CN111027859A - Driving risk prevention method and system based on motor vehicle state monitoring data mining - Google Patents

Driving risk prevention method and system based on motor vehicle state monitoring data mining Download PDF

Info

Publication number
CN111027859A
CN111027859A CN201911262661.0A CN201911262661A CN111027859A CN 111027859 A CN111027859 A CN 111027859A CN 201911262661 A CN201911262661 A CN 201911262661A CN 111027859 A CN111027859 A CN 111027859A
Authority
CN
China
Prior art keywords
vehicle state
motor vehicle
driving
data
driver
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.)
Granted
Application number
CN201911262661.0A
Other languages
Chinese (zh)
Other versions
CN111027859B (en
Inventor
张潮
姜良维
王敏
郑煜
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Traffic Management Research Institute of Ministry of Public Security
Original Assignee
Traffic Management Research Institute of Ministry of Public Security
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Traffic Management Research Institute of Ministry of Public Security filed Critical Traffic Management Research Institute of Ministry of Public Security
Priority to CN201911262661.0A priority Critical patent/CN111027859B/en
Publication of CN111027859A publication Critical patent/CN111027859A/en
Application granted granted Critical
Publication of CN111027859B publication Critical patent/CN111027859B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0635—Risk analysis of enterprise or organisation activities
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • G06F18/232—Non-hierarchical techniques
    • G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
    • G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
    • 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
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/044—Recurrent networks, e.g. Hopfield networks
    • 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
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • 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)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • General Physics & Mathematics (AREA)
  • Business, Economics & Management (AREA)
  • General Engineering & Computer Science (AREA)
  • Human Resources & Organizations (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • General Health & Medical Sciences (AREA)
  • Computing Systems (AREA)
  • Molecular Biology (AREA)
  • Computational Linguistics (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Biophysics (AREA)
  • Biomedical Technology (AREA)
  • Health & Medical Sciences (AREA)
  • Strategic Management (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Economics (AREA)
  • Probability & Statistics with Applications (AREA)
  • Evolutionary Biology (AREA)
  • Development Economics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Educational Administration (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Game Theory and Decision Science (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Marketing (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Tourism & Hospitality (AREA)
  • General Business, Economics & Management (AREA)
  • Traffic Control Systems (AREA)

Abstract

The invention provides a driving risk prevention system based on motor vehicle state monitoring data mining, which comprises: the data acquisition module is used for acquiring vehicle data and related data to form motor vehicle state full-element high-dimensional data which is divided by taking a vehicle as a unit according to time; the data compression module is used for performing dimension reduction compression on the motor vehicle state full-element high-dimensional data; the training module is used for training a large amount of motor vehicle state full-factor data after dimension reduction compression by using a deep neural network to obtain a neural network model capable of correspondingly outputting a driver driving risk value; the prediction analysis module reconstructs the massive motor vehicle state full-factor high-dimensional data into massive driver driving risk values by using a neural network model, and analyzes the massive driver driving risk values by using a machine learning method to obtain a driver driving risk threshold value; the real-time risk prompting module is used for outputting and displaying the current driving risk value of the driver; the invention provides scientific and effective driving instruction for the driver.

Description

Driving risk prevention method and system based on motor vehicle state monitoring data mining
Technical Field
The invention relates to the field of intelligent traffic application, in particular to a driving risk prevention method and system based on motor vehicle state monitoring data mining.
Background
With the rapid rise of the car networking technology, various current intelligent transportation technologies based on massive motor vehicle state monitoring data become research hotspots, and meanwhile, the intelligent transportation technologies are very worthy of research due to huge social values and wide application prospects.
On one hand, the existing research lacks the research based on the motor vehicle state monitoring data mining and driving risk prevention method and system. On the other hand, along with the development of big data and artificial intelligence technology, the method and the system for providing real-time, professional, intelligent, scientific and effective driving guidance for the driver can be called in the field of intelligent transportation.
Disclosure of Invention
The invention aims to overcome the defects in the prior art, and provides a driving risk prevention method and a driving risk prevention system based on motor vehicle state monitoring data mining, which can provide real-time risk value prompt for a driver and provide scientific and effective driving guidance.
The embodiment of the invention provides a driving risk prevention system based on motor vehicle state monitoring data mining, which comprises:
the data acquisition module is used for acquiring vehicle data and related data to form motor vehicle state full-element high-dimensional data which is divided by taking a vehicle as a unit according to time;
the data compression module is used for performing dimension reduction compression on the motor vehicle state full-element high-dimensional data;
the training module is used for training a large amount of motor vehicle state full-factor data after dimension reduction compression by using a deep neural network to obtain a neural network model capable of correspondingly outputting a driver driving risk value;
the prediction analysis module reconstructs the massive motor vehicle state full-factor high-dimensional data into massive driver driving risk values by using a neural network model, and analyzes the massive driver driving risk values by using a machine learning method to obtain a driver driving risk threshold value;
the real-time risk prompting module is used for loading the neural network model, accessing the current motor vehicle state full-element data, and outputting and displaying the current driving risk value of the driver; and prompting when the current driver driving risk value exceeds the driver driving risk value threshold value.
Furthermore, in the data compression module, the dimension reduction compression method adopts a line segment simplification compression algorithm.
Furthermore, in the training module, the deep neural network adopts a multilayer perceptron, and extracts the information whether the traffic accident happens in the motor vehicle state full-factor high-dimensional data as a two-classification label used in training.
Further, in the prediction analysis module, the machine learning method includes: and performing secondary classification on the driving risk value corresponding to the accident by the classification label by applying a K-means algorithm, and taking the minimum value in the large cluster as a driving risk value threshold of the driver.
Furthermore, the system also comprises a driving habit correcting module which is used for guiding the driving operation of the current driver and prompting the adjustment of the current motor vehicle state;
the training module is also required to select data to train a first dual RNN network model based on the driving operation and the motor vehicle state; the driving habit correcting module loads the first double-RNN network model and accesses the current accessed motor vehicle state full-factor data; the first dual-RNN network model comprises two recurrent neural networks RNNs; one predicts the driving operation and the vehicle state at the next moment according to the current driving operation of the driver and the vehicle state, the other predicts the optimal driving operation and the vehicle state adjusting direction at the next moment according to the current driving operation of the driver and the vehicle state, and when the difference between the optimal driving operation and the optimal vehicle state adjusting direction is large enough, the driver is prompted.
Furthermore, the system also comprises a driving condition risk prompting module which is used for giving driving guidance and risk decision according to the current state of the motor vehicle and the driving environment;
the training module is also required to select data to train a second dual-RNN network model based on the motor vehicle state and the driving environment; the driving condition risk prompting module loads the second double-RNN network model and accesses the current accessed motor vehicle state full-factor data; the second dual RNN network model includes two recurrent neural networks RNNs; one motor vehicle state is used for predicting a driving risk value at the next moment according to the current motor vehicle state and the current driving environment, the other motor vehicle state is used for predicting a motor vehicle state adjusting operation which is best beneficial to driving risk improvement at the next moment according to the current motor vehicle state and the current driving environment, and when the risk at the next moment is larger and the driving risk improvement expectation is better, the driving instruction and risk decision prompting are given to a driver.
The embodiment of the invention also provides a driving risk prevention method based on the mining of the motor vehicle state monitoring data, which comprises the following steps:
s1, collecting vehicle data and related data to form motor vehicle state full-element high-dimensional data divided by time by taking a vehicle as a unit;
s2, performing dimension reduction compression on the motor vehicle state full-factor high-dimensional data;
s3, training a large amount of motor vehicle state full-factor data after dimension reduction compression by using a deep neural network to obtain a neural network model capable of correspondingly outputting a driver driving risk value;
s4, reconstructing mass motor vehicle state full-element high-dimensional data into mass driver driving risk values by using a neural network model, and analyzing the values by using a machine learning method to obtain a driver driving risk threshold value;
s5, loading the neural network model, accessing the current motor vehicle state full-element data, outputting and displaying the current driver driving risk value; and prompting when the current driver driving risk value exceeds the driver driving risk value threshold value.
Further, in step S2, a segment simplified compression algorithm is used to make the portion in the trace at a time point represent the whole data.
Further, the data set after the dimension reduction compression is S'; step S3 specifically includes:
(1) extracting all the information of whether the motor vehicles have accidents or not in the S 'as two classification labels, and obtaining S' after the information is remained;
(2) and (4) using the multi-layer perceptron training to update the network parameters W of the multi-layer perceptron, and finally obtaining the neural network model M.
Further, step S4 specifically includes:
(1) inputting the full-element high-dimensional data without the information of whether the motor vehicle has an accident or not into a neural network model M to obtain and reconstruct a mass driver driving risk value set V;
(2) removing all corresponding parts which do not have accidents in the V to obtain V';
(3) performing two classifications on V' by applying a K-means algorithm, and taking the minimum value in the large cluster as a threshold value of a driving risk value of a driver
Figure BDA0002311979700000031
Further, the driving risk prevention method based on the motor vehicle state monitoring data mining further comprises the following steps:
s6, training a first double-RNN network model based on the driving operation and the vehicle state selection data; loading the first double RNN network model, accessing the current accessed motor vehicle state full-factor data, outputting the best possible driving operation at the next moment and the corresponding motor vehicle state adjustment direction, guiding the driving operation of the current driver and prompting the current motor vehicle state adjustment;
s7, training a second dual-RNN network model based on the vehicle state and the driving environment selected data; and loading the second dual-RNN network model, accessing the current accessed motor vehicle state full-element data, and outputting possible driving guidance and risk decision reminding at the next moment.
The invention has the advantages that: the invention supports providing real-time risk value prompt for a driver in any vehicle, selectively enables the driving habit correction module and the driving condition risk prompt module, provides real-time, professional, intelligent, scientific and effective driving guidance for the driver, helps the driver select driving operation scientifically to keep the driving safety of the motor vehicle, reduces the damage of wrong operation to the motor vehicle, gives prompt to the driver timely and accurately, treats various strange driving environments efficiently, and avoids driving danger caused by insufficient driving experience.
Drawings
FIG. 1 is a schematic diagram of a system according to an embodiment of the present invention.
FIG. 2 is a flowchart of an embodiment of a method of the present invention.
Detailed Description
The invention is further illustrated by the following specific figures and examples.
The embodiment of the invention provides a driving risk prevention system based on motor vehicle state monitoring data mining, as shown in fig. 1, comprising: the system comprises a data acquisition module, a data compression module, a training module, a prediction analysis module, a real-time risk prompting module, a driving habit correction module and a driving condition risk prompting module; wherein the latter two modules belong to the matched module;
in some embodiments, the data collection module, the data compression module, the training module, and the predictive analysis module may be disposed on a server or a high-performance computing device, such as a distributed computing cluster; of course, these modules may also be configured on the vehicle-mounted AI chip in case the performance of the vehicle-mounted AI chip is sufficient;
in some embodiments, the real-time risk prompting module, and optionally the driving habit correction module, the driving condition risk prompting module are configured on the vehicle-mounted AI unit; the vehicle-mounted AI unit can comprise a vehicle-mounted AI chip, a display screen and the like;
1) the data acquisition module is used for acquiring vehicle data and related data to form motor vehicle state full-element high-dimensional data which is divided by taking a vehicle as a unit according to time;
in some embodiments, mass data of a motor vehicle security networking supervision platform based on the internet of vehicles can be collected to form motor vehicle state full-element high-dimensional data which is divided by taking the vehicles as units according to time;
the vehicle unit is an object paired by a specific vehicle and a specific driver, the collected massive full-factor high-dimensional data is high-dimensional data with one dimension expanded along with the data richness expansion of a motor vehicle safety networking supervision platform based on the Internet of vehicles, and the vehicle unit comprises the following information: vehicle unique ID, driver identity ID, vehicle location status, vehicle speed, vehicle heading, driver operation, vehicle emissions, vehicle three-axis acceleration, vehicle three-axis angular velocity, weather conditions, road conditions, obd information, traffic accident information, time information.
2) The data compression module is used for performing dimension reduction compression on the motor vehicle state full-element high-dimensional data;
in some embodiments, the dimension reduction compression method may use the vehicle driving track as a main compression basis, and use a line segment simplified compression algorithm to make part of data in a section of track represent the whole data.
3) The training module is used for training a large amount of motor vehicle state full-factor data after dimension reduction compression by using a deep neural network to obtain a neural network model capable of correspondingly outputting a driver driving risk value;
in some embodiments, the deep neural network used is a multilayer perceptron, and information of whether a traffic accident occurs in the full-element high-dimensional data of the motor vehicle state is extracted to be used as a two-classification label used in training.
4) The prediction analysis module reconstructs the massive motor vehicle state full-factor high-dimensional data into massive driver driving risk values by using a neural network model, and analyzes the massive driver driving risk values by using a machine learning method to obtain a driver driving risk threshold value;
in some embodiments, the machine learning method may be that a K-means algorithm is applied to the driving risk value corresponding to the classification label as the accident to perform secondary classification, and the minimum value in the large cluster is taken as the driving risk value threshold of the driver.
5) The real-time risk prompting module is used for loading the neural network model, accessing the current motor vehicle state full-element data, and outputting and displaying the current driving risk value of the driver; prompting when the current driving risk value of the driver exceeds the driving risk value threshold value of the driver;
in some embodiments, when the current driver driving risk value exceeds the driver driving risk value threshold, a prompt may be made in red on the in-vehicle display screen; in other embodiments, a warning sound may be used for the prompt.
6) In some embodiments, the system may further include a driving habit correction module that guides the current driving operation of the driver and prompts the current vehicle state adjustment;
the training module is also required to select data to train a first dual RNN network model based on the driving operation and the motor vehicle state;
the driving habit correcting module loads the first double-RNN network model and accesses the current accessed motor vehicle state full-factor data;
the first dual-RNN network model comprises two recurrent neural networks RNNs; one predicts the driving operation and the vehicle state at the next moment according to the current driving operation of the driver and the vehicle state, and the other predicts the optimal driving operation and the vehicle state adjusting direction at the next moment according to the current driving operation of the driver and the vehicle state, and when the difference between the two is large enough (for example, by comparing whether the difference exceeds a set range), the optimal driving operation and the optimal vehicle state adjusting direction are prompted to the driver;
7) in some embodiments, the system may further include a driving condition risk prompting module that gives driving guidance and risk decision according to the current vehicle state and driving environment;
the training module is also required to select data to train a second dual-RNN network model based on the motor vehicle state and the driving environment;
the driving condition risk prompting module loads the second double-RNN network model and accesses the current accessed motor vehicle state full-factor data;
the second dual RNN network model includes two recurrent neural networks RNNs; one predicts the driving risk value at the next moment according to the current motor vehicle state and the current driving environment (such as weather conditions and road surface conditions), and the other predicts the motor vehicle state adjusting operation which is best beneficial to the driving risk improvement at the next moment according to the current motor vehicle state and the current driving environment (such as weather conditions and road surface conditions).
The embodiment of the invention also provides a driving risk prevention method based on the motor vehicle state monitoring data mining, which comprises the following steps:
s1, collecting vehicle data and related data to form motor vehicle state full-element high-dimensional data divided by time by taking a vehicle as a unit;
(1) let the full-element high-dimensional data unit vector s be defined as: s ═ I (I)v,Ip,T,E1,E2,,...Ei...,En) In which IvIndicating a vehicle unique ID, IpID indicating driver identity, T indicating time, EiRepresents data collected by other numerous vehicle networking-based motor vehicle safety networking supervision platforms, such as vehicle positioning state, vehicle speed, vehicle heading, driver operation, vehicle emission, vehicle three-axis acceleration, vehicle three-axis angular velocity, weather conditions, road conditions, obd information, traffic accident information and the like, and is high-dimensional data with one dimension capable of being expanded along with the expansion of the data richness of the vehicle networking-based motor vehicle safety networking supervision platforms;
(2) let the full-factor high-dimensional data S set be defined as: s ═ S1,t11,...,s1,t1n,s2,t21,...,s2,t2n,...si,ti1,..si,tii.,si,tin,...,sn,tn1,...,sn,tnnIn which s isi,tiiA full element high-dimensional data unit vector representing a vehicle unit at a certain moment of the ith motor vehicle;
s2, performing dimension reduction compression on the motor vehicle state full-factor high-dimensional data;
(1) the full-element high-dimensional data S set can be simplified by adopting line segmentsThe compression algorithm makes the part in the track of a period of time represent the whole data, and obtains S' ═ { S ═1,t1j,...,s1,t1k,s2,t2j,...,s2,t2k,...si,tij,..si,tii.,si,tik,...,sn,tnj,...,sn,tnk}；
S3, training a large amount of motor vehicle state full-factor data after dimension reduction compression by using a deep neural network to obtain a neural network model capable of correspondingly outputting a driver driving risk value;
(1) extracting all the information of whether the motor vehicles have accidents or not in the S 'as two classification labels, and obtaining S' after the information is remained;
(2) using a multilayer perceptron for training, updating a network parameter W of the multilayer perceptron, and finally obtaining a neural network model M;
s4, reconstructing mass motor vehicle state full-element high-dimensional data into mass driver driving risk values by using a neural network model, and analyzing the values by using a machine learning method to obtain a driver driving risk threshold value;
(1) inputting the full-element high-dimensional data without the information of whether the motor vehicle has an accident or not into a neural network model M, and obtaining and reconstructing a mass driver driving risk value set V ═ V1,t1j,...,v1,t1k,v2,t2j,...,v2,t2k,...vi,tij,..vi,tii.,vi,tik,...,vn,tnj,...,vn,tnk-wherein the risk value vector v is defined as v ═ α, where α is a real number, representing driving risk size;
(2) removing all corresponding parts which do not have accidents in the V to obtain V';
(3) performing two classifications on V' by applying a K-means algorithm, and taking the minimum value in the large cluster as a threshold value of a driving risk value of a driver
Figure BDA0002311979700000061
S5, loading the neural network model, accessing the current motor vehicle state full-element data, outputting and displaying the current driver driving risk value; prompting when the current driving risk value of the driver exceeds the driving risk value threshold value of the driver;
(1) selecting a low-power-consumption and high-computing-power vehicle-mounted AI unit based on a vehicle-mounted AI chip to load the neural network model M, accessing the current motor vehicle state full-element data, and outputting a current driving risk value α;
(2) when α passes over
Figure BDA0002311979700000062
The driving risk value can be prompted by red on a display screen;
optionally, the following two steps may also be included:
s6, training a first double-RNN network model based on the driving operation and the vehicle state selection data; loading the first double RNN network model, accessing the current accessed motor vehicle state full-factor data, outputting the best possible driving operation at the next moment and the corresponding motor vehicle state adjustment direction, guiding the driving operation of the current driver and prompting the current motor vehicle state adjustment;
if the driving is excellent, no guide information is prompted, and the information for confirming the driving technology can be output;
s7, training a second dual-RNN network model based on the vehicle state and the driving environment selected data; loading the second dual-RNN network model, accessing the current accessed motor vehicle state full-element data, and outputting possible driving guidance and risk decision reminding at the next moment;
if the risk level is low, no information is prompted, and driving information for reassurance can be output.
Finally, it should be noted that the above embodiments are only for illustrating the technical solutions of the present invention and not for limiting, and although the present invention has been described in detail with reference to examples, it should be understood by those skilled in the art that modifications or equivalent substitutions may be made on the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, which should be covered by the claims of the present invention.

Claims (11)

1. A driving risk prevention system based on motor vehicle state monitoring data mining is characterized by comprising:
the data acquisition module is used for acquiring vehicle data and related data to form motor vehicle state full-element high-dimensional data which is divided by taking a vehicle as a unit according to time;
the data compression module is used for performing dimension reduction compression on the motor vehicle state full-element high-dimensional data;
the training module is used for training a large amount of motor vehicle state full-factor data after dimension reduction compression by using a deep neural network to obtain a neural network model capable of correspondingly outputting a driver driving risk value;
the prediction analysis module reconstructs the massive motor vehicle state full-factor high-dimensional data into massive driver driving risk values by using a neural network model, and analyzes the massive driver driving risk values by using a machine learning method to obtain a driver driving risk threshold value;
the real-time risk prompting module is used for loading the neural network model, accessing the current motor vehicle state full-element data, and outputting and displaying the current driving risk value of the driver; and prompting when the current driver driving risk value exceeds the driver driving risk value threshold value.
2. The driving risk prevention system based on motor vehicle state monitoring data mining of claim 1,
in the data compression module, a dimension reduction compression method adopts a line segment simplified compression algorithm.
3. The driving risk prevention system based on the motor vehicle state monitoring data mining according to claim 1 or 2,
in the training module, the deep neural network adopts a multilayer perceptron, and extracts the information whether the traffic accident happens or not in the motor vehicle state full-factor high-dimensional data as a two-classification label used in training.
4. The driving risk prevention system based on motor vehicle state monitoring data mining of claim 3,
in the prediction analysis module, the machine learning method comprises the following steps: and performing secondary classification on the driving risk value corresponding to the accident by the classification label by applying a K-means algorithm, and taking the minimum value in the large cluster as a driving risk value threshold of the driver.
5. The driving risk prevention system based on the motor vehicle state monitoring data mining according to claim 1 or 2,
the system also comprises a driving habit correcting module which is used for guiding the driving operation of the current driver and prompting the adjustment of the current motor vehicle state;
the training module is also required to select data to train a first dual RNN network model based on the driving operation and the motor vehicle state; the driving habit correcting module loads the first double-RNN network model and accesses the current accessed motor vehicle state full-factor data; the first dual-RNN network model comprises two recurrent neural networks RNNs; one predicts the driving operation and the vehicle state at the next moment according to the current driving operation of the driver and the vehicle state, the other predicts the optimal driving operation and the vehicle state adjusting direction at the next moment according to the current driving operation of the driver and the vehicle state, and when the difference between the optimal driving operation and the optimal vehicle state adjusting direction is large enough, the driver is prompted.
6. The driving risk prevention system based on the motor vehicle state monitoring data mining according to claim 1 or 2,
the system also comprises a driving condition risk prompting module which is used for giving driving guidance and risk decision according to the current motor vehicle state and driving environment;
the training module is also required to select data to train a second dual-RNN network model based on the motor vehicle state and the driving environment; the driving condition risk prompting module loads the second double-RNN network model and accesses the current accessed motor vehicle state full-factor data; the second dual RNN network model includes two recurrent neural networks RNNs; one motor vehicle state is used for predicting a driving risk value at the next moment according to the current motor vehicle state and the current driving environment, the other motor vehicle state is used for predicting a motor vehicle state adjusting operation which is best beneficial to driving risk improvement at the next moment according to the current motor vehicle state and the current driving environment, and when the risk at the next moment is larger and the driving risk improvement expectation is better, the driving instruction and risk decision prompting are given to a driver.
7. A driving risk prevention method based on motor vehicle state monitoring data mining is characterized by comprising the following steps:
s1, collecting vehicle data and related data to form motor vehicle state full-element high-dimensional data divided by time by taking a vehicle as a unit;
s2, performing dimension reduction compression on the motor vehicle state full-factor high-dimensional data;
s3, training a large amount of motor vehicle state full-factor data after dimension reduction compression by using a deep neural network to obtain a neural network model capable of correspondingly outputting a driver driving risk value;
s4, reconstructing mass motor vehicle state full-element high-dimensional data into mass driver driving risk values by using a neural network model, and analyzing the values by using a machine learning method to obtain a driver driving risk threshold value;
s5, loading the neural network model, accessing the current motor vehicle state full-element data, outputting and displaying the current driver driving risk value; and prompting when the current driver driving risk value exceeds the driver driving risk value threshold value.
8. The driving risk prevention method based on the mining of vehicle state monitoring data according to claim 7,
in step S2, a segment simplified compression algorithm is used to make the portion of the trace at a time represent the whole data.
9. The driving risk prevention method based on the mining of vehicle state monitoring data according to claim 7,
the data set after the dimensionality reduction compression is S'; step S3 specifically includes:
(1) extracting all the information of whether the motor vehicles have accidents or not in the S 'as two classification labels, and obtaining S' after the information is remained;
(2) and (4) using the multi-layer perceptron training to update the network parameters W of the multi-layer perceptron, and finally obtaining the neural network model M.
10. The driving risk prevention method based on the mining of vehicle state monitoring data according to claim 8,
step S4 specifically includes:
(1) inputting the full-element high-dimensional data without the information of whether the motor vehicle has an accident or not into a neural network model M to obtain and reconstruct a mass driver driving risk value set V;
(2) removing all corresponding parts which do not have accidents in the V to obtain V';
(3) performing two classifications on V' by applying a K-means algorithm, and taking the minimum value in the large cluster as a threshold value of a driving risk value of a driver
Figure FDA0002311979690000031
11. The driving risk prevention method based on the mining of the motor vehicle state monitoring data according to claim 7, further comprising the steps of:
s6, training a first double-RNN network model based on the driving operation and the vehicle state selection data; loading the first double RNN network model, accessing the current accessed motor vehicle state full-factor data, outputting the best possible driving operation at the next moment and the corresponding motor vehicle state adjustment direction, guiding the driving operation of the current driver and prompting the current motor vehicle state adjustment;
s7, training a second dual-RNN network model based on the vehicle state and the driving environment selected data; and loading the second dual-RNN network model, accessing the current accessed motor vehicle state full-element data, and outputting possible driving guidance and risk decision reminding at the next moment.
CN201911262661.0A 2019-12-11 2019-12-11 Driving risk prevention method and system based on motor vehicle state monitoring data mining Active CN111027859B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201911262661.0A CN111027859B (en) 2019-12-11 2019-12-11 Driving risk prevention method and system based on motor vehicle state monitoring data mining

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201911262661.0A CN111027859B (en) 2019-12-11 2019-12-11 Driving risk prevention method and system based on motor vehicle state monitoring data mining

Publications (2)

Publication Number Publication Date
CN111027859A true CN111027859A (en) 2020-04-17
CN111027859B CN111027859B (en) 2021-11-30

Family

ID=70205435

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201911262661.0A Active CN111027859B (en) 2019-12-11 2019-12-11 Driving risk prevention method and system based on motor vehicle state monitoring data mining

Country Status (1)

Country Link
CN (1) CN111027859B (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112418400A (en) * 2020-11-20 2021-02-26 江苏驭道数据科技有限公司 Vehicle driving safety risk assessment method and system
CN112849156A (en) * 2021-04-25 2021-05-28 北京三快在线科技有限公司 Driving risk identification method and device

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107730028A (en) * 2017-09-18 2018-02-23 广东翼卡车联网服务有限公司 A kind of car accident recognition methods, car-mounted terminal and storage medium
CN108423006A (en) * 2018-02-02 2018-08-21 辽宁友邦网络科技有限公司 A kind of auxiliary driving warning method and system
KR101984284B1 (en) * 2017-11-28 2019-05-30 주식회사 제네시스랩 Automated Driver-Managing System Using Machine Learning Model, and Method Thereof
KR20190072077A (en) * 2017-12-15 2019-06-25 현대자동차주식회사 System and method for predicting vehicle accidents

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107730028A (en) * 2017-09-18 2018-02-23 广东翼卡车联网服务有限公司 A kind of car accident recognition methods, car-mounted terminal and storage medium
KR101984284B1 (en) * 2017-11-28 2019-05-30 주식회사 제네시스랩 Automated Driver-Managing System Using Machine Learning Model, and Method Thereof
KR20190072077A (en) * 2017-12-15 2019-06-25 현대자동차주식회사 System and method for predicting vehicle accidents
CN108423006A (en) * 2018-02-02 2018-08-21 辽宁友邦网络科技有限公司 A kind of auxiliary driving warning method and system

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
李洪涛等: "车车通信下基于ＲBF 神经网络的", 《重庆理工大学学报（自然科学）》 *
胡立伟等: "基于改进BP神经网络的城市交通拥塞环境下车辆运行风险识别研究", 《公路交通科技》 *

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112418400A (en) * 2020-11-20 2021-02-26 江苏驭道数据科技有限公司 Vehicle driving safety risk assessment method and system
CN112849156A (en) * 2021-04-25 2021-05-28 北京三快在线科技有限公司 Driving risk identification method and device
CN112849156B (en) * 2021-04-25 2021-07-30 北京三快在线科技有限公司 Driving risk identification method and device

Also Published As

Publication number Publication date
CN111027859B (en) 2021-11-30

Similar Documents

Publication Publication Date Title
US11836802B2 (en) Vehicle operation analytics, feedback, and enhancement
US11842404B2 (en) Enhancement using analytics based on vehicle kinematic data
DE112018004000B4 (en) MONITORING VEHICLE OPERATION RISK USING SENSOR UNITS
CN109711557B (en) Driving track prediction method, computer equipment and storage medium
CN107782564A (en) A kind of automatic driving vehicle evaluation system and method
CN106971540B (en) The behaviorist risk quantization method of driver under freeway tunnel environment
CN111259737B (en) Method and device for predicting failure of steering wheel of vehicle, electronic equipment and storage medium
CN110458214B (en) Driver replacement recognition method and device
CN111027859B (en) Driving risk prevention method and system based on motor vehicle state monitoring data mining
CN112287801A (en) Vehicle-mounted data processing method and device, server and readable storage medium
CN112435466B (en) Method and system for predicting take-over time of CACC vehicle changing into traditional vehicle under mixed traffic flow environment
Jain et al. Review of computational techniques for modelling eco-safe driving behavior
Deng et al. Research on operation characteristics and safety risk forecast of bus driven by multisource forewarning data
Lee et al. Individual stable driving pattern analysis for evaluating driver readiness at autonomous driving levels 2 and 3
DE102018215017A1 (en) Method for determining an operating strategy for a vehicle, system, control device for a vehicle and a vehicle
CN113947326A (en) Comprehensive vehicle risk classification and risk premium measurement and calculation method
CN114973212A (en) Fatigue driving stopping method based on visual features, steering wheel operation detection and active intervention
Bhalerao et al. Proposed design on driver behavioral analysis
Gupta et al. Unveiling gap acceptance behaviour during lane change with EDIV data: A deep dive into driving behaviour on expressway using a three level mixed effect linear regression approach
Michelaraki et al. Modelling the Safety Tolerance Zone: Recommendations from the i-DREAMS project
Haghshenas et al. Review of Applications of ML Approaches in Driver Behavior Analysis Using Qualitative and Quantitative Analysis
CN117076816B (en) Response prediction method, response prediction apparatus, computer device, storage medium, and program product
Maktoubian et al. Empirical Evaluation of Machine Learning Models for Fuel Consumption, Driver Identification, and Behavior Prediction
Talebloo et al. Dynamic and Systematic Survey of Deep Learning Approaches for Driving Behavior Analysis
Mosleh et al. Embedded machine learning-based road conditions and driving behavior monitoring.

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
GR01 Patent grant
GR01 Patent grant