and may include data of at least one dimension. The data of the at least one dimension includes at least one of data of the first sample vehicle in a vehicle travel dimension, or vehicle attribute data of the first sample vehicle in a vehicle attribute dimension.
The second sample travel data refers to travel data associated with a second sample vehicle and may include data of at least one dimension. The data of the at least one dimension includes at least one of data of the second sample vehicle in a vehicle travel dimension, or vehicle attribute data of the second sample vehicle in a vehicle attribute dimension.
Wherein the vehicle travel dimension includes at least one of a vehicle speed dimension, a vehicle distance dimension, or a vehicle POI (Point of Interest ) dimension.
The data in the vehicle traveling dimension includes at least one of a vehicle speed of the vehicle itself, a forward and backward vehicle speed of the vehicle, a road surface image collected by the vehicle, a forward and backward vehicle distance, a left and right vehicle distance, or vehicle POI data. The vehicle attribute data in the vehicle attribute dimension includes at least one of a vehicle type, a vehicle length, a height, a width, a vehicle core number, a vehicle actual number, a vehicle maximum speed per hour, or a vehicle displacement.
Specifically, the first sample vehicle performs data acquisition on the vehicle driving dimension to obtain corresponding first sample driving data. Further, the first sample vehicle performs data acquisition in the dimension of the vehicle attribute, namely, acquires attribute data of the vehicle itself, and obtains the vehicle attribute data. The related data and the vehicle attribute data acquired in the vehicle running dimension are taken as first sample running data of a first sample vehicle.
And the second sample vehicle performs data acquisition on the vehicle driving dimension to obtain corresponding second sample driving data. Further, the second sample vehicle performs data acquisition in the dimension of the vehicle attribute, namely, acquires attribute data of the vehicle itself, and obtains the vehicle attribute data. And taking the related data and the vehicle attribute data acquired in the vehicle driving dimension as second sample driving data of a second sample vehicle.
The computer device may determine that there is a first sample vehicle and a second sample vehicle in a driving order on the sample road and acquire first sample driving data of the first sample vehicle and second sample driving data of the second sample vehicle.
In step S204, sample road surface data acquired by a plurality of sample sensors related to the first sample vehicle for the sample road respectively is acquired.
The sample sensor is a sensor arranged in a sample road and is used for collecting data of vehicles in the sample road.
The plurality of sample sensors associated with the first sample vehicle include a target sensor closest to a travel position of the first sample vehicle and at least one of a forward sensor or a backward sensor corresponding to the target sensor. The position of the forward sensor on the sample path is before the position of the target sensor, and the position of the backward sensor on the sample path is after the position of the target sensor.
Or, the plurality of sample sensors associated with the first sample vehicle include a target sensor within a preset range of the first sample vehicle and at least one of a forward sensor or a backward sensor corresponding to the target sensor.
For example, the plurality of sample sensors associated with the first sample vehicle includes a target sensor that the first sample vehicle is passing by and at least one of a forward sensor or a backward sensor corresponding to the sample sensor that the first sample vehicle is passing by.
The sample road surface data refers to data obtained by data acquisition of a sample road and vehicles in the sample road by a sample sensor. The sample road surface data may include data of at least one dimension including at least one of road surface travel data of a road surface travel dimension in which the vehicle travels in the sample road, or road surface attribute data of the sample road in a road surface attribute dimension.
The road surface traveling data in the road surface traveling dimension includes at least one of a speed of a passing vehicle, a forward-backward vehicle speed of the vehicle, a forward-backward vehicle distance, a left-right vehicle distance, or vehicle POI (Point of interest) data.
The road surface attribute data in the road surface attribute dimension includes at least one of an image of a road surface of a road segment, an average vehicle speed, speed limit data, overspeed data, traffic accident image, traffic flow data, waiting time period, or road segment POI information.
Specifically, a sensor in a sample road collects vehicle running data of a passing vehicle in a road surface running dimension to obtain road surface running data of the vehicle in the road surface running dimension. The sensor in the sample road can also collect data of the sample road in the dimension of the road surface attribute to obtain the road surface attribute data.
For example, the sample sensor collects vehicle running data of a first sample vehicle passing by, obtains road surface running data corresponding to the first sample vehicle, and collects road surface attribute data of a sample road when the first sample vehicle passes by. At least one of road surface running data or road surface attribute data is used as sample road surface data corresponding to the sample sensor. And each sample sensor performs the same data acquisition processing on the passing vehicle, so that sample pavement data corresponding to each sample sensor can be obtained.
The computer device determines a target sensor closest to a travel position of the first sample vehicle in the sample road and at least one of a forward sensor or a backward sensor corresponding to the target sensor. The target sensor and the forward sensor are taken as a plurality of sample sensors related to the first sample vehicle, or the target sensor and the backward sensor are taken as a plurality of sample sensors related to the first sample vehicle, or the target sensor, the forward sensor and the backward sensor are taken as a plurality of sample sensors related to the first sample vehicle.
In this embodiment, the computer device determines a target sensor in the sample road within a preset range of the first sample vehicle, and determines at least one of a forward sensor or a backward sensor corresponding to the target sensor.
The computer device may obtain sample road surface data collected by each sample sensor, resulting in a plurality of sample road surface data.
As shown in fig. 3, the first sample vehicle and the second sample vehicle have the vehicle-mounted terminals mounted thereon. In a sample road, a vehicle-mounted terminal of a first sample vehicle performs data acquisition on the first sample vehicle to obtain first sample driving data. And the vehicle-mounted terminal of the second sample vehicle acquires data of the second sample vehicle to obtain second sample driving data. And the sensor on the sample road performs data acquisition on the sample road to obtain sample road surface data. As the first sample vehicle passes sensor B, sensor B and the forward sensor of sensor B (i.e., sensor C), and the backward sensor of sensor B (i.e., sensor a) are taken as the corresponding plurality of sample sensors of the first sample vehicle.
Step S206, determining a first association relationship between the first sample vehicle and the second sample vehicle according to the first sample driving data and the second sample driving data; and determining a second association relationship among the plurality of sample sensors according to the plurality of sample pavement data.
Wherein the first association represents a data correlation or a characteristic correlation of data between first sample travel data of the first sample vehicle and second sample travel data of the second sample vehicle, i.e. a correlation on travel data or a characteristic correlation on travel data of the first sample vehicle and the second sample vehicle. The second correlation represents a data correlation or a characteristic correlation of data between sample road surface data of the plurality of sample sensors, that is, a correlation on collected road surface data or a characteristic correlation on road surface data of the plurality of sample sensors.
The first association and the second association may be represented by a correlation matrix, which may specifically be represented by a covariance matrix. The first association may be characterized by a covariance matrix between the first sample travel data and the second sample travel data. The second association may be characterized by a covariance matrix between the plurality of sample road surface data.
Specifically, the computer device may determine, from the first sample travel data and the second sample travel data, a data correlation existing between the first sample travel data and the second sample travel data, and use the data correlation as the first association relationship between the first sample vehicle and the second sample vehicle. The computer device may determine, from the plurality of sample road surface data, a data correlation that exists between the plurality of sample road surface data, and use the data correlation as the second association relationship between the plurality of sample sensors.
In this embodiment, the computer device may construct a first correlation matrix according to the first sample driving data and the second sample driving data, and characterize a first association relationship between the first sample vehicle and the second sample vehicle by using the first correlation matrix. The computer device may construct a second correlation matrix from the plurality of sample road surface data, and characterize a second association between the plurality of sample sensors by the second correlation matrix.
For example, a first correlation covariance matrix is constructed from the first sample travel data and the second sample travel data, and a first association relationship between the first sample vehicle and the second sample vehicle is represented by the first correlation covariance matrix. Constructing a second correlation covariance matrix according to the plurality of sample pavement data, and representing a second association relationship among the plurality of sample sensors through the second correlation covariance matrix.
Step S208, a first initial weight parameter of the second sample vehicle and a second initial weight parameter of each sample sensor are acquired.
Wherein the first initial weight parameter refers to an initial value of a weight parameter corresponding to the second sample vehicle, which is adjusted in an iteration. The second initial weight parameter refers to an initial value of the weight parameter corresponding to the sample sensor, which is adjusted in the iteration.
Specifically, the computer device may initialize the weight parameter of the second sample vehicle and the weight parameter of each sample sensor to obtain a first initial weight parameter corresponding to the second sample vehicle and a second initial weight parameter corresponding to each sample sensor.
Step S210, based on the first initial weight parameter and each second initial weight parameter, carrying out cross iteration weight parameter adjustment by combining the first sample driving data, the second sample driving data, each sample road surface data, the first association relationship and the second association relationship to obtain a first target weight parameter corresponding to the second sample vehicle and a second target weight parameter corresponding to each sample sensor; the first target weight parameter and the second target weight parameter are used to determine a distance between at least two vehicles for which a driving sequence exists.
The first target weight parameter refers to a target value of the weight parameter corresponding to the second sample vehicle, and the target value is generated by adjusting an initial value in iteration. The second target weight parameter refers to a target value of the weight parameter corresponding to the sample sensor, which is generated by adjusting the initial value in the iteration.
Specifically, the computer equipment performs cross iteration weight parameter adjustment based on the first initial weight parameter and each second initial weight parameter and combines the first sample driving data, the second sample driving data, the pavement data of each sample, the first association relationship and the second association relationship to obtain an intermediate weight parameter of each iteration. The intermediate weight parameters include a first intermediate weight parameter corresponding to the second sample vehicle and a second intermediate weight parameter corresponding to each sample sensor.
In the first iteration, the computer equipment adjusts the first initial weight parameter and each second initial weight parameter based on the first initial weight parameter and each second initial weight parameter and combines the first sample driving data, the second sample driving data, the pavement data of each sample, the first association relationship and the second association relationship to obtain a first intermediate weight parameter of the second sample vehicle and a second intermediate weight parameter corresponding to each sample sensor.
And from the second iteration, taking the first intermediate weight parameter of the second sample vehicle obtained in the last iteration as the first initial weight parameter of the second sample vehicle in the next iteration, taking the second intermediate weight parameter corresponding to each sample sensor obtained in the last iteration as the second intermediate weight parameter corresponding to each sample sensor in the next iteration, and adjusting the weight parameters according to the processing mode until the iteration stopping condition is met, so as to obtain the first target weight parameter corresponding to the second sample vehicle and the second target weight parameter corresponding to each sample sensor.
The iteration stop condition may be satisfied in that the number of iterations reaches a preset number of iterations, or a difference in distance between a predicted distance and an expected distance generated in the iterations is smaller than a difference threshold. The desired distance refers to the true distance between the first sample vehicle and the second sample vehicle, used as a distance tag in a cross iteration.
The adjustment of the weight parameters of the cross iteration refers to the adjustment of the first initial weight parameter and each second initial weight parameter in the same iteration, the next iteration is performed after the adjustment of the first initial weight parameter and each second initial weight parameter is completed in the same iteration, and the like. The method comprises the steps of entering the next iteration after obtaining a first intermediate weight parameter and each second intermediate weight parameter of the previous iteration, and adjusting the first intermediate weight parameter and each second intermediate weight parameter obtained in the previous iteration in the next iteration until the iteration stopping condition is met, so that a first target weight parameter and each second target weight parameter are obtained. The order of adjustment of the first initial weight parameter and the second initial weight parameter in the same iteration is not limited.
The first target weight parameter and each second target weight parameter are used to determine a distance between at least two vehicles for which a driving sequence exists.
In this embodiment, feature extraction is performed on the first sample driving data and the second sample driving data to obtain a first sample feature corresponding to the first sample vehicle and a second sample feature corresponding to the second sample vehicle; respectively extracting the characteristics of each sample pavement data to obtain the sample pavement characteristics corresponding to each sample sensor; based on the first initial weight parameter and each second initial weight parameter, carrying out cross iteration weight parameter adjustment by combining the first sample characteristic, the second sample characteristic, each sample pavement characteristic, the first association relation and the second association relation to obtain a first target weight parameter corresponding to the second sample vehicle and a second target weight parameter corresponding to each sample sensor; the first target weight parameter and the second target weight parameter are used to determine a distance between at least two vehicles for which a driving sequence exists.
In this embodiment, first sample traveling data of a first sample vehicle and second sample traveling data of a second sample vehicle are acquired, and traveling orders of the first sample vehicle and the second sample vehicle exist on a sample road to determine a first association relationship between the first sample vehicle and the second sample vehicle based on the first sample traveling data and the second sample traveling data, thereby determining correlations between the traveling data acquired by the first sample vehicle and the second sample vehicle, respectively. Sample road surface data acquired by a plurality of sample sensors related to the first sample vehicle respectively for the sample road are acquired, so that a second association relationship among the plurality of sample sensors is determined according to the plurality of sample road surface data, and accordingly correlation among the plurality of sample sensors on the road surface data is determined.
Based on the first initial weight parameter of the second sample vehicle and the second initial weight parameter of each sample sensor, the first sample driving data, the second sample driving data, the pavement data of each sample, the first association relationship and the second association relationship are combined to carry out cross iterative weight parameter adjustment, and the weight calculation can be carried out by combining the correlation between the driving data of each vehicle and the correlation between the pavement data acquired by each sensor to obtain the first target weight parameter of the second sample vehicle and the second target weight parameter of each sample sensor, so that the problem of inaccurate weight parameters caused by repeated superposition calculation of the same data is avoided, and the accuracy of weight parameter calculation is effectively improved. The distance between at least two vehicles with the running sequence is determined by using the first target weight parameter and the second target weight parameter, so that the accuracy of the distance prediction between the vehicles can be effectively improved.
In one embodiment, the second sample vehicle comprises at least one of a forward sample vehicle or a backward sample vehicle of the first sample vehicle; the second sample travel data includes at least one of forward sample travel data of a forward sample vehicle or backward sample travel data of a backward sample vehicle; the first association relationship comprises at least one of a forward association relationship between the first sample vehicle and a forward sample vehicle or a backward association relationship between the first sample vehicle and a backward sample vehicle; the first initial weight parameter includes at least one of a forward initial weight parameter of the forward sample vehicle or a backward initial weight parameter of the backward sample vehicle; the first target weight parameter includes at least one of a forward target weight parameter of the forward sample vehicle, or a backward target weight parameter of the backward sample vehicle.
In one embodiment, the second sample vehicle comprises a forward sample vehicle and a backward sample vehicle of the first sample vehicle, the forward sample vehicle traveling on the sample road before the first sample vehicle traveling on the sample road, the backward sample vehicle traveling on the sample road after the first sample vehicle traveling on the sample road; the second sample travel data includes forward sample travel data of a forward sample vehicle and backward sample travel data of a backward sample vehicle; the first association relationship comprises a forward association relationship between the first sample vehicle and a forward sample vehicle and a backward association relationship between the first sample vehicle and a backward sample vehicle; the first initial weight parameters comprise forward initial weight parameters of the forward sample vehicle and backward initial weight parameters of the backward sample vehicle; the first target weight parameters include forward target weight parameters of the forward sample vehicle and backward target weight parameters of the backward sample vehicle.
Specifically, first sample travel data of a first sample vehicle, forward sample travel data of a forward sample vehicle of the first sample vehicle, and backward sample travel data of a backward sample vehicle of the first sample vehicle are acquired; the forward sample vehicle travel position on the sample road is before the first sample vehicle travel position, and the backward sample vehicle travel position on the sample road is after the first sample vehicle travel position. Sample road surface data acquired by a plurality of sample sensors respectively aiming at a sample road is acquired, wherein the plurality of sample sensors are related to a first sample vehicle. Determining a forward association relationship between the first sample vehicle and the forward sample vehicle according to the first sample driving data and the forward sample driving data; determining a backward association relationship between the first sample vehicle and the backward sample vehicle according to the first sample driving data and the backward sample driving data; and determining a second association relationship among the plurality of sample sensors according to the plurality of sample pavement data.
A forward initial weight parameter of the forward sample vehicle, a backward initial weight parameter of the backward sample vehicle, and a second initial weight parameter of each sample sensor are obtained. Based on the forward initial weight parameter, the backward initial weight parameter and each second initial weight parameter, carrying out cross iteration weight parameter adjustment by combining the first sample driving data, the forward sample driving data, the backward sample driving data, each sample pavement data, the forward association relationship, the backward association relationship and the second association relationship to obtain a forward target weight parameter corresponding to the forward sample vehicle, a backward target weight parameter corresponding to the backward sample vehicle and a second target weight parameter corresponding to each sample sensor; the forward target weight parameter, the backward target weight parameter, and the second target weight parameter are used to determine distances between at least three vehicles for which a driving order exists.
In one embodiment, determining a first association between the first sample vehicle and the second sample vehicle based on the first sample travel data and the second sample travel data comprises:
performing feature extraction of a plurality of first dimensions based on the first sample driving data to obtain first sample features of the first sample driving data in each first dimension; the plurality of first dimensions includes a vehicle travel dimension and a vehicle attribute dimension; performing feature extraction of a plurality of first dimensions based on the second sample driving data to obtain second sample features of the second sample driving data in each first dimension; and determining a first association relationship between the first sample vehicle and the second sample vehicle according to each first sample characteristic and each second sample characteristic.
The vehicle driving dimension comprises at least one dimension of vehicle speed, front and rear vehicle speed, road surface images acquired by the vehicle, front and rear vehicle distance, left and right vehicle distance or vehicle POI. The vehicle attribute dimension includes at least one dimension of a vehicle type, a vehicle length, a height, a width, a vehicle core number, a vehicle payload number, a vehicle maximum speed per hour, or a vehicle displacement.
Specifically, the first sample running data comprises data in a vehicle running dimension and a vehicle attribute dimension, the computer equipment performs feature extraction of the vehicle running dimension and the vehicle attribute dimension on the first sample running data to obtain vehicle running features corresponding to the first sample running data in the vehicle running dimension and vehicle attribute features corresponding to the first sample running data in the vehicle attribute dimension. The computer device uses the vehicle running characteristics and the vehicle attribute characteristics corresponding to the first sample running data as first sample characteristics of the first sample running data in the corresponding first dimension, namely as first sample characteristics of the first sample vehicle in the corresponding first dimension.
The second sample driving data comprises data on a vehicle driving dimension and a vehicle attribute dimension, and the computer equipment performs feature extraction on the vehicle driving dimension and the vehicle attribute dimension on the second sample driving data to obtain vehicle driving features corresponding to the second sample driving data on the vehicle driving dimension and vehicle attribute features corresponding to the second sample driving data on the vehicle attribute dimension. The computer equipment takes the vehicle running characteristics and the vehicle attribute characteristics corresponding to the second sample running data as second sample characteristics of the second sample running data in the corresponding second dimension, namely second sample characteristics of the second sample vehicle in the corresponding second dimension.
The computer device may construct a first correlation matrix from each first sample feature and each second sample feature, and characterize a first association between the first sample vehicle and the second sample vehicle by the first correlation matrix.
In this embodiment, the computer device may determine a sub-association relationship between a first sample feature and each second sample feature, and traverse each first sample feature to obtain a sub-association relationship between each first sample feature and each second sample feature, and construct a first association relationship between the first sample vehicle and the second sample vehicle according to the sub-association relationships. The sub-association between the first sample feature and the second sample feature may be characterized by a covariance, and the first correlation matrix may be a correlation covariance matrix.
In this embodiment, the vehicle driving features include at least one of a vehicle speed feature, a forward vehicle speed feature, a backward vehicle speed feature, a road surface image feature collected by the vehicle, a forward-backward vehicle distance feature, a left-right vehicle distance feature, or a vehicle POI (Point of interest) feature. The vehicle attribute features include at least one of a vehicle type feature, a vehicle volume feature, a vehicle core man count feature, a vehicle payload number feature, a vehicle highest speed per hour feature, or a vehicle displacement feature. Vehicle volume characteristics include length, height, and width.
In this embodiment, the plurality of first dimensions include a vehicle running dimension and a vehicle attribute dimension, and feature extraction of the plurality of first dimensions is performed based on the first sample running data to extract first sample features of the first sample running data in the vehicle running dimension and the vehicle attribute dimension, respectively. And carrying out feature extraction of a plurality of first dimensions based on the second sample driving data to obtain second sample features of the second sample driving data in the vehicle driving dimension and the vehicle attribute dimension respectively, so that a first association relation between the first sample vehicle and the second sample vehicle can be accurately constructed and determined according to the features of the first sample vehicle and the second sample vehicle in the vehicle driving dimension and the vehicle attribute dimension, and the feature correlation of the first sample vehicle and the second sample vehicle in different dimensions can be accurately represented through the association relation.
In one embodiment, determining a first association between the first sample vehicle and the second sample vehicle based on each of the first sample features and each of the second sample features includes:
determining sub-association relations between each first sample feature and each second sample feature; and determining a first association relationship between the first sample vehicle and the second sample vehicle according to each sub-association relationship.
The sub-association relationship characterizes the feature correlation between the first sample feature and the second sample feature.
In particular, the computer device may determine a feature correlation between one first sample feature and each second sample feature, respectively, by which the sub-association between the first sample feature and the second sample feature is characterized.
Traversing each first sample feature by the computer equipment to obtain feature correlation between each first sample feature and each second sample feature, and obtaining each sub-association relation. A first correlation matrix is constructed from each correlation.
In this embodiment, the sub-association relationship between the first sample feature and the second sample feature may be represented by a first covariance, and then the first correlation matrix may be a first correlation covariance matrix. The computer device calculates a first covariance of each first sample feature and each second sample feature, and constructs a first correlation covariance matrix between the first sample vehicle and the second sample vehicle according to the first covariance.
In this embodiment, sub-association relationships between each first sample feature and each second sample feature are determined, so that correlation between any two sample features is represented by the sub-association relationships, and feature correlations of the first sample vehicle and the second sample vehicle on each feature are accurately reflected according to the first association relationship determined by each sub-association relationship.
In one embodiment, determining a second association between the plurality of sample sensors from the plurality of sample road surface data includes:
respectively extracting the characteristics of a plurality of second dimensions based on each sample pavement data to obtain sample pavement characteristics of each sample pavement data respectively corresponding to each second dimension; the plurality of second dimensions includes a road surface travel dimension and a road surface attribute dimension; and determining a second association relationship among the plurality of sample sensors according to the plurality of sample pavement characteristics corresponding to each sample pavement data.
Wherein the road surface driving dimension comprises at least one of a vehicle speed dimension, a vehicle distance dimension or a vehicle POI dimension of the vehicle. The data of the vehicle speed dimension includes at least one of a vehicle speed of the passing vehicle, or a vehicle speed of the passing vehicle in a front-rear direction of the vehicle. The data of the vehicle distance dimension includes at least one of a distance between the passing vehicle and the front-rear vehicle, or a distance between the passing vehicle and the left-right vehicle distance.
The road attribute dimension includes at least one of an image dimension, a road speed limit dimension, a traffic volume dimension, a length of time dimension, or a road segment POI dimension.
The data of the image dimension includes at least one of an image of a road surface of a road segment or an image of a car accident of the road segment. The data of the road speed limit dimension includes at least one of average vehicle speed, speed limit data or over-speed data. The data for the traffic dimension includes traffic data. The data of the duration dimension includes a waiting duration of the vehicle.
Specifically, the sample road surface data includes data in a road surface travel dimension and a road surface property dimension. And the computer equipment respectively performs feature extraction on the road surface driving dimension and the road surface attribute dimension on each sample road surface data to obtain the road surface driving feature corresponding to each sample road surface data on the road surface driving dimension and the road surface attribute feature corresponding to each sample road surface data on the road surface attribute dimension. The computer device uses the road surface driving characteristics and the road surface attribute characteristics corresponding to the same sample road surface data as the sample road surface characteristics of the sample road surface data in the corresponding second dimension, namely the sample road surface characteristics of the sample sensor in the corresponding second dimension.
The computer device may construct a second correlation matrix based on the road surface characteristics of each sample, and characterize a second association between the plurality of sample sensors by the second correlation matrix.
In this embodiment, determining, according to a plurality of sample road surface features corresponding to each sample road surface data, a second association relationship between a plurality of sample sensors includes: determining sub-association relationships between the respective sample road surface features of each sample road surface data; and determining a second association relationship among the plurality of sample sensors according to each sub-association relationship.
The computer device may determine a sub-association between one sample road surface feature of one sample sensor and each sample road surface feature of the other sample sensors, respectively. Traversing each sample road surface characteristic of the sample sensor, thereby obtaining sub-association relations between each sample road surface characteristic and other sample road surface characteristics. And determining a second association relationship among the plurality of sample sensors according to each sub-association relationship.
In this embodiment, the sub-association relationship between the first sample feature and the second sample feature may be referred to as a first sub-association relationship, and the sub-association relationship between the respective sample road surface features may be referred to as a second sub-association relationship.
In this embodiment, the sub-association relationship between the road surface features of each sample may be represented by a second covariance, and then the second association relationship between the plurality of sample sensors may be represented by a second correlation covariance matrix formed by each second covariance.
The sample road surface characteristics of each sample road surface data in each second dimension can be shown in the table of fig. 4, for example, the sample road surface data 1 corresponds to 8 sample road surface characteristics, namely, a vehicle speed characteristic 1, a vehicle distance characteristic 1, a vehicle interest point characteristic 1, an image characteristic 1 of an acquired image, a road surface speed limit characteristic 1, a vehicle flow characteristicSign 1, waiting duration feature 1, and road section interest point feature 1. A second correlation covariance matrix characterization Ω as shown in fig. 4 may be constructed based on 8 sample road surface features corresponding to each of the sample road surface data 1, 2, 3 R 。
In this embodiment, the plurality of second dimensions include a road surface driving dimension and a road surface attribute dimension, and feature extraction of the plurality of second dimensions is performed based on each sample road surface data, so as to obtain features corresponding to each sample sensor in the road surface driving dimension and the road surface attribute dimension, so that a second association relationship of the plurality of sample sensors can be accurately constructed according to features corresponding to each sample sensor in the road surface driving dimension and the road surface attribute dimension, and feature correlation of the plurality of sample sensors in different dimensions can be accurately represented through the second association relationship.
In one embodiment, as shown in fig. 5, based on the first initial weight parameter and each second initial weight parameter, performing cross iterative weight parameter adjustment in combination with the first sample driving data, the second sample driving data, each sample road surface data, the first association relationship and the second association relationship to obtain a first target weight parameter corresponding to the second sample vehicle and a second target weight parameter corresponding to each sample sensor, including:
step S502, determining an initial prediction distance between the first sample vehicle and the second sample vehicle based on the first initial weight parameter and each second initial weight parameter, in combination with the first sample driving data, the second sample driving data, and the respective sample road surface data.
Specifically, the computer device may perform regression analysis processing based on the first initial weight parameter, the first sample travel data, and the second sample travel data, to obtain first sample regression data. And carrying out regression analysis processing based on each second initial weight parameter and each sample pavement data to obtain second sample regression data. And determining an initial predicted distance between the first sample vehicle and the second sample vehicle according to the first sample regression data and the second sample regression data.
In this embodiment, the computer device may multiply the first sample driving data and the second sample driving data with the first initial weight parameter, respectively, and sum the two products obtained by the multiplication to obtain the first sample regression data. For each sample sensor, multiplying the sample pavement data of the single sample sensor by the second initial weight parameter of the sample sensor to obtain a corresponding product of each sample sensor, and summing the products to obtain second sample regression data.
The computer device may sum the first sample regression data and the second sample regression data to obtain an initial predicted distance between the first sample vehicle and the second sample vehicle. The initial predicted distance is a predicted value of the distance between the first sample vehicle and the second sample vehicle obtained by performing distance prediction based on the first initial weight parameter and each of the second initial weight parameters.
Step S504, obtaining the expected distance between the first sample vehicle and the second sample vehicle, and carrying out iterative processing on the weight parameters based on the first initial weight parameters and the initial predicted distance and combining the expected distance, the first sample driving data, the second sample driving data and the first association relation to obtain the first target weight parameters corresponding to the second sample vehicle.
Wherein the expected distance refers to the actual distance between the first sample vehicle and the second sample vehicle, and the expected distance is used as a distance label to participate in the iterative processing of the weight parameters.
Specifically, the computer device obtains a desired distance between the first sample vehicle and the second sample vehicle and determines a distance difference between the desired distance and the initial predicted distance. And carrying out iterative processing on the weight parameters of the second sample vehicle based on the first initial weight parameters, the distance difference, the first sample driving data, the second sample driving data and the first association relation, and obtaining a first intermediate weight parameter corresponding to the second sample vehicle in each iteration. After the first intermediate weight parameter is obtained in the previous iteration, entering the next iteration, taking the first intermediate weight parameter obtained in the previous iteration as the first initial weight parameter in the next iteration, and adjusting the weight parameter of the second sample vehicle according to the processing mode until the iteration stopping condition is met, and obtaining the first target weight parameter corresponding to the second sample vehicle.
In this embodiment, the computer device determines a first product of the distance difference, the first sample traveling data, the second sample traveling data, and the first association relationship, and uses a difference between the first initial weight parameter and the first product as a first intermediate weight parameter, that is, a first intermediate weight parameter of the second sample vehicle after the iterative processing. Since the initial prediction distance is calculated based on the first initial weight parameter and each second initial weight parameter, the first initial weight parameter and each second initial weight parameter obtain a first intermediate weight parameter and a second intermediate weight parameter after iterative processing. In the next iteration, the first intermediate weight parameter obtained in the previous iteration is used as a first initial weight parameter in the next iteration, each second intermediate weight parameter obtained in the previous iteration is used as each second intermediate weight parameter in the next iteration, so that the initial predicted distance is updated in the next iteration, and the updated distance difference is calculated based on the expected distance and the updated initial predicted distance. And taking the difference value of the updated distance difference, the product of the first sample driving data, the second sample driving data and the first association relation and the first initial weight parameter as a first intermediate weight parameter of the next iteration. And the like, until the iteration stopping condition is met, obtaining a first target weight parameter corresponding to the second sample vehicle.
In one embodiment, the computer device obtains a first learning rate corresponding to the second sample vehicle, the first learning rate being a pre-set empirical value. The Learning rate (Learning rate) is an important super-parameter in supervised Learning and deep Learning, and determines whether and when an objective function can converge to a local minimum. The computer equipment determines a first product of a first learning rate, a distance difference, first sample driving data, second sample driving data and a first association relation, and takes a difference value between a first initial weight parameter and the first product as an updated weight parameter of the second sample vehicle, namely, the weight parameter of the second sample vehicle obtained through iterative processing.
Step S506, based on each second initial weight parameter and the initial predicted distance, carrying out iterative processing of the weight parameters by combining the expected distance, the pavement data of each sample and the second association relation to obtain a second target weight parameter corresponding to each sample sensor.
Specifically, the computer equipment performs iterative processing on the weight parameters of each sample sensor based on each second initial weight parameter, the distance difference, each sample pavement data and the second association relation, and obtains a second intermediate weight parameter corresponding to each sample sensor in each iteration. After the second intermediate weight parameter is obtained in the previous iteration, entering the next iteration, taking the second intermediate weight parameter obtained in the previous iteration as the second initial weight parameter in the next iteration, and adjusting the weight parameter of each sample sensor according to the processing mode until the iteration stopping condition is met, and obtaining the second target weight parameter corresponding to each sample sensor.
In this embodiment, the computer device determines a second product of the distance difference, the road surface data of each sample, and the second association relationship, and uses a difference between the second initial weight parameter and the second product as a second intermediate weight parameter, that is, a second intermediate weight parameter of each sample sensor obtained by iterative processing. Since the initial prediction distance is calculated based on the first initial weight parameter and each second initial weight parameter, the first initial weight parameter and each second initial weight parameter obtain a first intermediate weight parameter and a second intermediate weight parameter after iterative processing. In the next iteration, the first intermediate weight parameter obtained in the previous iteration is used as a first initial weight parameter in the next iteration, each second intermediate weight parameter obtained in the previous iteration is used as each second initial weight parameter in the next iteration, so that the initial predicted distance is updated in the next iteration, and the updated distance difference is calculated based on the expected distance and the updated initial predicted distance. And taking the updated second product of the distance difference, the pavement data of each sample and the second association relation and the difference value of the second initial weight parameter as a second intermediate weight parameter of the next iteration. And the like, until the iteration stopping condition is met, obtaining a second target weight parameter corresponding to each sample sensor.
In one embodiment, the computer device obtains a second learning rate corresponding to the sample sensor, the second learning rate being a pre-set empirical value. The first learning rate may be the same as or different from the second learning rate. The computer equipment determines a second product of a second learning rate, a distance difference and a second association relation of each sample pavement data, and takes a difference value between a second initial weight parameter and the second product as a second intermediate weight parameter, namely the second intermediate weight parameter of each sample sensor obtained through iterative processing.
In this embodiment, based on the first initial weight parameter and each second initial weight parameter, the first sample traveling data, the second sample traveling data, and the respective sample road surface data are combined to predict the distance between the first sample vehicle and the second sample vehicle, so as to obtain an initial predicted distance. The expected distance between the first sample vehicle and the second sample vehicle is obtained to be used as a distance label in iteration, the difference between the real distance and the distance predicted based on each weight parameter can be determined, the weight parameter of the second sample vehicle is iterated by combining the running data of each vehicle and the data correlation between the running data, and the weight parameter of the second sample vehicle is continuously optimized in multiple iterations, so that the first target weight parameter of the second sample vehicle is accurately obtained. And iterating the weight parameters of each sample sensor by combining the sample pavement data acquired by each sample sensor and the data correlation between the sample pavement data, so that the weight parameters of each sample sensor are continuously optimized in a plurality of iterations, and a second target weight parameter corresponding to each sample sensor is accurately obtained.
In one embodiment, the second sample vehicle includes a forward sample vehicle and a backward sample vehicle of the first sample vehicle; the driving position of the forward sample vehicle on the sample road is before the driving position of the first sample vehicle, and the driving position of the backward sample vehicle on the sample road is after the driving position of the first sample vehicle; the second sample travel data includes forward sample travel data of a forward sample vehicle and backward sample travel data of a backward sample vehicle; the initial prediction distance comprises a forward initial prediction distance and a backward initial prediction distance between the forward sample vehicle and the backward sample vehicle and between the backward sample vehicle and the first sample vehicle respectively; the expected distance comprises a forward expected distance and a backward expected distance between the forward sample vehicle and the backward sample vehicle and the first sample vehicle respectively; the first association relation comprises a forward association relation and a backward association relation between the forward sample vehicle and the backward sample vehicle and the first sample vehicle respectively; the first initial weight parameters comprise forward initial weight parameters of the forward sample vehicle and backward initial weight parameters of the backward sample vehicle; the first target weight parameters include forward target weight parameters of the forward sample vehicle and backward target weight parameters of the backward sample vehicle.
The initial prediction distance comprises a forward initial prediction distance between the first sample vehicle and the forward sample vehicle and a backward initial prediction distance between the first sample vehicle and the backward sample vehicle; the desired distance includes a forward desired distance between the first sample vehicle and the forward sample vehicle, and a rearward desired distance between the first sample vehicle and the rearward sample vehicle; the first association relationship includes a forward association relationship between the first sample vehicle and the forward sample vehicle, and a backward association relationship between the first sample vehicle and the backward sample vehicle.
Specifically, first sample travel data of a first sample vehicle, forward sample travel data of a forward sample vehicle of the first sample vehicle, and backward sample travel data of a backward sample vehicle of the first sample vehicle are acquired; the forward sample vehicle travel position on the sample road is before the first sample vehicle travel position, and the backward sample vehicle travel position on the sample road is after the first sample vehicle travel position.
A forward initial weight parameter of the forward sample vehicle, a backward initial weight parameter of the backward sample vehicle, and a second initial weight parameter of each sample sensor are obtained.
And determining a forward prediction distance between the first sample vehicle and the forward sample vehicle and a backward prediction distance between the first sample vehicle and the backward sample vehicle based on the forward initial weight parameter, the backward initial weight parameter and each second initial weight parameter in combination with the first sample driving data, the forward sample driving data, the backward sample driving data and each sample road surface data. A forward desired distance between the first sample vehicle and the forward sample vehicle, and a backward desired distance between the first sample vehicle and the backward sample vehicle are obtained. And carrying out iterative processing on the weight parameters based on the forward initial weight parameters, the backward initial weight parameters, the forward predicted distance and the backward predicted distance, and combining the forward expected distance, the backward expected distance, the first sample driving data, the forward sample driving data, the backward sample driving data, the forward association relationship and the backward association relationship to obtain forward target weight parameters corresponding to the forward sample vehicles and backward target weight parameters corresponding to the backward sample vehicles.
And carrying out iterative processing on the weight parameters based on each second initial weight parameter, the forward predicted distance and the backward predicted distance by combining the forward expected distance, the backward expected distance, the pavement data of each sample and the second association relation to obtain a second target weight parameter corresponding to each sample sensor.
In this embodiment, based on the forward initial weight parameter, the backward initial weight parameter, and each second initial weight parameter, the distances between the first sample vehicle and the forward sample vehicle and between the first sample vehicle and the backward sample vehicle are predicted by combining the first sample travel data, the forward sample travel data, the backward sample travel data, and the respective sample road surface data, so as to obtain a forward predicted distance and a backward predicted distance. The method comprises the steps of obtaining a forward expected distance and a backward expected distance as distance labels in iteration, determining the difference between the real forward distance and the forward distance predicted based on the weight parameters of a forward sample vehicle and the weight parameters of each sample sensor, and the difference between the real backward distance and the backward distance predicted based on the weight parameters of a backward sample vehicle and the weight parameters of each sample sensor, combining the data correlation between the running data of each vehicle and the running data, and iterating the weight parameters of the forward sample vehicle, the backward sample vehicle and each sample sensor according to the data correlation between the running data of each vehicle and the data correlation between the running data of each sample sensor, so that the weight parameters of each sample vehicle, each sample sensor are continuously optimized in multiple iterations, and the target weight parameters corresponding to each of the forward sample vehicle, the backward sample vehicle and each sample sensor are accurately obtained.
In one embodiment, determining an initial predicted distance between the first sample vehicle and the second sample vehicle based on the first initial weight parameter and each of the second initial weight parameters in combination with the first sample travel data, the second sample travel data, and the respective sample road surface data comprises:
based on the forward initial weight parameters and each second initial weight parameter, combining the first sample driving data, the forward sample driving data and the pavement data of each sample, and determining the forward initial prediction distance between the first sample vehicle and the forward sample vehicle;
and determining a backward initial prediction distance between the first sample vehicle and the backward sample vehicle by combining the first sample driving data, the backward sample driving data and the pavement data of each sample based on the backward initial weight parameters and each second initial weight parameter.
Specifically, the computer device may perform regression analysis processing based on the forward initial weight parameter, the first sample travel data, and the forward sample travel data to obtain forward sample regression data. And carrying out regression analysis processing based on each second initial weight parameter and each sample pavement data to obtain second sample regression data. And determining the forward initial prediction distance between the first sample vehicle and the forward sample vehicle according to the forward sample regression data and the second sample regression data.
The computer device may perform regression analysis processing based on the backward initial weight parameter, the first sample travel data, and the backward sample travel data to obtain backward sample regression data. And carrying out regression analysis processing based on each second initial weight parameter and each sample pavement data to obtain second sample regression data. And determining a backward initial prediction distance between the first sample vehicle and the backward sample vehicle according to the backward sample regression data and the second sample regression data.
In this embodiment, the computer device may multiply the first sample driving data and the forward sample driving data with the forward initial weight parameter, respectively, and sum the two products obtained by the multiplication to obtain the forward sample regression data. For each sample sensor, multiplying the sample pavement data of the single sample sensor by the second initial weight parameter of the sample sensor to obtain a corresponding product of each sample sensor, and summing the products to obtain second sample regression data. The computer equipment acquires a forward random sequence, and performs summation processing on the forward sample regression data, the second sample regression data and the forward random sequence to obtain a forward initial prediction distance between the first sample vehicle and the forward sample vehicle. The forward random sequence indicates that the distance between the first sample vehicle and the forward sample vehicle obeys a standard normal distribution.
The computer device may multiply the first sample travel data and the backward sample travel data with the backward initial weight parameter, respectively, and sum the two products obtained by the multiplication to obtain backward sample regression data. And the computer equipment acquires a backward random sequence, and performs summation processing on the backward sample regression data, the second sample regression data and the backward random sequence to obtain a backward initial prediction distance between the first sample vehicle and the backward sample vehicle. The backward random sequence indicates that the distance between the first sample vehicle and the backward sample vehicle obeys a standard normal distribution.
For example, the computer device may calculate the forward initial predicted distance by the following forward distance formula:
the computer device may calculate the backward initial predicted distance by the following backward distance formula:
in this embodiment, based on the forward initial weight parameter and each second initial weight parameter, the forward initial prediction distance between the first sample vehicle and the forward sample vehicle can be accurately calculated by combining the first sample travel data, the forward sample travel data, and the respective sample road surface data. Based on the backward initial weight parameters and each second initial weight parameter, the backward initial prediction distance between the first sample vehicle and the backward sample vehicle can be accurately calculated by combining the first sample driving data, the backward sample driving data and the sample road surface data, so that the weight parameters of the forward and backward vehicles and the weight parameters of the sample sensor are adjusted according to the prediction distance between the first sample vehicle and the forward and backward vehicles to obtain the optimal solution of each weight parameter.
In one embodiment, based on the first initial weight parameter and the initial predicted distance, performing iterative processing of the weight parameter in combination with the expected distance, the first sample driving data, the second sample driving data and the first association relationship to obtain a first target weight parameter corresponding to the second sample vehicle, including:
based on the forward initial weight parameter and the forward initial prediction distance, carrying out iterative processing on the weight parameter by combining the forward expected distance, the first sample driving data, the forward sample driving data and the forward association relation to obtain a forward target weight parameter corresponding to the forward sample vehicle;
and carrying out iterative processing on the weight parameters based on the backward initial weight parameters and the backward initial prediction distance and combining the backward expected distance, the first sample driving data, the backward sample driving data and the backward association relation to obtain backward target weight parameters corresponding to the backward sample vehicle.
Specifically, the computer device obtains a forward expected distance between the first sample vehicle and the forward sample vehicle and determines a forward distance difference between the forward expected distance and a forward initial predicted distance. And carrying out iterative processing on the weight parameters of the forward sample vehicle based on the forward initial weight parameters, the forward distance difference, the first sample driving data, the forward sample driving data and the forward association relation, and obtaining forward intermediate weight parameters corresponding to the forward sample vehicle in each iteration. After the previous iteration process obtains the forward intermediate weight parameter, the next iteration is carried out, and in the next iteration, the forward intermediate weight parameter obtained in the previous iteration is used as the forward initial weight parameter in the next iteration, and the weight parameter of the forward sample vehicle is adjusted according to the processing mode until the iteration stop condition is met, so that the forward target weight parameter corresponding to the forward sample vehicle is obtained.
The computer device obtains a backward desired distance between the first sample vehicle and the backward sample vehicle and determines a backward distance difference between the backward desired distance and the backward initial predicted distance. And carrying out iterative processing on the weight parameters of the backward sample vehicle based on the backward initial weight parameters, the backward distance difference, the first sample driving data, the backward sample driving data and the backward association relation, and obtaining backward intermediate weight parameters corresponding to the backward sample vehicle in each iteration. And after the last iteration process obtains the backward intermediate weight parameter, entering the next iteration, taking the backward intermediate weight parameter obtained in the last iteration as the backward initial weight parameter in the next iteration, and adjusting the weight parameter of the backward sample vehicle according to the processing mode until the backward target weight parameter corresponding to the backward sample vehicle is obtained when the iteration stop condition is met.
In this embodiment, the computer device determines a forward product of the forward distance difference, the first sample traveling data, the forward sample traveling data, and the forward association relation, and takes a difference between the forward initial weight parameter and the forward product as a forward intermediate weight parameter. Since the forward initial prediction distance is calculated based on the forward initial weight parameter and each second initial weight parameter, the forward initial weight parameter and each second initial weight parameter obtain a forward intermediate weight parameter and a second intermediate weight parameter after iterative processing. In the next iteration, the forward intermediate weight parameter obtained in the previous iteration is used as a forward initial weight parameter in the next iteration, each second intermediate weight parameter obtained in the previous iteration is used as each second initial weight parameter in the next iteration, so that the forward initial predicted distance is updated in the next iteration, and the updated forward distance difference is calculated based on the forward expected distance and the updated forward initial predicted distance. And taking the difference value of the forward initial weight parameter and the product of the updated forward distance difference, the first sample driving data, the forward sample driving data and the forward association relation as a forward intermediate weight parameter of the next iteration. And the like, until the iteration stop condition is met, obtaining a forward target weight parameter corresponding to the forward sample vehicle.
The iterative process of the backward initial weight parameters of the rear vehicle is similar to that of the forward initial weight parameters of the forward vehicle, and the forward target weight parameters corresponding to the forward sample vehicle can be obtained by adjusting the data of all the forward vehicles to the data of the backward vehicle.
In this embodiment, based on the forward initial weight parameter and the forward initial prediction distance, the forward expected distance, the first sample running data, the forward sample running data and the forward association relation are combined to perform iterative processing of the weight parameter, so that the weight parameter of the forward sample vehicle is optimized by combining the data in each aspect in multiple iterations, and an optimal solution of the weight parameter corresponding to the forward sample vehicle, that is, the forward target weight parameter, is obtained. And (3) based on the backward initial weight parameter and the backward initial prediction distance, carrying out iterative processing of the weight parameter by combining the backward expected distance, the first sample driving data, the backward sample driving data and the backward association relation, and optimizing the weight parameter of the backward sample vehicle by combining the data in each aspect in multiple iterations to obtain an optimal solution of the weight parameter corresponding to the backward sample vehicle, namely the backward target weight parameter.
In one embodiment, based on each second initial weight parameter and the initial predicted distance, performing iterative processing of the weight parameters in combination with the expected distance, the pavement data of each sample and the second association relationship to obtain a second target weight parameter corresponding to each sample sensor, including:
and carrying out iterative processing on the weight parameters based on each second initial weight parameter, the forward initial prediction distance and the backward initial prediction distance and combining the forward expected distance and the backward expected distance, the pavement data of each sample and the second association relation to obtain a second target weight parameter corresponding to each sample sensor.
Specifically, the computer device calculates a global initial predicted distance between the forward sample vehicle and the backward sample vehicle, and a global desired distance between the forward sample vehicle and the backward sample vehicle, based on the forward initial predicted distance, the backward initial predicted distance, the forward desired distance, and the backward desired distance. And calculating the global distance difference according to the global initial predicted distance and the global expected distance. The global distance difference characterizes a difference between the global expected distance and the global initial predicted distance.
The computer equipment performs iterative processing on the weight parameters of each sample sensor based on each second initial weight parameter, the global distance difference, each sample pavement data and the second association relation, and obtains second intermediate weight parameters corresponding to each sample sensor in each iteration. After the second intermediate weight parameter is obtained in the previous iteration, entering the next iteration, taking the second intermediate weight parameter obtained in the previous iteration as the second initial weight parameter in the next iteration, and adjusting the weight parameter of each sample sensor according to the processing mode until the iteration stopping condition is met, and obtaining the second target weight parameter corresponding to each sample sensor.
In this embodiment, based on each second initial weight parameter, the forward initial predicted distance and the backward initial predicted distance, iterative processing of the weight parameters is performed in combination with the forward expected distance and the backward expected distance, the pavement data of each sample and the second association relationship, so that the weight parameters of each sample sensor are optimized in multiple iterations in combination with the data of each aspect, and an optimal solution of the weight parameters corresponding to each sample sensor, namely each second target weight parameter, is obtained.
In one embodiment, based on each second initial weight parameter, a forward initial predicted distance and a backward initial predicted distance, and in combination with the forward expected distance and the backward expected distance, each sample road surface data and a second association relationship, performing iterative processing of the weight parameters to obtain a second target weight parameter corresponding to each sample sensor, including:
determining a global expected distance according to the forward expected distance and the backward expected distance;
determining a global initial predicted distance according to the forward initial predicted distance and the backward initial predicted distance;
and carrying out iterative processing on the weight parameters based on each second initial weight parameter and the global initial prediction distance and combining the global expected distance, the pavement data of each sample and the second association relation to obtain a second target weight parameter corresponding to each sample sensor.
Specifically, the computer device takes the sum of the forward initial predicted distance and the backward initial predicted distance as the global initial predicted distance between the forward sample vehicle and the backward sample vehicle. The sum of the forward expected distance and the backward expected distance is taken as the global expected distance between the forward sample vehicle and the backward sample vehicle. And taking the difference value between the global expected distance and the global initial predicted distance as the global distance difference.
The computer equipment determines a second product of the global distance difference, the pavement data of each sample and a second association relation, and takes the difference between a second initial weight parameter and the second product as a second intermediate weight parameter, namely the second intermediate weight parameter of each sample sensor obtained through iterative processing. Since the global initial prediction distance is calculated based on the forward initial weight parameter, the backward initial weight parameter and each second initial weight parameter, the forward initial weight parameter, the backward initial weight parameter and each second initial weight parameter are iteratively processed to obtain a forward intermediate weight parameter, a backward intermediate weight parameter and a second intermediate weight parameter. In the next iteration, the forward intermediate weight parameter, the backward intermediate weight parameter and each second intermediate weight parameter obtained in the previous iteration are respectively used as the forward initial weight parameter, the backward initial weight parameter and each second initial weight parameter in the next iteration, so that the global initial prediction distance is updated in the next iteration, and the updated global distance difference is calculated based on the global expected distance and the updated global initial prediction distance. And taking the updated second product of the global distance difference, the pavement data of each sample and the second association relation and the difference value of the second initial weight parameter as a second intermediate weight parameter of the next iteration. And the like, until the iteration stopping condition is met, obtaining a second target weight parameter corresponding to each sample sensor.
In this embodiment, the global initial predicted distance between the front and rear vehicles of the first sample vehicle is calculated according to the forward expected distance and the backward expected distance, and the global expected distance between the front and rear vehicles of the first sample vehicle is calculated according to the forward initial predicted distance and the backward initial predicted distance, so that the influence of the front and rear vehicles on the first sample vehicle can be considered at the same time. And (3) carrying out iterative processing of the weight parameters based on each second initial weight parameter and the global initial prediction distance and combining the global expected distance, each sample pavement data and the second association relation, so that the weight parameters of each sample sensor are optimized according to the data of the first sample vehicle, the data of the front and rear vehicles, the correlation among the sensors, the vehicle data acquired by each sample sensor and other multi-aspect data in multiple iterations, and the optimal solution of the weight parameters corresponding to each sample sensor is obtained, so that each obtained second target weight parameter is suitable for a scene of distance prediction between the first vehicle and the front and rear vehicles when the first vehicle has the front and rear vehicles at the same time.
In one embodiment, as shown in fig. 6, there is provided a vehicle data processing method, which is exemplified as being applied to a computer device (the computer device may be a terminal or a server in fig. 1), including the steps of:
step S602, acquiring first driving data of a first vehicle and second driving data of a second vehicle; the first vehicle and the second vehicle have a traveling sequence on the target road.
Wherein, the first vehicle and the second vehicle refer to vehicles needing to perform distance detection. The second vehicle includes at least one of a forward vehicle or a rearward vehicle of the first vehicle. The forward vehicle travel position on the target road is before the first vehicle travel position, and the backward vehicle travel position on the target road is after the first vehicle travel position. The driving sequence refers to the order in which each vehicle is driven on the road.
The first travel data refers to travel data associated with a first vehicle and may include data of at least one dimension. The data of the at least one dimension includes at least one of data of the first vehicle in a vehicle travel dimension, or vehicle attribute data of the first vehicle in a vehicle attribute dimension.
The second travel data refers to travel data associated with a second vehicle and may include data of at least one dimension. The data of the at least one dimension includes at least one of data of the second vehicle in a vehicle travel dimension, or vehicle attribute data of the second vehicle in a vehicle attribute dimension.
The data in the vehicle travel dimension includes at least one of a vehicle own speed, a front-rear vehicle speed, a road surface image collected by the vehicle, or vehicle POI data. The vehicle attribute data in the vehicle attribute dimension includes at least one of a vehicle type, a vehicle length, a height, a width, a vehicle core number, a vehicle payload number, a vehicle maximum speed per hour, or a vehicle displacement.
Specifically, the first vehicle performs data acquisition on the vehicle driving dimension to obtain corresponding first driving data. Further, the first vehicle performs data acquisition in the dimension of the vehicle attribute, namely, acquires attribute data of the vehicle itself, and obtains the vehicle attribute data. The related data and the vehicle attribute data acquired in the vehicle driving dimension are taken as first driving data of the first vehicle.
And the second vehicle performs data acquisition on the vehicle driving dimension to obtain corresponding second driving data. Further, the second vehicle performs data acquisition in the dimension of the vehicle attribute, namely, acquires attribute data of the vehicle itself, and obtains the vehicle attribute data. And taking the related data and the vehicle attribute data acquired in the vehicle driving dimension as second driving data of a second vehicle.
The computer device may determine that there is a driving sequence of the first vehicle and the second vehicle on the target road, and acquire first driving data of the first vehicle, and second driving data of the second vehicle.
In step S604, target road surface data acquired by a plurality of sensors associated with the first vehicle for the target road, respectively, is acquired.
The sensor is arranged in the target road and is used for collecting data of vehicles in the target road.
The plurality of sensors associated with the first vehicle includes a target sensor closest to a travel position of the first vehicle and at least one of a forward sensor or a backward sensor corresponding to the target sensor. The position of the forward sensor on the target road is before the position of the target sensor, and the position of the backward sensor on the target road is after the position of the target sensor.
Or, the plurality of sensors associated with the first vehicle include an object sensor within a preset range of the first vehicle and at least one of a forward sensor or a backward sensor corresponding to the object sensor.
For example, the plurality of sensors associated with the first vehicle includes a target sensor that the first vehicle is passing by and at least one of a forward sensor or a backward sensor corresponding to the sensor that the first vehicle is passing by.
The road surface data refers to data obtained by data acquisition of vehicles passing through a target road by a sensor. The road surface data may include data of at least one dimension including at least one of road surface travel data in a road surface travel dimension in which the vehicle travels in the target road, or road surface attribute data in a road surface attribute dimension of the target road.
The road surface travel data in the road surface travel dimension includes at least one of a speed of the passing vehicle, a front-rear vehicle speed, a front-rear vehicle distance, a left-right vehicle distance, or vehicle POI data.
The road surface attribute data in the road surface attribute dimension includes at least one of an image of a road surface of a road segment, an average vehicle speed, speed limit data, overspeed data, traffic accident image, traffic flow data, waiting time period, or road segment POI information.
Specifically, a sensor in a target road collects vehicle running data of a passing vehicle in a road surface running dimension to obtain road surface running data of the vehicle in the road surface running dimension. The sensor in the target road can also acquire data of the target road in the dimension of the road surface attribute to obtain the road surface attribute data.
For example, the sensor collects vehicle running data of a first vehicle passing by, obtains road surface running data corresponding to the first vehicle, and collects road surface attribute data corresponding to the target road when the first vehicle passes by. At least one of road surface driving data and road surface attribute data is used as road surface data corresponding to the sensor. And each sensor performs the same data acquisition processing on the passing vehicle, so that the pavement data corresponding to each sensor can be obtained.
The computer device determines a target sensor closest to a travel position of the first vehicle in the target road, and at least one of a forward sensor or a backward sensor corresponding to the target sensor. The target sensor and the forward sensor are taken as a plurality of sensors associated with the first vehicle, or the target sensor and the backward sensor are taken as a plurality of sensors associated with the first vehicle, or the target sensor, the forward sensor and the backward sensor are taken as a plurality of sensors associated with the first vehicle.
In this embodiment, the computer device determines a target sensor in a preset range of the first vehicle in the road, and determines at least one of a forward sensor or a backward sensor corresponding to the target sensor.
The computer device may obtain the road surface data collected by each sensor, resulting in a plurality of road surface data.
Step S606, according to the driving sequence, obtaining a first target weight parameter obtained by pre-training corresponding to the second vehicle and a second target weight parameter obtained by pre-training corresponding to each sensor.
The first target weight parameter refers to a target value of a weight parameter corresponding to a second vehicle, and the second target weight parameter refers to a target value of a weight parameter corresponding to a sensor. The target value of the weight parameter is generated by adjusting the initial value of the weight parameter in the pre-training.
The first target weight parameter is obtained by performing iterative adjustment on a first initial weight parameter corresponding to the second sample vehicle in pre-training. The second target weight parameter is obtained by iterative adjustment of a second initial weight parameter corresponding to the sample sensor in the pre-training.
The pre-training process comprises the following steps: acquiring first sample driving data of a first sample vehicle and second sample driving data of a second sample vehicle; the first sample vehicle and the second sample vehicle have a driving sequence on the sample road; acquiring sample pavement data which are acquired by a plurality of sample sensors related to a first sample vehicle and are respectively aimed at a sample road; determining a first association relationship between the first sample vehicle and the second sample vehicle according to the first sample driving data and the second sample driving data; determining a second association relationship among the plurality of sample sensors according to the plurality of sample pavement data; acquiring a first initial weight parameter of a second sample vehicle and a second initial weight parameter of each sample sensor; based on the first initial weight parameter and each second initial weight parameter, the first sample driving data, the second sample driving data, the pavement data of each sample, the first association relationship and the second association relationship are combined, and the weight parameter adjustment of the cross iteration is carried out to obtain a first target weight parameter corresponding to the second sample vehicle and a second target weight parameter corresponding to each sample sensor.
Specifically, the computer equipment acquires a first target weight parameter obtained through pre-training corresponding to the second vehicle and a second target weight parameter obtained through pre-training corresponding to each sensor according to the driving sequence.
Step S608, performing regression analysis processing based on the first driving data, the second driving data and the first target weight parameter to obtain first regression data; and carrying out regression analysis processing based on the pavement data and the second target weight parameters to obtain second regression data.
Specifically, the computer device may multiply the first travel data and the second travel data with the first target weight parameter, respectively, and sum the two products obtained by the multiplication to obtain the first regression data. For each sensor, multiplying the road surface data of the single sensor by the second target weight parameter of the sensor to obtain a corresponding product of each sensor, and summing the products to obtain second regression data.
Step S610, determining a vehicle distance between the first vehicle and the second vehicle according to the first regression data and the second regression data.
Specifically, the computer device may sum the first regression data and the second regression data to obtain a vehicle distance between the first vehicle and the second vehicle.
In this embodiment, the computer device performs a summation process on the first regression data, the second regression data, and a preset random sequence to obtain a vehicle distance between the first vehicle and the second vehicle. The preset random sequence indicates that the vehicle distance is subject to a standard normal distribution.
In this embodiment, feature extraction is performed on the first driving data and the second driving data respectively to obtain a first feature corresponding to the first vehicle and a second feature corresponding to the second vehicle; respectively extracting the characteristics of each road surface data to obtain the road surface characteristics corresponding to each sensor; carrying out regression analysis processing based on the first characteristic, the second characteristic and the first target weight parameter to obtain first regression data; and carrying out regression analysis processing based on the road surface characteristics and the second target weight parameters to obtain second regression data.
In this embodiment, second driving data of a first vehicle and a second vehicle, which have driving sequences on a target road, and first target weight parameters obtained by pre-training corresponding to the second vehicle are obtained, and regression analysis processing is performed based on the first driving data, the second driving data and the first target weight parameters, so that the first target weight parameters are used as regression coefficients to regress the first driving data and the second driving data, and correlation between a vehicle distance to be solved and driving data of each vehicle, namely, first regression data is obtained. Acquiring target road surface data which are acquired by a plurality of sensors related to a first vehicle and respectively aiming at a target road, and respectively corresponding second target weight parameters which are obtained by pretraining each sensor, and carrying out regression analysis processing based on the road surface data and the second target weight parameters, so that the second target weight parameters are used as regression coefficients to carry out regression on the road surface data of the sensors, and obtaining the correlation between the vehicle distance to be solved and the data acquired by the sensors, namely second regression data. According to the correlation between the vehicle distance to be solved and the running data of each vehicle and the correlation between the vehicle distance to be solved and the data acquired by each sensor, the vehicle distance between the first vehicle and the second vehicle can be calculated more accurately.
In one embodiment, the method further comprises: and prompting when the vehicle distance between the first vehicle and the second vehicle meets the distance prompting condition.
The distance presentation condition refers to a condition for presenting a distance between vehicles in relation to a distance between vehicles. The distance cue condition may specifically be that the distance between vehicles is less than a distance threshold.
When the distance between the first vehicle and the second vehicle meets the distance prompt condition, the distance between the first vehicle and the second vehicle is short, and potential safety hazards of collision of the vehicles exist, voice prompt is carried out through the vehicle-mounted terminal of the first vehicle, or a vehicle owner is prompted through the screen of the vehicle-mounted terminal of the first vehicle, collision between the vehicles is effectively prevented, and traffic safety is improved.
In one embodiment, the second vehicle includes a forward vehicle and a rearward vehicle of the first vehicle, the forward vehicle traveling on the target road before the first vehicle traveling, and the rearward vehicle traveling on the target road after the first vehicle traveling; the second traveling data includes forward traveling data of a forward vehicle and backward traveling data of a backward vehicle, the first target weight parameter includes a forward target weight parameter of the forward vehicle and a backward target weight parameter of the backward vehicle, the first regression data includes forward regression data of the forward vehicle and backward regression data of the backward vehicle, the vehicle distance includes a forward distance between the first vehicle and the forward vehicle, and a backward distance between the first vehicle and the backward vehicle.
Specifically, first running data of a first vehicle, forward running data of a forward vehicle of the first vehicle, and backward running data of a backward vehicle of the first vehicle are acquired; the forward vehicle travel position on the target road is before the first vehicle travel position, and the backward vehicle travel position on the target road is after the first vehicle travel position.
Target road surface data acquired by a plurality of sensors associated with a first vehicle for a target road, respectively, is acquired. And acquiring a forward target weight parameter obtained by pre-training corresponding to the forward vehicle, a backward target weight parameter obtained by pre-training corresponding to the backward vehicle and a second target weight parameter obtained by pre-training corresponding to each sensor according to the driving sequence.
And carrying out regression analysis processing based on the first driving data, the forward driving data, the backward driving data, the forward target weight parameter and the backward target weight parameter to obtain forward regression data of the forward vehicle and backward regression data of the backward vehicle. And carrying out regression analysis processing based on the pavement data and the second target weight parameters to obtain second regression data. And determining the forward distance between the first vehicle and the forward vehicle and the backward distance between the first vehicle and the backward vehicle according to the forward regression data, the backward regression data and the second regression data. In this embodiment, respective driving data of a first vehicle, a forward vehicle and a backward vehicle of the first vehicle, which have a driving sequence on a target road, are acquired, and respective target weight parameters are obtained through pre-training corresponding to the forward vehicle and the backward vehicle, and regression analysis processing is performed based on the driving data of each vehicle and the target weight parameters corresponding to each vehicle, so that the target weight parameters are used as regression coefficients to regress the respective driving data, and correlation between a vehicle distance to be solved and the driving data of each vehicle is obtained. Acquiring target road surface data which are acquired by a plurality of sensors related to a first vehicle and respectively aiming at a target road, and respectively corresponding second target weight parameters which are obtained by pretraining each sensor, and carrying out regression analysis processing based on the road surface data and the second target weight parameters, thereby taking the second target weight parameters as regression coefficients to carry out regression on the road surface data of the sensors so as to obtain the correlation between the vehicle distance to be solved and the data acquired by each sensor. According to the correlation between the forward distance to be solved and the running data of the forward vehicle and the correlation between the forward distance to be solved and the data acquired by each sensor, the forward distance between the first vehicle and the forward vehicle can be calculated more accurately. According to the correlation between the backward distance to be solved and the running data of the forward vehicle and the correlation between the backward distance to be solved and the data acquired by each sensor, the forward distance between the first vehicle and the backward vehicle can be calculated more accurately.
In one embodiment, performing regression analysis based on the first travel data, the second travel data, and the first target weight parameter to obtain first regression data includes:
carrying out regression analysis processing according to the first driving data, the forward driving data and the forward target weight parameter to obtain forward regression data; performing regression analysis processing according to the first driving data, the backward driving data and the backward target weight parameter to obtain backward regression data;
determining a vehicle distance between the first vehicle and the second vehicle based on the first regression data and the second regression data, comprising:
determining a forward distance between the first vehicle and the forward vehicle according to the forward regression data and the second regression data; and determining the backward distance between the first vehicle and the backward vehicle according to the backward regression data and the second regression data.
Specifically, the computer device may multiply the first traveling data and the forward traveling data with the forward target weight parameter, respectively, and sum the two products obtained by the multiplication to obtain the forward regression data. For each sensor, multiplying the road surface data of the single sensor by the second target weight parameter of the sensor to obtain a corresponding product of each sensor, and summing the products to obtain second regression data. The computer equipment acquires a forward random sequence, and sums the forward regression data, the second regression data and the forward random sequence to obtain the forward distance between the first vehicle and the forward vehicle. The forward random sequence indicates that the distance between the first vehicle and the forward vehicle obeys a standard normal distribution.
The computer device may multiply the first traveling data and the backward traveling data with the backward target weight parameter, respectively, and sum the two products obtained by the multiplication to obtain backward regression data. And the computer equipment acquires a backward random sequence, and sums the backward regression data, the second regression data and the backward random sequence to obtain the backward distance between the first vehicle and the backward vehicle. The backward random sequence indicates that the distance between the first vehicle and the backward vehicle obeys a standard normal distribution.
In this embodiment, the forward distance between the first vehicle and the forward vehicle may be calculated by a forward distance formula:
the backward distance between the first vehicle and the backward vehicle can be calculated by a backward distance formula:
wherein X is c,k The first travel data, the forward travel data and the backward travel data are represented, or the first feature corresponding to the first travel data, the forward feature corresponding to the forward travel data and the backward feature corresponding to the backward travel data are represented. When k=0 represents the first vehicle, then X c,k First travel data or first characteristics representing a first vehicle; when k=1 represents a forward vehicle, then X c,k Representing forward travel data or forward characteristicsSign of the disease; when k= -1 represents a backward vehicle, then X c,k Representing backward travel data or backward characteristics. R represents a sensor, X R,i Representing the road surface data or road surface characteristics of the ith sensor.
In this embodiment, the forward distance between the first vehicle and the forward vehicle may be calculated by a forward distance model, and the forward distance model calculates the input data by using a forward distance formula to obtain the forward distance. The backward distance between the first vehicle and the backward vehicle can be calculated through a backward distance model, and the backward distance model calculates input data by using a backward distance formula to obtain the backward distance.
In this embodiment, regression analysis processing is performed according to the first driving data, the forward driving data and the forward target weight parameter, so that the forward target weight parameter is used as a regression coefficient, and regression processing is performed on the first driving data and the forward driving data to obtain forward regression data, which is the correlation between the forward distance to be solved and the driving data of the vehicle. And carrying out regression analysis processing based on the road surface data and the second target weight parameters, so that the second target weight parameters are used as regression coefficients to carry out regression on the road surface data of the sensors, and obtaining the correlation between the vehicle distance to be solved and the data acquired by the sensors, namely second regression data. According to the correlation between the forward distance to be solved and the running data of the first vehicle and the forward vehicle and the correlation between the forward distance to be solved and the data acquired by each sensor, the forward distance between the first vehicle and the vehicle in front of the first vehicle can be calculated more accurately.
And carrying out regression analysis processing according to the first traveling data, the backward traveling data and the backward target weight parameter, so that the backward target weight parameter is used as a regression coefficient, and carrying out regression processing on the first traveling data and the backward traveling data to obtain backward regression data which is the correlation between the backward distance to be solved and the traveling data of the vehicle. According to the correlation between the backward distance to be solved and the running data of the first vehicle and the backward vehicle and the correlation between the backward distance to be solved and the data acquired by each sensor, the backward distance between the first vehicle and the vehicle behind the first vehicle can be calculated more accurately.
In one embodiment, the method further comprises: and prompting when at least one of the forward distance and the backward distance meets the distance prompting condition.
The distance prompt condition may specifically be that the forward distance between the vehicles is smaller than a distance threshold, or that the backward distance between the vehicles is smaller than the distance threshold.
When the forward distance between the first vehicle and the forward vehicle meets the distance prompt condition, the distance between the first vehicle and the forward vehicle is short, and the first vehicle has potential safety hazards of collision with the forward vehicle, so that the first vehicle is prompted through the vehicle-mounted terminal of the first vehicle to effectively prevent collision with the forward vehicle. When the backward distance between the first vehicle and the backward vehicle meets the distance prompt condition, the distance between the first vehicle and the backward vehicle is short, the first vehicle has potential safety hazards of collision with the backward vehicle, and the vehicle-mounted terminal of the first vehicle prompts the first vehicle to effectively prevent the first vehicle from colliding with the backward vehicle.
As shown in fig. 7, the vehicle-mounted terminal of the first vehicle performs data acquisition on the first vehicle to obtain first driving data. The forward vehicle and the backward vehicle of the first vehicle also acquire data through the forward vehicle and the backward vehicle to obtain forward running data and backward running data. And the vehicle machine end of the second vehicle acquires data of the second vehicle to obtain second driving data. And the sensors on the road acquire data aiming at the road to obtain road surface data. When a first vehicle passes by the sensor 2, the sensor 2 and the forward sensor 3 of the sensor 2, and the backward sensor 1 of the sensor 2 are regarded as a plurality of sensors corresponding to the first vehicle. And each sensor performs data acquisition to obtain each road surface data. The forward distance between the first vehicle and the forward vehicle, and the backward distance between the first vehicle and the backward vehicle are calculated in the processing manner of the above-described embodiment. When the forward distance or the backward distance is less than 1 meter, no prompt is sent out. When the forward distance or the backward distance is smaller than 1 meter, a prompt is sent out through the vehicle machine end to remind the vehicle owner of the first vehicle of hidden danger of vehicle collision.
In one embodiment, there is provided a vehicle data processing method applied to a computer device, including:
Pre-training stage:
acquiring first sample travel data of a first sample vehicle, forward sample travel data of a forward sample vehicle of the first sample vehicle, and backward sample travel data of a backward sample vehicle of the first sample vehicle; the forward sample vehicle travel position on the sample road is before the first sample vehicle travel position, and the backward sample vehicle travel position on the sample road is after the first sample vehicle travel position.
Sample road surface data acquired by a plurality of sample sensors respectively aiming at a sample road is acquired, wherein the plurality of sample sensors are related to a first sample vehicle.
Performing feature extraction of a plurality of first dimensions based on the first sample driving data to obtain first sample features of the first sample driving data in each first dimension; the plurality of first dimensions includes a vehicle travel dimension and a vehicle attribute dimension; and respectively extracting the characteristics of the forward sample running data and the backward sample running data in a plurality of first dimensions to obtain the forward sample characteristics of the forward sample running data in each first dimension and the backward sample characteristics of the backward sample running data in each first dimension.
Determining a forward association relationship between the first sample vehicle and the forward sample vehicle according to each first sample characteristic and each forward sample characteristic; and determining a backward association relationship between the first sample vehicle and the backward sample vehicle according to each first sample characteristic and each backward sample characteristic.
Respectively extracting the characteristics of a plurality of second dimensions based on each sample pavement data to obtain sample pavement characteristics of each sample pavement data respectively corresponding to each second dimension; the plurality of second dimensions includes a road surface travel dimension and a road surface attribute dimension; and determining a second association relationship among the plurality of sample sensors according to the plurality of sample pavement characteristics corresponding to each sample pavement data.
A forward initial weight parameter of the forward sample vehicle, a backward initial weight parameter of the backward sample vehicle, and a second initial weight parameter of each sample sensor are obtained.
Based on the forward initial weight parameters and each second initial weight parameter, combining the first sample characteristics, the forward sample characteristics and the pavement characteristics of each sample, and determining the forward initial prediction distance between the first sample vehicle and the forward sample vehicle;
and determining a backward initial prediction distance between the first sample vehicle and the backward sample vehicle based on the backward initial weight parameter and each second initial weight parameter by combining the first sample characteristic, the backward sample characteristic and each sample pavement characteristic.
Based on the forward initial weight parameter and the forward initial prediction distance, carrying out iterative processing on the weight parameter by combining the forward expected distance, the first sample characteristic, the forward sample characteristic and the forward association relation to obtain a forward target weight parameter corresponding to the forward sample vehicle;
And carrying out iterative processing on the weight parameters based on the backward initial weight parameters and the backward initial prediction distance and combining the backward expected distance, the first sample characteristics, the backward sample characteristics and the backward association relation to obtain backward target weight parameters corresponding to the backward sample vehicle.
Determining a global expected distance according to the forward expected distance and the backward expected distance; determining a global initial predicted distance according to the forward initial predicted distance and the backward initial predicted distance; and carrying out iterative processing on the weight parameters based on each second initial weight parameter and the global initial prediction distance and combining the global expected distance, the pavement characteristics of each sample and the second association relation to obtain a second target weight parameter corresponding to each sample sensor.
The application stage comprises the following steps:
acquiring a first feature of a first vehicle, a forward feature of a forward vehicle of the first vehicle, and a backward feature of a backward vehicle of the first vehicle; the forward vehicle travel position on the target road is before the first vehicle travel position, and the backward vehicle travel position on the target road is after the first vehicle travel position.
Target road surface data acquired by a plurality of sensors associated with a first vehicle for a target road, respectively, is acquired. And acquiring a forward target weight parameter obtained by pre-training corresponding to the forward vehicle, a backward target weight parameter obtained by pre-training corresponding to the backward vehicle and a second target weight parameter obtained by pre-training corresponding to each sensor according to the driving sequence.
Carrying out regression analysis processing according to the first characteristic, the forward characteristic and the forward target weight parameter to obtain forward regression data; carrying out regression analysis processing based on the road surface features and the second target weight parameters to obtain second regression data; and determining the forward distance between the first vehicle and the forward vehicle according to the forward regression data and the second regression data.
Carrying out regression analysis processing according to the first characteristic, the backward characteristic and the backward target weight parameter to obtain backward regression data; and determining the backward distance between the first vehicle and the backward vehicle according to the backward regression data and the second regression data.
And prompting when at least one of the forward distance and the backward distance meets the distance prompting condition.
In this embodiment, in the pre-training stage, first sample running data of a first sample vehicle and second sample running data of a second sample vehicle are acquired, running sequences of the first sample vehicle and the second sample vehicle exist on a sample road, and first sample characteristics of the first sample running data in a vehicle running dimension and a vehicle attribute dimension are extracted. And carrying out feature extraction of a plurality of first dimensions based on the second sample driving data to obtain second sample features of the second sample driving data in the vehicle driving dimension and the vehicle attribute dimension respectively, so that a first association relation between the first sample vehicle and the second sample vehicle can be accurately constructed and determined according to the features of the first sample vehicle and the second sample vehicle in the vehicle driving dimension and the vehicle attribute dimension, and the feature correlation of the first sample vehicle and the second sample vehicle in different dimensions can be accurately represented through the association relation. And respectively extracting the characteristics of a plurality of second dimensions based on the pavement data of each sample to obtain the corresponding characteristics of each sample sensor in the pavement driving dimension and the pavement attribute dimension, so that the second association relation of the plurality of sample sensors can be accurately constructed according to the corresponding characteristics of each sample sensor in the pavement driving dimension and the pavement attribute dimension, and the characteristic correlation of the plurality of sample sensors in different dimensions can be accurately represented through the second association relation.
Based on the first initial weight parameter of the second sample vehicle and the second initial weight parameter of each sample sensor, the first sample characteristic, the second sample characteristic, the pavement characteristics of each sample, the first association relationship and the second association relationship are combined, the weight parameter adjustment of the cross iteration is carried out, the weight calculation can be carried out by combining the correlation between the characteristics of each vehicle and the correlation between the pavement characteristics collected by each sensor, the first target weight parameter of the second sample vehicle and the second target weight parameter of each sample sensor are obtained, the problem of inaccurate weight parameters caused by repeated superposition calculation of the same characteristics is avoided, and the accuracy of the weight parameter calculation is effectively improved. The distance between at least two vehicles with the running sequence is determined by using the first target weight parameter and the second target weight parameter, so that the accuracy of the distance prediction between the vehicles can be effectively improved.
In the application stage, regression analysis processing is carried out according to the first feature, the forward feature and the forward target weight parameter, so that the forward target weight parameter is used as a regression coefficient, the first feature and the forward feature are subjected to regression processing, and the correlation between the forward distance to be solved and the feature of the vehicle, namely the forward regression feature, is obtained. And carrying out regression analysis processing based on the road surface features and the second target weight parameters, so that the second target weight parameters are used as regression coefficients to carry out regression on the road surface features of the sensors, and obtaining the correlation between the vehicle distance to be solved and the features acquired by the sensors, namely the second regression features. The forward distance between the first vehicle and the vehicle preceding the first vehicle can be calculated more accurately according to the correlation between the forward distance to be solved and the first vehicle, the characteristics of the forward vehicle and the correlation between the characteristics acquired by each sensor.
And carrying out regression analysis processing according to the first feature, the backward feature and the backward target weight parameter, thereby taking the backward target weight parameter as a regression coefficient, carrying out regression processing on the first feature and the backward feature, and obtaining a backward regression feature which is the correlation between the backward distance to be solved and the feature of the vehicle. According to the correlation between the backward distance to be solved and the characteristics of the first vehicle and the backward vehicle and the correlation between the backward distance to be solved and the characteristics acquired by each sensor, the backward distance between the first vehicle and the vehicle behind the first vehicle can be calculated more accurately.
When the forward distance between the first vehicle and the forward vehicle meets the distance prompt condition, the distance between the first vehicle and the forward vehicle is short, the first vehicle has potential safety hazards colliding with the forward vehicle, when the backward distance between the first vehicle and the backward vehicle meets the distance prompt condition, the distance between the first vehicle and the backward vehicle is short, the first vehicle has potential safety hazards colliding with the backward vehicle, and the vehicle-mounted terminal of the first vehicle prompts to effectively prevent the first vehicle from colliding with the backward vehicle.
In one embodiment, an application scenario of a vehicle data processing method is provided, and the application scenario is applied to a vehicle machine side of a vehicle, and if a distance between a first sample vehicle and a forward sample vehicle of the first sample vehicle and a backward sample vehicle of the first sample vehicle needs to be calculated in the application scenario, a processing flow of the application scenario is shown in fig. 8, and the application scenario includes model training prediction of step S802 to step S818 and model application stages of step S820 to step S822.
Model training prediction refers to using a machine learning or deep learning model, performing iterative computation on feature data and labels of the model by adopting a gradient descent method to obtain model weights, and predicting by using the model weights and prediction samples, wherein the model weights and the prediction samples are specifically as follows:
step S802, data input stage. The data input section includes three sections: road information data collected by a vehicle terminal; secondly, the sample vehicle configures parameter data; thirdly, road surface data acquired by the sample sensor. Road information data collected by the vehicle machine end and sample vehicle self-configuration parameter data form sample driving data. Namely, road information data collected by a vehicle end of a first sample vehicle and vehicle self-configuration parameter data form first sample running data, road information data collected by a vehicle end of a forward sample vehicle and vehicle self-configuration parameter data form forward sample running data, and road information data collected by a vehicle end of a backward sample vehicle and vehicle self-configuration parameter data form backward sample running data.
The method comprises the following steps:
the road information data collected by the vehicle terminal is as follows: the vehicle speed, the front and rear vehicle speed, road surface images acquired by the vehicle, the distance between the front and rear vehicles, the distance between the left and right vehicles, vehicle POI data and the like.
The vehicle configuration parameters include: vehicle type, vehicle length, height, width, number of vehicles on demand, highest speed per hour of vehicle, vehicle displacement, etc.
The roadside sensor collects the data as follows: an image of a road surface of a road section, road information, average vehicle speed, speed limit data, overspeed data, traffic accident image, traffic flow data, waiting time length, road section POI information, and the like.
Step S804, a total distance model and a total loss function construction stage.
Total distance model: d=y+z+e (1)
Total loss model:
step S806, forward distance model and loss function construction stage.
Forward distance formula, i.e. forward distance model:
forward loss function model:
step S808, a backward distance model and a loss function construction stage.
A backward distance formula, namely a backward distance model:
model of backward loss function:
step S810, a forward-backward distance loss cross model construction phase, i.e., a distance loss cross model construction phase of a forward vehicle and a backward vehicle.
Forward-backward distance loss crossover model:
wherein during the pre-training phase D represents the total distance between the forward sample vehicle and the backward sample vehicle, i.e. the global desired distance. Z represents the backward distance between the current sample vehicle and the backward sample vehicle, i.e., the backward desired distance. The current sample vehicle is the first sample vehicle. Y represents the forward distance between the current sample vehicle and the forward sample vehicle, i.e., the forward desired distance. k represents a kth vehicle (k=0 represents a current sample vehicle, k= -1 represents a backward sample vehicle of the current sample vehicle, and k=1 represents a forward sample vehicle of the current sample vehicle). X is X C,k Sample characteristics corresponding to sample driving data collected at the vehicle machine side of the kth vehicle, including characteristics of road data and characteristics of configuration data of the vehicle, W C,k Weight parameters corresponding to vehicle body end of forward sample vehicle are represented, V C,k And (5) representing the weight parameters corresponding to the vehicle-mounted end of the backward sample vehicle. W (W) R,i Weight parameter X corresponding to ith sample sensor of road section R,i And representing the sample road surface characteristics corresponding to the sample road surface data acquired by the ith sample sensor of the road section. n denotes that the current sample vehicle has n sample sensors passing through the road segment, The weight parameter representing the i-th sample sensor (i= -1 represents the backward sample sensor of the current sample sensor, i=0 represents the current sample sensor, and i=1 represents the forward sample sensor of the current sample sensor). E represents a random number sequence with total distance obeying a standard normal distribution, E Y Representing a sequence of random numbers whose forward distances follow a standard normal distribution, E Z A random number sequence representing that the backward distance obeys a standard normal distribution. L (L) D Representing the total loss function>Distance loss function representing forward sample vehicle, < ->Distance loss function representing a backward sample vehicle, < ->Represents the cross-loss function, alpha, between forward and backward distances k Representing the parameter of the kth vehicle beta i Representing the parameters of the ith sample sensor.
It will be appreciated that during the application phase, Z represents the rearward distance between the current vehicle and the rearward vehicle. Y represents the forward distance between the current vehicle and the forward vehicle. k represents a kth vehicle (k=0 represents a current vehicle, k= -1 represents a backward vehicle of the current vehicle, k=1 represents a forward vehicle of the current vehicle, and the current vehicle is a first vehicle). X is X C,k The characteristic of the running data acquired by the vehicle machine side of the kth vehicle comprises a first characteristic of the first vehicle, a forward characteristic of the forward vehicle or a backward characteristic of the backward vehicle. W (W) C,k Representing weight parameters corresponding to the vehicle-machine side of the forward vehicle, namely target weight parametersV C,k Weight parameters corresponding to the vehicle body side of the backward vehicle, namely target weight parameters +.>W R,i Weight parameter corresponding to the i-th sensor representing the road section, i.e. target weight parameter +.>X R,i And representing the road surface characteristics corresponding to the road surface data acquired by the ith sensor of the road section.
Step S812, federal learning model construction phase. Federal learning combines distributed machine learning, cryptography, incentive mechanisms based on financial rules, and game theory to solve the problem of use of decentralized data. Deducing according to the distance model and the loss function of S804-S210, thereby constructing a federal learning gradient descent method parameter estimation model based on the correlation of the front and rear vehicle characteristics and the correlation of the adjacent sensor characteristics, wherein the federal learning gradient descent method parameter estimation model is as follows:
wherein, in the pre-training stage, W R Representing the weight parameter, W, corresponding to the sample sensor C Representing weight parameters, V, corresponding to a forward sample vehicle C And representing the weight parameters corresponding to the backward sample vehicle.Representing the total predicted distance between the forward sample vehicle and the rearward sample vehicle. />Representing the forward predicted distance between the current sample vehicle and the forward sample vehicle. />Representing a backward predicted distance between the current sample vehicle and the backward sample vehicle.
It will be appreciated that, in the application phase, the forward distance between the current vehicle and the forward vehicle is calculated using the above formula (3), i.e., the forward distance, and the backward distance between the current vehicle and the backward vehicle is calculated using the above formula (5), i.e., the backward distance. The current vehicle is the first vehicle. In the application phase, W R Representing the weight parameter, W, corresponding to the sensor C Representing weight parameters, V, corresponding to a forward vehicle C And the weight parameters corresponding to the backward vehicles are represented.
Step S814, the adjacent sensor correlation covariance matrix construction phase. And respectively extracting the characteristics of the first sample driving data, the second sample driving data and the pavement data of each sample to correspondingly obtain the first sample characteristics, the second sample characteristics and the pavement characteristics of each sample.
Sample road surface characteristics X input to each sample sensor R Calculating the correlation between the sample pavement characteristics of each sample sensor, and obtaining a correlation covariance matrix of the sample pavement characteristics of the sample sensors:
step S816, a correlation covariance matrix construction phase between the forward sample vehicle and the backward sample vehicle. Inputting sample characteristics of each of a forward sample vehicle, a first sample vehicle, and a backward sample vehicle
X C,k (k= -1,0, 1) calculating a forward correlation covariance matrix between the forward sample vehicle and the first sample vehicle:
and a backward correlation covariance matrix between the backward sample vehicle and the first sample vehicle:
step S818, model weight cross iteration solving stage. First initialize the weight parameter W of the forward sample vehicle C Obtaining forward initial weight parameters W C,0 Initializing the weight parameter W of each sample sensor R Obtaining a second initial weight parameter W R,0 Thereby according to the forward distance formulaObtain forward prediction distance->Initial value of (i.e. forward initial prediction distance +)>Reinitializing the weight parameter V of the backward sample vehicle C Obtaining backward initial weight parameter V C,0 Thereby according to the backward distance formula +.>Obtain backward prediction distance->Initial value of (i.e. backward initial prediction distance +)>Thereby obtaining the global initial prediction distance of the forward sample vehicle and the backward sample vehicleReinitializing the weight parameters W of the sample sensors R Obtaining a second initial weight parameter W R,0 Sample road surface characteristic X substituted into sample sensor R And a correlation matrix Ω by the formula +.>Calculating the weight parameter W of the first step R,1 The method comprises the steps of carrying out a first treatment on the surface of the Substituting the weight parameter W of the sample sensor obtained in the first step R,1 Thereby obtaining Y R The value Y of (2) R,1 Substituting the characteristic data X of the sample vehicle C,Y Correlation matrix Σ C,Y By the formulaCalculating a weight parameter W for a first step of a forward sample vehicle C,1 The method comprises the steps of carrying out a first treatment on the surface of the Substituting the weight parameter W of the sample sensor obtained in the first step R,1 Thereby obtaining Z R The value Z of (2) R,1 Substituting the characteristic X of the sample vehicle C,Z Correlation matrix sigma C,Z By the formula->Calculating a weight parameter V for a first step of a backward sample vehicle C,1 . And so on, obtaining target weight parameters corresponding to each sample sensor by a cross iteration method>Target weight parameter corresponding to forward sample vehicle>Target weight parameter corresponding to backward sample vehicle +.>
Step S820, the distance prediction phase, i.e. the application phase. Inputting target weight parameters obtained by cross iteration solution of vehicle-machine end in step S818And->Inputting a road surface characteristic corresponding to the road surface data acquired by each sensor, a first characteristic corresponding to the first driving data of the current vehicle (namely, a first vehicle) and a forward characteristic corresponding to the forward driving data of the forward vehicle, and substituting a forward distance formula: />And obtaining the forward distance output by the forward distance model. Inputting the road surface characteristics corresponding to the road surface data acquired by each sensor, the first characteristics corresponding to the first driving data of the current vehicle and the backward characteristics corresponding to the backward driving data of the backward vehicle, and substituting the characteristics into a backward distance formula And obtaining the backward distance output by the backward distance model.
Step S822, safe distance judging and prompting. The forward distance and the backward distance calculated in step S820 are input. Setting a distance threshold (for example, 5 meters), and if the forward distance is greater than the threshold, indicating that the forward distance between the current vehicle and the forward vehicle is safe, and not needing prompting; otherwise, indicating that the forward distance between the current vehicle and the forward vehicle is unsafe, and prompting the vehicle owner through vehicle-mounted voice at the vehicle end or a screen at the vehicle end; if the backward distance is greater than the threshold value, the backward distance between the current vehicle and the backward vehicle is safe, and prompt is not needed; otherwise, the backward distance between the current vehicle and the backward vehicle is unsafe, and the vehicle owner is prompted through the vehicle-mounted voice at the vehicle end or the vehicle-mounted screen at the vehicle end.
The vehicle data processing method of the embodiment is a vehicle-road cooperation safety distance prediction scheme based on multiple vehicles and multiple sensors, and can solve the problem that data acquired by front and rear vehicles and adjacent road side sensors have correlation on the basis of federal learning of vehicle-road cooperation. Specifically, a correlation matrix of the sensors is constructed based on the consideration that there is correlation between data acquired by adjacent sensors, but there is no correlation between data acquired by non-adjacent sensors. And, consider the correlation problem of the front and back car driving characteristic, and based on the correlation of the front and back car, have constructed the covariance matrix of the front and back car correlation. The embodiment adopts a mechanism of a federal learning algorithm, and can effectively protect the data security problem between the vehicle and the sensor. Based on longitudinal federal learning of front and rear vehicles, longitudinal federal learning of adjacent sensors and transverse federal learning between vehicles and sensors, a hybrid federal learning model is constructed, comprising a front and rear vehicle total distance model and a loss function, a front vehicle distance model and a loss function, a rear vehicle distance model and a loss function and a front and rear vehicle distance crossover loss function. And according to the constructed correlation matrix and each loss function, a gradient descent method is used to push to an iteration formula for obtaining weight parameter estimation based on federal learning, so that optimal weight parameters of forward and backward vehicles and each sensor are solved, and the problem of safe distance prediction with correlation between multiple vehicles and multiple sensors is effectively solved. The method in the embodiment fully considers the influence of the correlation of the front and rear vehicles and the correlation of the adjacent sensors on the safety distance of the front and rear vehicles, can calculate the safety distance of the front and rear vehicles more accurately, and effectively prevents the front and rear vehicles from collision.
It should be understood that, although the steps in the flowcharts related to the above embodiments are sequentially shown as indicated by arrows, these steps are not necessarily sequentially performed in the order indicated by the arrows. The steps are not strictly limited to the order of execution unless explicitly recited herein, and the steps may be executed in other orders. Moreover, at least some of the steps in the flowcharts described in the above embodiments may include a plurality of steps or a plurality of stages, which are not necessarily performed at the same time, but may be performed at different times, and the order of the steps or stages is not necessarily performed sequentially, but may be performed alternately or alternately with at least some of the other steps or stages.
Based on the same inventive concept, the embodiment of the application also provides a vehicle data processing device for realizing the vehicle data processing method. The implementation of the solution provided by the device is similar to the implementation described in the above method, so the specific limitation of one or more embodiments of the vehicle data processing device provided below may refer to the limitation of the vehicle data processing method hereinabove, and will not be repeated herein.
In one embodiment, as shown in fig. 9, there is provided a vehicle data processing apparatus 900 including:
a sample travel data acquisition module 902 for acquiring first sample travel data of a first sample vehicle and second sample travel data of a second sample vehicle; the first sample vehicle and the second sample vehicle have a driving order on the sample road.
A sample road surface data obtaining module 904, configured to obtain sample road surface data collected by a plurality of sample sensors related to the first sample vehicle for sample roads, respectively.
An association determining module 906, configured to determine a first association between the first sample vehicle and the second sample vehicle according to the first sample driving data and the second sample driving data; and determining a second association relationship among the plurality of sample sensors according to the plurality of sample pavement data.
The weight parameter acquisition module 908 is configured to acquire a first initial weight parameter of the second sample vehicle and a second initial weight parameter of each sample sensor.
The weight parameter adjustment module 910 is configured to perform cross iterative weight parameter adjustment based on the first initial weight parameter and each second initial weight parameter, and combine the first sample driving data, the second sample driving data, each sample road surface data, the first association relationship and the second association relationship to obtain a first target weight parameter corresponding to the second sample vehicle and a second target weight parameter corresponding to each sample sensor; the first target weight parameter and the second target weight parameter are used to determine a distance between at least two vehicles for which a driving sequence exists.
In this embodiment, first sample traveling data of a first sample vehicle and second sample traveling data of a second sample vehicle are acquired, and traveling orders of the first sample vehicle and the second sample vehicle exist on a sample road to determine a first association relationship between the first sample vehicle and the second sample vehicle based on the first sample traveling data and the second sample traveling data, thereby determining correlations between the traveling data acquired by the first sample vehicle and the second sample vehicle, respectively. Sample road surface data acquired by a plurality of sample sensors related to the first sample vehicle respectively for the sample road are acquired, so that a second association relationship among the plurality of sample sensors is determined according to the plurality of sample road surface data, and accordingly correlation among the plurality of sample sensors on the road surface data is determined.
Based on the first initial weight parameter of the second sample vehicle and the second initial weight parameter of each sample sensor, the first sample driving data, the second sample driving data, the pavement data of each sample, the first association relationship and the second association relationship are combined to carry out cross iterative weight parameter adjustment, and the weight calculation can be carried out by combining the correlation between the driving data of each vehicle and the correlation between the pavement data acquired by each sensor to obtain the first target weight parameter of the second sample vehicle and the second target weight parameter of each sample sensor, so that the problem of inaccurate weight parameters caused by repeated superposition calculation of the same data is avoided, and the accuracy of weight parameter calculation is effectively improved. The distance between at least two vehicles with the running sequence is determined by using the first target weight parameter and the second target weight parameter, so that the accuracy of the distance prediction between the vehicles can be effectively improved.
In one embodiment, the association determining module 906 is further configured to perform feature extraction of a plurality of first dimensions based on the first sample driving data, to obtain first sample features of the first sample driving data in each first dimension; the plurality of first dimensions includes a vehicle travel dimension and a vehicle attribute dimension; performing feature extraction of a plurality of first dimensions based on the second sample driving data to obtain second sample features of the second sample driving data in each first dimension; and determining a first association relationship between the first sample vehicle and the second sample vehicle according to each first sample characteristic and each second sample characteristic.
In this embodiment, the plurality of first dimensions include a vehicle running dimension and a vehicle attribute dimension, and feature extraction of the plurality of first dimensions is performed based on the first sample running data to extract first sample features of the first sample running data in the vehicle running dimension and the vehicle attribute dimension, respectively. And carrying out feature extraction of a plurality of first dimensions based on the second sample driving data to obtain second sample features of the second sample driving data in the vehicle driving dimension and the vehicle attribute dimension respectively, so that a first association relation between the first sample vehicle and the second sample vehicle can be accurately constructed and determined according to the features of the first sample vehicle and the second sample vehicle in the vehicle driving dimension and the vehicle attribute dimension, and the feature correlation of the first sample vehicle and the second sample vehicle in different dimensions can be accurately represented through the association relation.
In one embodiment, the association determining module 906 is further configured to determine a sub-association relationship between each first sample feature and each second sample feature; and determining a first association relationship between the first sample vehicle and the second sample vehicle according to each sub-association relationship.
In this embodiment, sub-association relationships between each first sample feature and each second sample feature are determined, so that correlation between any two sample features is represented by the sub-association relationships, and feature correlations of the first sample vehicle and the second sample vehicle on each feature are accurately reflected according to the first association relationship determined by each sub-association relationship.
In one embodiment, the association determining module 906 is further configured to perform feature extraction of a plurality of second dimensions based on each sample road surface data, to obtain sample road surface features of each sample road surface data corresponding to each second dimension; the plurality of second dimensions includes a road surface travel dimension and a road surface attribute dimension; and determining a second association relationship among the plurality of sample sensors according to the plurality of sample pavement characteristics corresponding to each sample pavement data.
In this embodiment, the plurality of second dimensions include a road surface driving dimension and a road surface attribute dimension, and feature extraction of the plurality of second dimensions is performed based on each sample road surface data, so as to obtain features corresponding to each sample sensor in the road surface driving dimension and the road surface attribute dimension, so that a second association relationship of the plurality of sample sensors can be accurately constructed according to features corresponding to each sample sensor in the road surface driving dimension and the road surface attribute dimension, and feature correlation of the plurality of sample sensors in different dimensions can be accurately represented through the second association relationship.
In one embodiment, the weight parameter adjustment module 910 is further configured to determine an initial predicted distance between the first sample vehicle and the second sample vehicle based on the first initial weight parameter and each of the second initial weight parameters, in combination with the first sample driving data, the second sample driving data, and the respective sample road surface data; acquiring an expected distance between a first sample vehicle and a second sample vehicle, and performing iterative processing of weight parameters based on a first initial weight parameter and an initial predicted distance by combining the expected distance, first sample running data, second sample running data and a first association relationship to obtain a first target weight parameter corresponding to the second sample vehicle; and carrying out iterative processing on the weight parameters based on each second initial weight parameter and the initial predicted distance by combining the expected distance, the pavement data of each sample and the second association relation to obtain a second target weight parameter corresponding to each sample sensor.
In this embodiment, based on the first initial weight parameter and each second initial weight parameter, the first sample traveling data, the second sample traveling data, and the respective sample road surface data are combined to predict the distance between the first sample vehicle and the second sample vehicle, so as to obtain an initial predicted distance. The expected distance between the first sample vehicle and the second sample vehicle is obtained to be used as a distance label in iteration, the difference between the real distance and the distance predicted based on each weight parameter can be determined, the weight parameter of the second sample vehicle is iterated by combining the running data of each vehicle and the data correlation between the running data, and the weight parameter of the second sample vehicle is continuously optimized in multiple iterations, so that the first target weight parameter of the second sample vehicle is accurately obtained. And iterating the weight parameters of each sample sensor by combining the sample pavement data acquired by each sample sensor and the data correlation between the sample pavement data, so that the weight parameters of each sample sensor are continuously optimized in a plurality of iterations, and a second target weight parameter corresponding to each sample sensor is accurately obtained.
In one embodiment, the second sample vehicle includes a forward sample vehicle and a backward sample vehicle of the first sample vehicle; the driving position of the forward sample vehicle on the sample road is before the driving position of the first sample vehicle, and the driving position of the backward sample vehicle on the sample road is after the driving position of the first sample vehicle; the second sample travel data includes forward sample travel data of a forward sample vehicle and backward sample travel data of a backward sample vehicle; the initial prediction distance comprises a forward initial prediction distance and a backward initial prediction distance between the forward sample vehicle and the backward sample vehicle and between the backward sample vehicle and the first sample vehicle respectively; the expected distance comprises a forward expected distance and a backward expected distance between the forward sample vehicle and the backward sample vehicle and the first sample vehicle respectively; the first association relation comprises a forward association relation and a backward association relation between the forward sample vehicle and the backward sample vehicle and the first sample vehicle respectively; the first initial weight parameters comprise forward initial weight parameters of the forward sample vehicle and backward initial weight parameters of the backward sample vehicle; the first target weight parameters include forward target weight parameters of the forward sample vehicle and backward target weight parameters of the backward sample vehicle.
In this embodiment, based on the forward initial weight parameter, the backward initial weight parameter, and each second initial weight parameter, the distances between the first sample vehicle and the forward sample vehicle and between the first sample vehicle and the backward sample vehicle are predicted by combining the first sample travel data, the forward sample travel data, the backward sample travel data, and the respective sample road surface data, so as to obtain a forward predicted distance and a backward predicted distance. The method comprises the steps of obtaining a forward expected distance and a backward expected distance as distance labels in iteration, determining the difference between the real forward distance and the forward distance predicted based on the weight parameters of a forward sample vehicle and the weight parameters of each sample sensor, and the difference between the real backward distance and the backward distance predicted based on the weight parameters of a backward sample vehicle and the weight parameters of each sample sensor, combining the data correlation between the running data of each vehicle and the running data, and iterating the weight parameters of the forward sample vehicle, the backward sample vehicle and each sample sensor according to the data correlation between the running data of each vehicle and the data correlation between the running data of each sample sensor, so that the weight parameters of each sample vehicle, each sample sensor are continuously optimized in multiple iterations, and the target weight parameters corresponding to each of the forward sample vehicle, the backward sample vehicle and each sample sensor are accurately obtained.
In one embodiment, the weight parameter adjustment module 910 is further configured to determine, based on the forward initial weight parameter and each of the second initial weight parameters, a forward initial prediction distance between the first sample vehicle and the forward sample vehicle in combination with the first sample driving data, the forward sample driving data, and the respective sample road surface data; and determining a backward initial prediction distance between the first sample vehicle and the backward sample vehicle by combining the first sample driving data, the backward sample driving data and the pavement data of each sample based on the backward initial weight parameters and each second initial weight parameter.
In this embodiment, based on the forward initial weight parameter and each second initial weight parameter, the forward initial prediction distance between the first sample vehicle and the forward sample vehicle can be accurately calculated by combining the first sample travel data, the forward sample travel data, and the respective sample road surface data. Based on the backward initial weight parameters and each second initial weight parameter, the backward initial prediction distance between the first sample vehicle and the backward sample vehicle can be accurately calculated by combining the first sample driving data, the backward sample driving data and the sample road surface data, so that the weight parameters of the forward and backward vehicles and the weight parameters of the sample sensor are adjusted according to the prediction distance between the first sample vehicle and the forward and backward vehicles to obtain the optimal solution of each weight parameter.
In one embodiment, the weight parameter adjustment module 910 is further configured to perform iterative processing of the weight parameter based on the forward initial weight parameter and the forward initial predicted distance, and in combination with the forward expected distance, the first sample traveling data, the forward sample traveling data, and the forward association relationship, to obtain a forward target weight parameter corresponding to the forward sample vehicle; and carrying out iterative processing on the weight parameters based on the backward initial weight parameters and the backward initial prediction distance and combining the backward expected distance, the first sample driving data, the backward sample driving data and the backward association relation to obtain backward target weight parameters corresponding to the backward sample vehicle.
In this embodiment, based on the forward initial weight parameter and the forward initial prediction distance, the forward expected distance, the first sample running data, the forward sample running data and the forward association relation are combined to perform iterative processing of the weight parameter, so that the weight parameter of the forward sample vehicle is optimized by combining the data in each aspect in multiple iterations, and an optimal solution of the weight parameter corresponding to the forward sample vehicle, that is, the forward target weight parameter, is obtained. And (3) based on the backward initial weight parameter and the backward initial prediction distance, carrying out iterative processing of the weight parameter by combining the backward expected distance, the first sample driving data, the backward sample driving data and the backward association relation, and optimizing the weight parameter of the backward sample vehicle by combining the data in each aspect in multiple iterations to obtain an optimal solution of the weight parameter corresponding to the backward sample vehicle, namely the backward target weight parameter.
In one embodiment, the weight parameter adjustment module 910 is further configured to perform iterative processing of the weight parameters based on each second initial weight parameter, the forward initial predicted distance and the backward initial predicted distance, and in combination with the forward expected distance and the backward expected distance, each sample road surface data and the second association relationship, to obtain a second target weight parameter corresponding to each sample sensor.
In this embodiment, based on each second initial weight parameter, the forward initial predicted distance and the backward initial predicted distance, iterative processing of the weight parameters is performed in combination with the forward expected distance and the backward expected distance, the pavement data of each sample and the second association relationship, so that the weight parameters of each sample sensor are optimized in multiple iterations in combination with the data of each aspect, and an optimal solution of the weight parameters corresponding to each sample sensor, namely each second target weight parameter, is obtained.
In one embodiment, the weight parameter adjustment module 910 is further configured to determine a global desired distance according to the forward desired distance and the backward desired distance; determining a global initial predicted distance according to the forward initial predicted distance and the backward initial predicted distance; and carrying out iterative processing on the weight parameters based on each second initial weight parameter and the global initial prediction distance and combining the global expected distance, the pavement data of each sample and the second association relation to obtain a second target weight parameter corresponding to each sample sensor.
In this embodiment, the global initial predicted distance between the front and rear vehicles of the first sample vehicle is calculated according to the forward expected distance and the backward expected distance, and the global expected distance between the front and rear vehicles of the first sample vehicle is calculated according to the forward initial predicted distance and the backward initial predicted distance, so that the influence of the front and rear vehicles on the first sample vehicle can be considered at the same time. And (3) carrying out iterative processing of the weight parameters based on each second initial weight parameter and the global initial prediction distance and combining the global expected distance, each sample pavement data and the second association relation, so that the weight parameters of each sample sensor are optimized according to the data of the first sample vehicle, the data of the front and rear vehicles, the correlation among the sensors, the vehicle data acquired by each sample sensor and other multi-aspect data in multiple iterations, and the optimal solution of the weight parameters corresponding to each sample sensor is obtained, so that each obtained second target weight parameter is suitable for a scene of distance prediction between the first vehicle and the front and rear vehicles when the first vehicle has the front and rear vehicles at the same time.
In one embodiment, as shown in fig. 10, there is provided a vehicle data processing apparatus 1000 including:
a driving data obtaining module 1002, configured to obtain first driving data of a first vehicle and second driving data of a second vehicle; the first vehicle and the second vehicle have a traveling sequence on the target road.
The road surface data acquisition module 1004 is configured to acquire target road surface data acquired by a plurality of sensors related to the first vehicle for the target road, respectively.
The target weight obtaining module 1006 is configured to obtain, according to a driving order, a first target weight parameter obtained by pre-training corresponding to the second vehicle and a second target weight parameter obtained by pre-training corresponding to each sensor;
the processing module 1008 is configured to perform regression analysis processing based on the first driving data, the second driving data, and the first target weight parameter, to obtain first regression data; and carrying out regression analysis processing based on the pavement data and the second target weight parameters to obtain second regression data.
The distance determining module 1010 is configured to determine a vehicle distance between the first vehicle and the second vehicle according to the first regression data and the second regression data.
In this embodiment, second driving data of a first vehicle and a second vehicle, which have driving sequences on a target road, and first target weight parameters obtained by pre-training corresponding to the second vehicle are obtained, and regression analysis processing is performed based on the first driving data, the second driving data and the first target weight parameters, so that the first target weight parameters are used as regression coefficients to regress the first driving data and the second driving data, and correlation between a vehicle distance to be solved and driving data of each vehicle, namely, first regression data is obtained. Acquiring target road surface data which are acquired by a plurality of sensors related to a first vehicle and respectively aiming at a target road, and respectively corresponding second target weight parameters which are obtained by pretraining each sensor, and carrying out regression analysis processing based on the road surface data and the second target weight parameters, so that the second target weight parameters are used as regression coefficients to carry out regression on the road surface data of the sensors, and obtaining the correlation between the vehicle distance to be solved and the data acquired by the sensors, namely second regression data. According to the correlation between the vehicle distance to be solved and the running data of each vehicle and the correlation between the vehicle distance to be solved and the data acquired by each sensor, the vehicle distance between the first vehicle and the second vehicle can be calculated more accurately.
In one embodiment, the second vehicle includes a forward vehicle and a rearward vehicle of the first vehicle, the forward vehicle traveling on the target road before the first vehicle traveling, and the rearward vehicle traveling on the target road after the first vehicle traveling; the second traveling data includes forward traveling data of a forward vehicle and backward traveling data of a backward vehicle, the first target weight parameter includes a forward target weight parameter of the forward vehicle and a backward target weight parameter of the backward vehicle, the first regression data includes forward regression data of the forward vehicle and backward regression data of the backward vehicle, the vehicle distance includes a forward distance between the first vehicle and the forward vehicle, and a backward distance between the first vehicle and the backward vehicle.
In this embodiment, respective driving data of a first vehicle, a forward vehicle and a backward vehicle of the first vehicle, which have a driving sequence on a target road, are acquired, and respective target weight parameters are obtained through pre-training corresponding to the forward vehicle and the backward vehicle, and regression analysis processing is performed based on the driving data of each vehicle and the target weight parameters corresponding to each vehicle, so that the target weight parameters are used as regression coefficients to regress the respective driving data, and correlation between a vehicle distance to be solved and the driving data of each vehicle is obtained. Acquiring target road surface data which are acquired by a plurality of sensors related to a first vehicle and respectively aiming at a target road, and respectively corresponding second target weight parameters which are obtained by pretraining each sensor, and carrying out regression analysis processing based on the road surface data and the second target weight parameters, thereby taking the second target weight parameters as regression coefficients to carry out regression on the road surface data of the sensors so as to obtain the correlation between the vehicle distance to be solved and the data acquired by each sensor. According to the correlation between the forward distance to be solved and the running data of the forward vehicle and the correlation between the forward distance to be solved and the data acquired by each sensor, the forward distance between the first vehicle and the forward vehicle can be calculated more accurately. According to the correlation between the backward distance to be solved and the running data of the forward vehicle and the correlation between the backward distance to be solved and the data acquired by each sensor, the forward distance between the first vehicle and the backward vehicle can be calculated more accurately.
In one embodiment, the processing module 1008 is further configured to perform regression analysis according to the first driving data, the forward driving data, and the forward target weight parameter to obtain forward regression data; performing regression analysis processing according to the first driving data, the backward driving data and the backward target weight parameter to obtain backward regression data;
the distance determining module 1010 is further configured to determine a forward distance between the first vehicle and the forward vehicle according to the forward regression data and the second regression data; and determining the backward distance between the first vehicle and the backward vehicle according to the backward regression data and the second regression data.
In this embodiment, regression analysis processing is performed according to the first driving data, the forward driving data and the forward target weight parameter, so that the forward target weight parameter is used as a regression coefficient, and regression processing is performed on the first driving data and the forward driving data to obtain forward regression data, which is the correlation between the forward distance to be solved and the driving data of the vehicle. And carrying out regression analysis processing based on the road surface data and the second target weight parameters, so that the second target weight parameters are used as regression coefficients to carry out regression on the road surface data of the sensors, and obtaining the correlation between the vehicle distance to be solved and the data acquired by the sensors, namely second regression data. According to the correlation between the forward distance to be solved and the running data of the first vehicle and the forward vehicle and the correlation between the forward distance to be solved and the data acquired by each sensor, the forward distance between the first vehicle and the vehicle in front of the first vehicle can be calculated more accurately.
And carrying out regression analysis processing according to the first traveling data, the backward traveling data and the backward target weight parameter, so that the backward target weight parameter is used as a regression coefficient, and carrying out regression processing on the first traveling data and the backward traveling data to obtain backward regression data which is the correlation between the backward distance to be solved and the traveling data of the vehicle. According to the correlation between the backward distance to be solved and the running data of the first vehicle and the backward vehicle and the correlation between the backward distance to be solved and the data acquired by each sensor, the backward distance between the first vehicle and the vehicle behind the first vehicle can be calculated more accurately.
In one embodiment, the apparatus further comprises a prompt module; the prompt module is used for prompting when the distance between the vehicles meets the distance prompt condition.
In one embodiment, the apparatus further comprises a prompt module; the prompting module is used for prompting when at least one of the forward distance and the backward distance meets the distance prompting condition.
The respective modules in the above-described vehicle data processing apparatus may be implemented in whole or in part by software, hardware, and a combination thereof. The above modules may be embedded in hardware or may be independent of a processor in the computer device, or may be stored in software in a memory in the computer device, so that the processor may call and execute operations corresponding to the above modules.
In one embodiment, a computer device is provided, which may be a terminal or a server. Taking the terminal as an example, the internal structure of the terminal can be shown in fig. 11. The computer device includes a processor, a memory, an input/output interface, a communication interface, a display unit, and an input means. The processor, the memory and the input/output interface are connected through a system bus, and the communication interface, the display unit and the input device are connected to the system bus through the input/output interface. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The input/output interface of the computer device is used to exchange information between the processor and the external device. The communication interface of the computer device is used for carrying out wired or wireless communication with an external terminal, and the wireless mode can be realized through WIFI, a mobile cellular network, NFC (near field communication) or other technologies. The computer program is executed by a processor to implement a vehicle data processing method. The display unit of the computer equipment is used for forming a visual picture, and can be a display screen, a projection device or a virtual reality imaging device, wherein the display screen can be a liquid crystal display screen or an electronic ink display screen, the input device of the computer equipment can be a touch layer covered on the display screen, can also be a key, a track ball or a touch pad arranged on a shell of the computer equipment, and can also be an external keyboard, a touch pad or a mouse and the like.
It will be appreciated by those skilled in the art that the structure shown in FIG. 11 is merely a block diagram of some of the structures associated with the present inventive arrangements and is not limiting of the computer device to which the present inventive arrangements may be applied, and that a particular computer device may include more or fewer components than shown, or may combine some of the components, or have a different arrangement of components.
In an embodiment, there is also provided a computer device comprising a memory and a processor, the memory having stored therein a computer program, the processor implementing the steps of the method embodiments described above when the computer program is executed.
In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored which, when executed by a processor, carries out the steps of the method embodiments described above.
In an embodiment, a computer program product is provided, comprising a computer program which, when executed by a processor, implements the steps of the method embodiments described above.
It should be noted that, the user information (including but not limited to user equipment information, user personal information, etc.) and the data (including but not limited to data for analysis, stored data, presented data, etc.) related to the present application are information and data authorized by the user or sufficiently authorized by each party, and the collection, use and processing of the related data need to comply with the related laws and regulations and standards of the related country and region.
Those skilled in the art will appreciate that implementing all or part of the above described methods may be accomplished by way of a computer program stored on a non-transitory computer readable storage medium, which when executed, may comprise the steps of the embodiments of the methods described above. Any reference to memory, database, or other medium used in embodiments provided herein may include at least one of non-volatile and volatile memory. The nonvolatile Memory may include Read-Only Memory (ROM), magnetic tape, floppy disk, flash Memory, optical Memory, high density embedded nonvolatile Memory, resistive random access Memory (ReRAM), magnetic random access Memory (Magnetoresistive RandomAccess Memory, MRAM), ferroelectric Memory (Ferroelectric RandomAccess Memory, FRAM), phase change Memory (Phase Change Memory, PCM), graphene Memory, and the like. Volatile memory can include random access memory (RandomAccess Memory, RAM) or external cache memory, and the like. By way of illustration, and not limitation, RAM can be in the form of a variety of forms, such as static random access memory (Static RandomAccess Memory, SRAM) or dynamic random access memory (Dynamic RandomAccess Memory, DRAM), and the like. The databases referred to in the embodiments provided herein may include at least one of a relational database and a non-relational database. The non-relational database may include, but is not limited to, a blockchain-based distributed database, and the like. The processor referred to in the embodiments provided in the present application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic unit, a data processing logic unit based on quantum computing, or the like, but is not limited thereto.
The technical features of the above embodiments may be arbitrarily combined, and all possible combinations of the technical features in the above embodiments are not described for brevity of description, however, as long as there is no contradiction between the combinations of the technical features, they should be considered as the scope of the description.
The foregoing examples illustrate only a few embodiments of the application and are described in detail herein without thereby limiting the scope of the application. It should be noted that it will be apparent to those skilled in the art that several variations and modifications can be made without departing from the spirit of the application, which are all within the scope of the application. Accordingly, the scope of the application should be assessed as that of the appended claims.
Claims (15)
1. A vehicle data processing method, characterized in that the method comprises:
acquiring first sample driving data of a first sample vehicle and second sample driving data of a second sample vehicle; the first sample vehicle and the second sample vehicle have a driving sequence on a sample road;
acquiring sample pavement data acquired by a plurality of sample sensors related to the first sample vehicle for the sample road respectively;
Determining a first association relationship between the first sample vehicle and the second sample vehicle according to the first sample driving data and the second sample driving data; determining a second association relationship among the plurality of sample sensors according to the plurality of sample pavement data;
acquiring a first initial weight parameter of the second sample vehicle and a second initial weight parameter of each sample sensor;
based on the first initial weight parameter and each second initial weight parameter, carrying out cross iteration weight parameter adjustment by combining the first sample driving data, the second sample driving data, each sample pavement data, the first association relationship and the second association relationship to obtain a first target weight parameter corresponding to the second sample vehicle and a second target weight parameter corresponding to each sample sensor; the first target weight parameter and the second target weight parameter are used to determine a distance between at least two vehicles for which a driving sequence exists.
2. The method of claim 1, wherein the determining a first association between the first sample vehicle and the second sample vehicle based on the first sample travel data and the second sample travel data comprises:
Performing feature extraction of a plurality of first dimensions based on the first sample driving data to obtain first sample features of the first sample driving data in each first dimension; the plurality of first dimensions includes a vehicle travel dimension and a vehicle attribute dimension;
performing feature extraction of the plurality of first dimensions based on the second sample travel data to obtain second sample features of the second sample travel data in each first dimension;
and determining a first association relationship between the first sample vehicle and the second sample vehicle according to each first sample characteristic and each second sample characteristic.
3. The method of claim 2, wherein determining a first association between the first sample vehicle and the second sample vehicle based on each of the first sample feature and each of the second sample feature comprises:
determining sub-association relations between each first sample feature and each second sample feature;
and determining a first association relationship between the first sample vehicle and the second sample vehicle according to each sub-association relationship.
4. The method of claim 1, wherein determining a second association between the plurality of sample sensors from the plurality of sample road surface data comprises:
Respectively extracting a plurality of characteristics of second dimensions based on each sample pavement data to obtain sample pavement characteristics of each sample pavement data corresponding to each second dimension; the plurality of second dimensions includes a road surface travel dimension and a road surface attribute dimension;
and determining a second association relationship among the plurality of sample sensors according to a plurality of sample pavement characteristics corresponding to each sample pavement data.
5. The method according to any one of claims 1 to 4, wherein the performing, based on the first initial weight parameter and each of the second initial weight parameters, the cross-iterated weight parameter adjustment in combination with the first sample driving data, the second sample driving data, each of the sample road surface data, the first association relationship, and the second association relationship, to obtain a first target weight parameter corresponding to the second sample vehicle and a second target weight parameter corresponding to each of the sample sensors includes:
determining an initial predicted distance between the first sample vehicle and the second sample vehicle based on the first initial weight parameter and each of the second initial weight parameters in combination with the first sample travel data, the second sample travel data, and each of the sample road surface data;
Acquiring an expected distance between the first sample vehicle and the second sample vehicle, and performing iterative processing of weight parameters based on the first initial weight parameter and the initial predicted distance and combining the expected distance, the first sample driving data, the second sample driving data and the first association relation to obtain a first target weight parameter corresponding to the second sample vehicle;
and carrying out iterative processing on the weight parameters based on each second initial weight parameter and the initial predicted distance and combining the expected distance, each sample pavement data and the second association relation to obtain a second target weight parameter corresponding to each sample sensor.
6. The method of claim 5, wherein the second sample vehicle comprises a forward sample vehicle and a backward sample vehicle of the first sample vehicle; the forward sample vehicle is located before the first sample vehicle in its travel position on the sample road, and the backward sample vehicle is located after the first sample vehicle in its travel position on the sample road; the second sample travel data includes forward sample travel data of the forward sample vehicle and backward sample travel data of the backward sample vehicle; the initial prediction distance comprises a forward initial prediction distance and a backward initial prediction distance between the forward sample vehicle and the backward sample vehicle and between the backward sample vehicle and the first sample vehicle respectively; the expected distance comprises a forward expected distance and a backward expected distance between the forward sample vehicle and the backward sample vehicle and the first sample vehicle respectively;
The first association relationship comprises a forward association relationship and a backward association relationship between the forward sample vehicle and the backward sample vehicle and the first sample vehicle respectively; the first initial weight parameters include forward initial weight parameters of the forward sample vehicle and backward initial weight parameters of the backward sample vehicle; the first target weight parameters include forward target weight parameters of the forward sample vehicle and backward target weight parameters of the backward sample vehicle.
7. The method of claim 6, wherein said determining an initial predicted distance between said first sample vehicle and said second sample vehicle based on said first initial weight parameter and each said second initial weight parameter in combination with said first sample travel data, said second sample travel data, each said sample road surface data, comprises:
based on the forward initial weight parameter and each of the second initial weight parameters, combining the first sample driving data, the forward sample driving data and each of the sample road surface data to determine a forward initial prediction distance between the first sample vehicle and the forward sample vehicle;
And determining a backward initial prediction distance between the first sample vehicle and the backward sample vehicle by combining the first sample driving data, the backward sample driving data and the sample pavement data based on the backward initial weight parameters and each second initial weight parameter.
8. The method of claim 6, wherein the performing the iterative processing of the weight parameters based on the first initial weight parameter and the initial predicted distance in combination with the desired distance, the first sample travel data, the second sample travel data, and the first association relationship to obtain the first target weight parameter corresponding to the second sample vehicle comprises:
based on the forward initial weight parameter and the forward initial predicted distance, carrying out iterative processing on the weight parameter by combining the forward expected distance, the first sample driving data, the forward sample driving data and the forward association relation to obtain a forward target weight parameter corresponding to the forward sample vehicle;
and carrying out iterative processing on the weight parameters based on the backward initial weight parameters and the backward initial predicted distance and combining the backward expected distance, the first sample driving data, the backward sample driving data and the backward association relation to obtain backward target weight parameters corresponding to the backward sample vehicle.
9. The method of claim 6, wherein the performing, based on each of the second initial weight parameters and the initial predicted distance, the iterative processing of the weight parameters in combination with the expected distance, each of the sample road surface data, and the second association relationship to obtain a second target weight parameter corresponding to each of the sample sensors includes:
and carrying out iterative processing on weight parameters based on each second initial weight parameter, the forward initial predicted distance and the backward initial predicted distance and combining the forward expected distance, the backward expected distance, each sample pavement data and the second association relation to obtain a second target weight parameter corresponding to each sample sensor.
10. A vehicle data processing method, characterized in that the method comprises:
acquiring first driving data of a first vehicle and second driving data of a second vehicle; the first vehicle and the second vehicle have a driving sequence on a target road;
acquiring target road surface data acquired by a plurality of sensors related to the first vehicle aiming at the target road respectively;
according to the driving sequence, acquiring a first target weight parameter obtained by pre-training corresponding to the second vehicle and a second target weight parameter obtained by pre-training corresponding to each sensor;
Performing regression analysis processing based on the first driving data, the second driving data and the first target weight parameter to obtain first regression data;
carrying out regression analysis processing based on the pavement data and each second target weight parameter to obtain second regression data;
and determining the vehicle distance between the first vehicle and the second vehicle according to the first regression data and the second regression data.
11. A vehicle data processing apparatus, characterized in that the apparatus comprises:
the sample driving data acquisition module is used for acquiring first sample driving data of the first sample vehicle and second sample driving data of the second sample vehicle; the first sample vehicle and the second sample vehicle have a driving sequence on a sample road;
a sample road surface data acquisition module, configured to acquire sample road surface data acquired by a plurality of sample sensors related to the first sample vehicle for the sample road respectively;
the association relation determining module is used for determining a first association relation between the first sample vehicle and the second sample vehicle according to the first sample driving data and the second sample driving data; determining a second association relationship among the plurality of sample sensors according to the plurality of sample pavement data;
A weight parameter acquisition module for acquiring a first initial weight parameter of the second sample vehicle and a second initial weight parameter of each sample sensor;
the weight parameter adjustment module is used for carrying out cross iteration weight parameter adjustment by combining the first sample driving data, the second sample driving data, the sample pavement data, the first association relationship and the second association relationship based on the first initial weight parameter and each second initial weight parameter to obtain a first target weight parameter corresponding to the second sample vehicle and a second target weight parameter corresponding to each sample sensor; the first target weight parameter and the second target weight parameter are used to determine a distance between at least two vehicles for which a driving sequence exists.
12. A vehicle data processing apparatus, characterized in that the apparatus comprises:
the driving data acquisition module is used for acquiring first driving data of the first vehicle and second driving data of the second vehicle; the first vehicle and the second vehicle have a driving sequence on a target road;
the road surface data acquisition module is used for acquiring target road surface data acquired by a plurality of sensors related to the first vehicle aiming at the target road respectively;
The target weight acquisition module is used for acquiring a first target weight parameter obtained by pre-training corresponding to the second vehicle and a second target weight parameter obtained by pre-training corresponding to each sensor according to the driving sequence;
the processing module is used for carrying out regression analysis processing based on the first driving data, the second driving data and the first target weight parameter to obtain first regression data; carrying out regression analysis processing based on the pavement data and each second target weight parameter to obtain second regression data;
and the distance determining module is used for determining the vehicle distance between the first vehicle and the second vehicle according to the first regression data and the second regression data.
13. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor implements the steps of the method of any one of claims 1 to 10 when the computer program is executed.
14. A computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the steps of the method of any of claims 1 to 10.
15. A computer program product comprising a computer program, characterized in that the computer program, when executed by a processor, implements the steps of the method of any one of claims 1 to 10.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211039706.XA CN117033882A (en) | 2022-08-29 | 2022-08-29 | Vehicle data processing method, device, computer equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211039706.XA CN117033882A (en) | 2022-08-29 | 2022-08-29 | Vehicle data processing method, device, computer equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117033882A true CN117033882A (en) | 2023-11-10 |
Family
ID=88632358
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211039706.XA Pending CN117033882A (en) | 2022-08-29 | 2022-08-29 | Vehicle data processing method, device, computer equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117033882A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117250871A (en) * | 2023-11-20 | 2023-12-19 | 暨南大学 | Man-machine cooperation safety assessment method and device based on decentralised federal learning |
-
2022
- 2022-08-29 CN CN202211039706.XA patent/CN117033882A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117250871A (en) * | 2023-11-20 | 2023-12-19 | 暨南大学 | Man-machine cooperation safety assessment method and device based on decentralised federal learning |
| CN117250871B (en) * | 2023-11-20 | 2024-03-08 | 暨南大学 | Man-machine cooperation safety assessment method and device based on decentralised federal learning |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP4152204A1 (en) | Lane line detection method, and related apparatus | |
| CN113264066B (en) | Obstacle track prediction method and device, automatic driving vehicle and road side equipment | |
| EP3186662B1 (en) | Measuring traffic speed in a road network | |
| CN111860227B (en) | Method, apparatus and computer storage medium for training trajectory planning model | |
| EP2575079A2 (en) | Method and apparatus for processing images | |
| CN111612489B (en) | Order quantity prediction method and device and electronic equipment | |
| Shen et al. | A convolutional neural‐network‐based pedestrian counting model for various crowded scenes | |
| US20150363706A1 (en) | Fusion of data from heterogeneous sources | |
| Wei et al. | Survey of connected automated vehicle perception mode: from autonomy to interaction | |
| CN112784954A (en) | Method and device for determining neural network | |
| An et al. | DGInet: Dynamic graph and interaction-aware convolutional network for vehicle trajectory prediction | |
| CN114556375A (en) | Cross-batch normalization | |
| CN113643532A (en) | Regional traffic prediction method and device | |
| CN117033882A (en) | Vehicle data processing method, device, computer equipment and storage medium | |
| JP2022116884A (en) | Model generation device, estimation device, model generation method, and model generation program | |
| CN110189537B (en) | Parking guidance method, device and equipment based on space-time characteristics and storage medium | |
| Chen et al. | Multitarget vehicle tracking and motion state estimation using a novel driving environment perception system of intelligent vehicles | |
| WO2023000261A1 (en) | Regional traffic prediction method and device | |
| Tabibiazar et al. | Kernel-based modeling and optimization for density estimation in transportation systems using floating car data | |
| Tabibiazar et al. | Kernel-based optimization for traffic density estimation in ITS | |
| WO2021150166A1 (en) | Determining a route between an origin and a destination | |
| US11657268B1 (en) | Training neural networks to assign scores | |
| CN116824870A (en) | Road segment flow prediction method, device, equipment and storage medium | |
| Schennings | Deep convolutional neural networks for real-time single frame monocular depth estimation | |
| Aboah et al. | Ai-based framework for understanding car following behaviors of drivers in a naturalistic driving environment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |