CN113256986A - Traffic analysis method, related device and readable storage medium - Google Patents
Traffic analysis method, related device and readable storage medium Download PDFInfo
- Publication number
- CN113256986A CN113256986A CN202110727130.5A CN202110727130A CN113256986A CN 113256986 A CN113256986 A CN 113256986A CN 202110727130 A CN202110727130 A CN 202110727130A CN 113256986 A CN113256986 A CN 113256986A
- Authority
- CN
- China
- Prior art keywords
- tensor
- traffic
- factor matrix
- traffic data
- road network
- 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
Links
- 238000004458 analytical method Methods 0.000 title claims abstract description 43
- 238000003860 storage Methods 0.000 title claims abstract description 10
- 238000000034 method Methods 0.000 claims abstract description 39
- 230000007246 mechanism Effects 0.000 claims abstract description 13
- 239000011159 matrix material Substances 0.000 claims description 109
- 230000006870 function Effects 0.000 claims description 29
- 230000004927 fusion Effects 0.000 claims description 19
- 238000000354 decomposition reaction Methods 0.000 claims description 6
- 239000013598 vector Substances 0.000 claims description 6
- 238000012545 processing Methods 0.000 claims description 4
- 230000002123 temporal effect Effects 0.000 claims 1
- 238000001514 detection method Methods 0.000 description 7
- 238000010586 diagram Methods 0.000 description 7
- 238000004364 calculation method Methods 0.000 description 5
- 238000000605 extraction Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 4
- 238000005457 optimization Methods 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 238000004422 calculation algorithm Methods 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000004590 computer program Methods 0.000 description 2
- 238000013527 convolutional neural network Methods 0.000 description 2
- 239000013256 coordination polymer Substances 0.000 description 2
- 238000007499 fusion processing Methods 0.000 description 2
- 238000011478 gradient descent method Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 230000002776 aggregation Effects 0.000 description 1
- 238000004220 aggregation Methods 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 230000001502 supplementing effect Effects 0.000 description 1
- 238000012549 training Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
- 230000003442 weekly effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
- G08G1/0125—Traffic data processing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
- G08G1/0125—Traffic data processing
- G08G1/0129—Traffic data processing for creating historical data or processing based on historical data
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
- G08G1/0137—Measuring and analyzing of parameters relative to traffic conditions for specific applications
Landscapes
- Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Analytical Chemistry (AREA)
- Chemical & Material Sciences (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- General Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Traffic Control Systems (AREA)
Abstract
The application provides a traffic analysis method, related equipment and a readable storage medium, and belongs to the field of traffic. The method comprises the following steps: generating a first tensor according to traffic data detected by each detector in a target road network within a first time length, wherein the length of a space dimension of the first tensor is S road sections in the target road network, the length of a time dimension is T time points in the first time length, the length of a traffic data dimension is R-type traffic data detected by the detectors, S and T are positive integers, and R is an integer greater than 1; fusing the space dimensionality and the traffic data dimensionality in the first tensor by using an attention mechanism to obtain a second tensor; determining target information of a target road network according to the third tensor; wherein the third tensor is the second tensor or is determined based on the second tensor; the target information includes at least one of: traffic patterns, traffic laws, and future traffic conditions. The method and the device utilize the multi-dimensional traffic data to perform traffic analysis, and accuracy of the traffic analysis is improved.
Description
Technical Field
The embodiment of the application relates to the field of traffic, in particular to a traffic analysis method, related equipment and a readable storage medium.
Background
With the rapid development of the traffic field, the traffic jam is increasingly intensified. In order to ensure that traffic management measures are reasonable and effective, it is necessary to acquire traffic data on a road network so as to analyze traffic congestion by using the acquired traffic data. In the prior art, only single traffic data is used for traffic analysis, so that the accuracy of the traffic analysis is low.
Disclosure of Invention
The embodiment of the application provides a traffic analysis method, related equipment and a readable storage medium, so as to solve the problem that the accuracy of traffic analysis is low in the prior art.
To solve the above problem, the present application is implemented as follows:
in a first aspect, an embodiment of the present application provides a traffic analysis method, where the method includes:
generating a first tensor according to traffic data detected by each detector in a target road network within a first time length, wherein the length of a space dimension of the first tensor is S road sections in the target road network, the length of a time dimension is T time points in the first time length, the length of a traffic data dimension is R-type traffic data detected by the detectors, S and T are positive integers, and R is an integer greater than 1;
fusing the space dimensionality and the traffic data dimensionality in the first tensor by using an attention mechanism to obtain a second tensor;
determining target information of the target road network according to the third tensor;
wherein the third tensor is the second tensor or determined based on the second tensor; the target information includes at least one of: traffic patterns, traffic laws, and future traffic conditions.
In a second aspect, an embodiment of the present application further provides a traffic analysis apparatus, including:
the system comprises a generating module, a calculating module and a processing module, wherein the generating module is used for generating a first tensor according to traffic data detected by each detector in a target road network within a first time length, the length of a space dimension of the first tensor is S road sections in the target road network, the length of a time dimension is T time points in the first time length, the length of a traffic data dimension is R-type traffic data detected by the detector, S and T are positive integers, and R is an integer greater than 1;
the fusion module is used for fusing the space dimensionality and the traffic data dimensionality in the first tensor by using an attention mechanism to obtain a second tensor;
a first determining module, configured to determine, according to a third tensor, target information of the target road network;
wherein the third tensor is the second tensor or determined based on the second tensor; the target information includes at least one of: traffic patterns, traffic laws, and future traffic conditions.
In a third aspect, an embodiment of the present application further provides an electronic device, including: a transceiver, a memory, a processor, and a program stored on the memory and executable on the processor; the processor is configured to read a program in the memory to implement the method according to the first aspect.
In a fourth aspect, embodiments of the present application further provide a readable storage medium for storing a program, where the program is executed by a processor to implement the method according to the first aspect.
In the embodiment of the application, the electronic equipment performs space-time splicing on the multi-dimensional traffic data detected by the detector to obtain a first sheet quantity; then, carrying out fusion processing on the spatial dimension and the traffic data dimension in the first tensor to obtain a second tensor, and further determining target information of the target road network according to the second tensor, wherein the target information comprises at least one of the following items: traffic patterns, traffic laws, and future traffic conditions. Therefore, the electronic device of the embodiment of the application comprehensively utilizes the multi-dimensional traffic data detected by the detector to analyze the traffic of the target road network, so that the accuracy of traffic analysis can be improved.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed to be used in the description of the embodiments of the present application will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without inventive exercise.
Fig. 1 is a schematic flow chart of a traffic analysis method provided in an embodiment of the present application;
FIG. 2 is a schematic diagram of a first tensor provided by an embodiment of the present application;
FIG. 3 is a schematic diagram of feature extraction of tensors provided by an embodiment of the present application;
FIG. 4 is a schematic diagram illustrating fusion of tensors provided by an embodiment of the present application;
FIG. 5 is a schematic diagram of a first model provided by an embodiment of the present application;
FIG. 6 is a schematic structural diagram of a traffic analysis device provided in the practice of the present application;
fig. 7 is a schematic structural diagram of an electronic device provided in this application.
Detailed Description
The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are some, but not all, embodiments of the present application. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
The terms "first," "second," and the like in the embodiments of the present application are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus. Further, as used herein, "and/or" means at least one of the connected objects, e.g., a and/or B and/or C, means 7 cases including a alone, B alone, C alone, and both a and B present, B and C present, both a and C present, and A, B and C present.
The following describes a traffic analysis method provided in an embodiment of the present application.
The traffic analysis method of the embodiment of the application can be executed by the electronic device. In practical applications, the electronic Device may be a server, a Mobile phone, a Tablet Personal Computer (Tablet Personal Computer), a Laptop Computer (Laptop Computer), a Personal Digital Assistant (PDA), a Mobile Internet Device (MID), a Wearable Device (Wearable Device), or a vehicle-mounted Device.
Referring to fig. 1, fig. 1 is a schematic flow chart of a traffic analysis method provided in an embodiment of the present application. As shown in fig. 1, the following steps may be included:
In practical application, the detector is arranged on each lane of each road section of the target road network. Alternatively, a detector can be arranged on a lane of a road section, i.e. a detector can be used to detect traffic data of a lane of a road section.
Each detector may detect a value of R-type traffic data, which may include, but is not limited to, data reflecting traffic conditions, such as an average speed, an occupancy rate, the number of vehicles traveling, and a lane occupancy time of each lane, according to a preset frequency. In the embodiment of the present application, the R-type traffic data may also be referred to as R-type traffic state.
In this step, the electronic device may obtain the overall traffic state of the target network within the first duration by obtaining traffic data detected by each detector in the target network within the first duration, and further may sense the traffic state of the target network within the first duration, and may predict a future traffic state of the target network by using the sensed traffic state, where the first preset duration may be set in advance according to a requirement, such as 1 hour or 1 day.
After acquiring traffic data detected by each detector in a target network within a first time period, an electronic device may generate a first tensor, where the first tensor is a multidimensional traffic data space-time tensor, and the first tensor includes all traffic data detected by the detectors of each road segment in the target network within the first time period.
It is to be understood that, in the case that the target network includes S road segments, the detector performs T detections within the first duration, and the detector detects R-type traffic data, the first tensor is an sxtx R tensor, where S represents a length of a spatial dimension; t represents the length of the time dimension; r represents the length size of the traffic data dimension.
The millimeter wave radar is considered to have the following advantages: the influence of the environment is small, such as the influence of haze, rain, snow and light is small; the detection coverage is large, the visual angle can reach 120 degrees, and the detection distance can reach 250 meters; the engineering is simple; no extra computational power requirement is required; the maintenance cost is low. The detector in the embodiment of the application can be a millimeter wave radar detector, so that the traffic data detected by the detector is more accurate, and the accuracy of traffic analysis can be further improved. Of course, it is understood that the detector in the embodiment of the present application may also be other types of detectors, such as a coil detector or a geomagnetic detector, and the like, which may be determined according to actual requirements and is not limited in the embodiment of the present application.
And 102, fusing the space dimensionality and the traffic data dimensionality in the first tensor by using an attention mechanism to obtain a second tensor.
In this embodiment of the application, after the electronic device generates the first vector, the electronic device may further capture and identify the regional features of the spatial dimension and the traffic data dimension in the first vector by using an attention mechanism, so as to improve accuracy of regional feature identification and capture. Since the more accurate the feature recognition and capture are, the better the analysis effect of the traffic state is, the accuracy of the traffic state analysis can be further improved by analyzing the traffic state using the second tensor obtained by fusion.
And 103, determining target information of the target road network according to the third tensor.
Wherein the third tensor is the second tensor or determined based on the second tensor; the target information includes at least one of: traffic patterns, traffic laws, and future traffic conditions.
In a specific implementation, in a first implementation, the electronic device may determine the target information of the target road network directly based on the second tensor obtained by fusion. In a second implementation manner, the electronic device may further detect whether there is a missing value in the second tensor, make up for the second tensor when there is a missing value in the second tensor, and determine the target information of the target road network by using a third tensor obtained by making up for the missing value. It can be seen that the accuracy of the target information of the target road network determined by the second implementation is higher than that of the first implementation.
In the embodiment of the application, the traffic modes can be characterized by congestion degrees, and the congestion degrees of different traffic modes are different. Traffic laws can be characterized by the average of various traffic data within a target time or target space. The future traffic state can be characterized by a multidimensional traffic data space-time tensor.
According to the traffic analysis method, the electronic equipment carries out space-time splicing on the multi-dimensional traffic data detected by the detector to obtain a first quantity; then, carrying out fusion processing on the spatial dimension and the traffic data dimension in the first tensor to obtain a second tensor, and further determining target information of the target road network according to the second tensor, wherein the target information comprises at least one of the following items: traffic patterns, traffic laws, and future traffic conditions. Therefore, the electronic device of the embodiment of the application comprehensively utilizes the multi-dimensional traffic data detected by the detector to analyze the traffic of the target road network, so that the accuracy of traffic analysis can be improved.
In this embodiment of the present application, optionally, the generating a first vector according to traffic data detected by each detector in the target road network within a first time period includes:
generating a time sequence corresponding to T time points detected by each detector in a target road network and a space sequence corresponding to S road sections in the target road network;
and splicing the value of the R-type traffic data corresponding to each time point in the time sequence and the value of the R-type traffic data corresponding to each road section in the space sequence to obtain a first quantity.
In this optional embodiment, after acquiring the traffic data detected by each detector in the target network within the first duration, the electronic device may perform space-time stitching on the acquired traffic data to generate the first tensor.
In particular implementations, the values of traffic data have a high correlation with the time dimension. After each unit time, the value of each traffic data of each lane of the road network changes due to the driving-in, driving-out, parking waiting, accident collision and the like of the vehicle. Thus, the electronic device may generate a time series including the T time points in chronological order of detection by the detector.
Traffic data such as lane speed and occupancy change between each road section of the same road network and each lane of the same road section due to alternate driving of vehicles or interaction between front and rear vehicles and left and right vehicles. Thus, the electronic device may generate a spatial sequence comprising the S road segments in a spatial order of the S road segments.
And after the electronic equipment generates the time sequence and the space sequence, forming a grid graph with time information and space relative relation based on the dimensions of time and space. The R-type traffic data can be represented by using R grids, and the R grids form a multidimensional space-time tensor of traffic states, i.e. a first tensor, according to the space-time correspondence, wherein each grid represents one type of traffic data, and each grid in the grids represents a value of one type of traffic data of one lane at one time point. For convenience of understanding, please refer to fig. 2, in fig. 2, the traffic data dimension of the first tensor is 3, but the value of the traffic data dimension is not limited thereby.
By the mode, the space-time splicing of the traffic data of the target road network in the first duration can be realized, and the first tensor is obtained, so that the traffic state obtained by the first tensor analysis can better accord with the actual traffic state, and the accuracy of traffic analysis can be improved.
In this embodiment of the application, optionally, the fusing, by using an attention mechanism, the spatial dimension and the traffic data dimension in the first tensor to obtain a second tensor includes:
extracting the regional characteristics of the space dimensionality and the traffic data dimensionality in the first tensor to obtain an intermediate tensor;
inputting the intermediate tensor into an attention fusion network to obtain the weight value of each regional characteristic in the intermediate tensor;
and multiplying the weight value of each regional characteristic by the first tensor to obtain a second tensor.
In a specific implementation, the electronic device may extract the regional features of the spatial dimension and the traffic data dimension in the first vector by using a feature extraction model. The feature extraction model may be, but is not limited to, a convolutional neural network. The convolutional neural network can extract the regional characteristics of the input tensor by using convolutional kernels with different weights. Convolution kernels of different sizes can capture region features of different scales.
As shown in fig. 3, in the feature extraction model, the first tensor is filtered and calculated by a plurality of convolution kernels to obtain an intermediate tensor pi, and the size of the intermediate tensor pi is H × W × T, where H represents the length of a spatial dimension, W represents the length of a traffic data dimension, and T represents the length of a time dimension.
The electronic device can input the intermediate tensor pi into the attention fusion network to obtain the weight value epsilon of each region feature in the intermediate tensor. As shown in fig. 4, the attention fusion network may obtain the weight value epsilon of each region feature in the intermediate tensor by averaging each plane of the intermediate tensor pi and compressing by 1 × 1 × T.
Then, as shown in fig. 4, the electronic device may multiply the weight value epsilon of each region feature by the first tensor to obtain a second tensor of depth fusion.
Through the mode, the electronic equipment can endow different weights to the characteristics of different areas of the first tensor, and then the multidimensional traffic data tensor with the attention mechanism is obtained through fusion, so that the accuracy of traffic analysis can be further improved.
In this embodiment of the application, optionally, after the fusion of the spatial dimension and the traffic data dimension in the first tensor by using the attention mechanism to obtain the second tensor, before the determining the target information of the target road network according to the third tensor, the method further includes:
detecting whether the second tensor has an empty value;
and under the condition that the second tensor has the vacancy value, utilizing an interpolation method based on a weight value to fill the vacancy value to obtain a third tensor.
In this alternative embodiment, it is contemplated that the detector may be missing acquired data due to at least one of the following factors: the detector is disturbed by shelters, buildings or bad weather; debugging of the detector, errors of the built-in program and the like. In order to improve the accuracy of the traffic analysis, missing data may be complemented. The filling-up can be to utilize the collected traffic data to perform numerical interpolation or reasonable prediction at the missing part, so that the data is reasonable and complete.
In specific implementation, the vacancy value can be filled in the following modes:
1) and searching the vacancy value of each dimension layer, and positioning the position of the vacancy value. Given a traffic data variable riE distribution of R over spatiotemporal layers:
judging the vacancy value as c =(s)1,tk,ri) I.e. in space tkAt a time s1Is short of traffic data riThe value of (c).
2) And (4) adopting an interpolation method based on weight to fill the vacancy value. The vacancy values are related to the variable values of the state of the front-back time sequence, the left-right space sequence and the variable, and therefore the expectation is obtained for the variables at the front-back position, the left-right position and the left-right position of the variable in a weighting mode to fill in the vacancy values at the center of the vacancy.
Wherein the weighting coefficient alpha1、α2、α3、α4The influence degree of the vacancy value by the surrounding spatiotemporal variables is determined, and the influence degree can be set according to actual requirements, and the method is not limited in the embodiment of the application. In general, the greater the degree of influence of the surrounding variables on the traffic state value at the vacancy, the higher the weighting coefficient.
It should be noted that the above formula for c is only an example, and the calculation manner of c is not limited thereby, for example, c may also be calculated by the following formula:
the following describes the determination of object information of an object road network in the present embodiment.
Firstly, the target information comprises a traffic mode.
In this case, optionally, the determining the target information of the target road network according to the third tensor includes:
carrying out tensor decomposition on the third tensor to obtain a first factor matrix corresponding to a time dimension, a second factor matrix corresponding to a space dimension and a third factor matrix corresponding to the traffic data;
and determining the traffic mode of the target road network according to the first factor matrix, the second factor matrix and the third factor matrix.
In this alternative embodiment, the electronic device determines the traffic pattern using tensor (CP) decomposition. In other embodiments, the electronic device may determine the traffic pattern using the third tensor in other manners, for example, the electronic device may determine the traffic pattern using a second model, where the input of the second model is the third tensor, and the output is the first factor matrix, the second factor matrix, and the third factor matrix, and the embodiment of the present application is not limited to the manner of determining the traffic pattern according to the third tensor.
The traffic pattern, which may also be referred to as a travel pattern or a traffic behavior pattern, reflects the inherent similarity between the traffic states of individuals or road segments within a given time-space range for each individual or road segment traveled. In practical applications, individuals or road segments with significantly similar links may be classified as one and the same traffic pattern.
Assuming that the length of the spatial dimension of the third tensor is n road segments, the length of the time dimension is m time points, the length of the traffic data dimension is i-type traffic data, n and m are positive integers, and i is an integer greater than 1.
Then, the size of a first factor matrix a corresponding to the time dimension obtained by CP decomposition of the third tensor is i × Q, the size of a second factor matrix B corresponding to the space dimension is n × Q, and the size of a third factor matrix C corresponding to the traffic data is i × Q, that is, the first factor matrix is an m × Q matrix, the second factor matrix B is an n × Q matrix, and the third factor matrix C is an i × Q matrix. And Q represents the number of the traffic modes of the target road network, and is a positive integer.
Then, the electronic device may determine a traffic mode of the target road network based on the decomposed first factor matrix, the decomposed second factor matrix, and the decomposed third factor matrix.
Optionally, the determining the traffic mode of the target road network according to the first factor matrix, the second factor matrix and the third factor matrix includes:
obtaining Q groups of traffic data sequences according to the first factor matrix, the second factor matrix and the third factor matrix;
and acquiring Q traffic modes corresponding to the Q groups of traffic data sequences.
In a specific implementation, the electronic device may sequentially extract and expand the first factor matrix a, the second factor matrix B, and the third factor matrix C obtained Q times according to the row or column Q =1,2,3, …, so as to obtain Q sets of traffic data sequences (which may also be referred to as a travel mode sequence or a state sequence). Each set of travel modes comprises three vectors of m × 1, n × 1, i × 1, and each set of travel modes can be used to determine one traffic mode, so that Q sets of traffic data sequences can reveal traffic modes with different Q types of features.
The solving of the first factor matrix a, the second factor matrix B and the third factor matrix C can be implemented by:
optionally, the determining the traffic mode of the target road network according to the first factor matrix, the second factor matrix and the third factor matrix includes:
constructing an objective function, wherein the objective function is used for reflecting a difference between a first value and a second value, the first value is a value of the third tensor, and the second value is calculated based on the first factor matrix, the second factor matrix and the third factor matrix;
optimizing the first factor matrix, the second factor matrix and the third factor matrix according to the objective function;
and determining the traffic mode of the target road network according to the optimized first factor matrix, the optimized second factor matrix and the optimized third factor matrix.
Taking the third tensor as X, the relationship between X and the first factor matrix a, the second factor matrix B, and the third factor matrix C can be expressed by the following formula:
the relationship of the elements in X to the elements in the first, second and third factor matrices a, B, C can be represented by the following formula:
constructing an objective function f, the objective function f reflecting X and the second value
The difference between them. Alternatively, the objective function f may be represented by the following formula:
the objective function f can be simplified as:
to optimize the factor matrix, the optimization decision variable a can be solved by using a gradient descent methodmr，bnr，cir. Optionally, the partial derivatives of the decision variables may be calculated first, the decision variables are updated by using the partial derivatives, and the updating is iterated for a plurality of times until the objective function converges, that is, the optimum is achieved, that is, the difference between the first value and the second value is minimized, and the second value is infinitely close to the first value. It can be understood that when the objective function is optimal, each factor matrix may be optimal, and the optimal first factor matrix, second factor matrix, and third factor matrix may be utilized to determine the traffic mode of the target road network, so as to further improve the accuracy of determining the traffic mode.
The updating of the individual decision variables can be achieved by the following formula:
secondly, the target information comprises traffic laws.
The traffic space-time law is reflected in a given space-time range, and in each individual trip or trip road section, the individual trip or the road section presents similar or same states in a specific time period and space. Traffic laws can be characterized by the average of various traffic data within a target time or target space.
In this case, optionally, the determining the target information of the target road network according to the third tensor comprises at least one of the following items:
1) acquiring an average value of kth-class traffic data of the target road network in a target time length;
2) acquiring an average value of kth-class traffic data of a target space of the target road network in the first time period;
wherein the time unit of the target duration comprises at least one of: day; week; year; the spatial unit of the target space includes at least one of: a lane; a road segment; a road network; k ranges from 1 to i.
For 1), the electronic device may perform daily and/or weekly and/or yearly traffic status analysis.
On a time scale, taking the hour, the seven days of the week and the twelve months of the year detected by the detector as time units (also called as time scale or aggregation unit) respectively, various traffic data are counted:
wherein, akAnd the average value of the ith traffic state of the target road network in the (o, p) time period is shown.
When the time scale is day, the continuous statistical time of (o, p) is 1 hour, all the state values of 24 hours in one day are calculated, and a day traffic state analysis table is obtained.
When the time scale is week, the continuous statistical time of (o, p) is 1 day, and the state values of all 7 days in the week are calculated to obtain a week traffic state analysis table.
When the time scale is year, (o, p) is 1 month, all the state values of 12 months in one year are calculated to obtain an annual traffic state analysis table.
For 2), the electronic device may perform lane-level and/or road segment-level and/or road network-level traffic status analysis.
On the spatial scale, taking the lanes, the road sections and the road network detected by the detector as a collection unit respectively, counting various traffic states:
wherein, bkIs shown at tnThe (q, u) th road segment at the time, the average of the i-th traffic state. The traffic state table of a certain lane level, a certain road section or a certain road network level in the city at a certain time can be obtained by using the formula.
And thirdly, the target information comprises a future traffic state.
Based on the third tensor, a prediction of traffic status at a future time may be made. And predicting the traffic state, wherein the traffic state is reflected in a given time range in the future, and various traffic states in the road section detected by the millimeter wave radar are predicted. The multi-dimensional traffic data tensor established based on the data detected by the detector has strong space-time dependency relationship, and can make accurate prediction on the future traffic state by utilizing the close space-time dependency relationship.
Optionally, the determining the target information of the target road network according to the third tensor comprises:
inputting the third tensor into a trained first model, and predicting to obtain a fourth tensor, wherein the fourth tensor is used for reflecting the traffic state of the target road network in a second time length, and the second time length is the next time length of the first time length;
wherein a first hidden state when the first model predicts the fourth tensor is determined by the third tensor and a historical hidden state, the historical hidden state is determined by a hidden state of the first model in a third duration, and the third duration is a duration previous to the first duration; the hidden state of the first model is used to determine the output of the first model.
In this optional embodiment, the traffic state at a future time is predicted by using a first model, where the input of the first model is the third tensor, and the output is a future time length tensor corresponding to the third tensor, and the third tensor can be used for reflecting the traffic state of the target road network within a narrow time length.
Structure of the first model referring to fig. 5, the first model predicting future traffic conditions may include the following three steps:
1) various variables are input.
For convenience of understanding, the time duration is referred to as the time of day, and the future time duration is the time of the futureAnd (6) engraving. Sorting variables into sequential inputs x according to time series(1),x(2),…,x(t)And each represents a third tensor calculated by the electronic device at time t. Other variables of the prediction model are:
h(t)representing the hidden state of the model at time t, h(t)By input x at corresponding time(t)And hidden state h at the previous moment(t-1)And (4) jointly determining.
o(t)Output of the table model at time t, o(t)Hidden state h only present by model(t)And (6) determining.
L(t)Representing the loss function of the model at time t.
y(t)Representing the true value y at time t(t)。
U, V, W represent the shared weight matrix in the prediction model.
2) Forward propagation computation.
Based on the prescribed variables, forward propagation calculation of the model is performed. For any time t, the hidden state h of the time(t)From the multidimensional traffic state tensor x of that moment(t)And hidden state h at the previous moment(t-1)Obtaining:
by pairs
Carrying out twice sharing weight matrix operation on hidden variables at the moment to obtain an estimated value of the moment to the multidimensional traffic state
。
3) And (4) backward propagation optimization.
In order to accurately predict the future traffic state, the estimated value of the traffic state obtained at the last moment needs to be measured
And the real traffic state value y(t)The difference between the two is quantified and a loss function (which may also be referred to as a difference function) L is established between the two.
Optionally, the inputting the third tensor into the trained first model, and before predicting the fourth tensor, the method further includes:
acquiring a sample tensor;
inputting the sample tensor into an untrained first model to obtain a fifth tensor;
determining a loss function corresponding to the first model according to the sample tensor and the fifth tensor;
and adjusting the weight value of the first model according to the loss function to obtain the trained first model.
The loss function can be calculated by the following formula:
and (3) solving the partial derivatives of the decision variables c and V of the difference function L through a gradient descent algorithm, performing iterative computation for multiple rounds, and continuously updating c and V to optimize the model so as to realize accurate prediction.
And (3) gradually solving partial derivatives of the decision variables U, W and b of the difference function L by a gradual gradient descent method, iterating for multiple rounds, and continuously updating to optimize the model so as to realize accurate prediction.
Gradually updating W:
updating U:
updating b:
and during model training, performing iterative optimization on model parameters based on the loss function, and when the loss function is smaller than a specified threshold value, not updating. And inputting the multidimensional traffic state variable of the moment into the model, and calculating to obtain the multidimensional traffic state variable of the corresponding space of the next moment, thereby obtaining a prediction result.
The various optional implementations described in the embodiments of the present application may be implemented in combination with each other or implemented separately without conflicting with each other, and the embodiments of the present application are not limited to this.
The embodiment of the application comprises the following contents:
and the electronic equipment collects the detection data of each millimeter wave radar detection point of the target road network. And splicing time and space dimensions of continuous millimeter wave radar data sets in the road network according to the result of positioning and matching the pile number of the millimeter wave radar with the map, and restoring the traffic state of the lane level. The traffic state information contained in each data tensor is not less than two types, including but not limited to the information of average speed, occupancy rate, number of running vehicles and the like of each lane in the road network.
And the electronic equipment outputs the spliced and fused millimeter wave radar data. According to the millimeter wave radar multi-source data fusion algorithm, the lane-level data tensors formed by splicing are subjected to deep fusion, and a millimeter wave radar multi-dimensional traffic state tensor with multi-source information fused is obtained.
And the millimeter wave radar data formed after the integration of the electronic equipment defect filling.
The electronic device outputs a predicted traffic state value for the lane at a future time.
The embodiment of the application has at least the following beneficial effects:
the method and the device have the advantages that an effective data format aiming at millimeter wave radar data is adopted, an effective urban road network lane-level algorithm is provided based on the effective data format, the travel mode analysis of traffic network data can be realized, and the effect of accurately predicting the traffic state of the specified road at a certain time interval in the future can be realized.
According to the method, the data of all time ranges and all area ranges covered by the millimeter wave radar are accurately spliced, so that the deep fusion of the multi-source data generated by the millimeter wave radar is realized, and the accuracy of perception and prediction is improved; and the missing data is processed, so that the method is suitable for the data problem existing in a real scene, and lays a foundation for further use of millimeter wave radar data in the follow-up process.
The data analysis and prediction method of the multi-dimensional millimeter wave radar can adapt to traffic road side equipment transmission in a 5G scene, and meanwhile can be oriented to distributed calculation and processing of data in edge calculation. The method and the device can not only realize travel mode analysis of traffic network data in the prior art environment, but also adapt to rapid traffic data processing and calculation in future emerging technical scenes.
Referring to fig. 6, fig. 6 is a structural diagram of a traffic analysis apparatus according to an embodiment of the present application. As shown in fig. 6, the traffic analysis device 600 includes:
a generating module 601, configured to generate a first tensor according to traffic data detected by each detector in a target road network within a first time duration, where a length of a spatial dimension of the first tensor is S road segments in the target road network, a length of a time dimension is T time points in the first time duration, a length of a traffic data dimension is R-class traffic data detected by the detector, S and T are positive integers, and R is an integer greater than 1;
a fusion module 602, configured to fuse a spatial dimension and a traffic data dimension in the first tensor by using an attention mechanism to obtain a second tensor;
a first determining module 603, configured to determine, according to a third tensor, target information of the target road network;
wherein the third tensor is the second tensor or determined based on the second tensor; the target information includes at least one of: traffic patterns, traffic laws, and future traffic conditions.
Optionally, the generating module 601 includes:
the generation submodule is used for generating a time sequence corresponding to the T time points detected by each detector in the target road network and a space sequence corresponding to the S road sections in the target road network;
and the splicing submodule is used for splicing the value of the R-type traffic data corresponding to each time point in the time sequence and the value of the R-type traffic data corresponding to each road section in the space sequence to obtain a first quantity.
Optionally, the fusion module 602 includes:
the extraction submodule is used for extracting the regional characteristics of the space dimensionality and the traffic data dimensionality in the first tensor to obtain an intermediate tensor;
the first obtaining submodule is used for inputting the middle tensor into an attention fusion network to obtain the weight value of each region feature in the middle tensor;
and the second obtaining submodule is used for multiplying the weight value of each regional characteristic by the first tensor to obtain a second tensor.
Optionally, the traffic analysis apparatus 600 further comprises:
a detection module, configured to detect whether the second tensor has an empty value;
and the first acquisition module is used for supplementing the vacancy value by using an interpolation method based on a weight value under the condition that the vacancy value exists in the second tensor, so as to obtain a third tensor.
Optionally, in a case that the target information includes a traffic mode, the first determining module 603 includes:
the decomposition submodule is used for carrying out tensor decomposition on the third tensor to obtain a first factor matrix corresponding to a time dimension, a second factor matrix corresponding to a space dimension and a third factor matrix corresponding to the traffic data;
and the determining submodule is used for determining the traffic mode of the target road network according to the first factor matrix, the second factor matrix and the third factor matrix.
Optionally, the determining sub-module includes:
a constructing unit, configured to construct an objective function, where the objective function is configured to reflect a difference between a first value and a second value, the first value is a value of the third tensor, and the second value is calculated based on the first factor matrix, the second factor matrix, and the third factor matrix;
an optimization unit, configured to optimize the first factor matrix, the second factor matrix, and the third factor matrix according to the objective function;
and the determining unit is used for determining the traffic mode of the target road network according to the optimized first factor matrix, the optimized second factor matrix and the optimized third factor matrix.
Optionally, the length of the spatial dimension of the third tensor is n road segments, the length of the time dimension is m time points, the length of the traffic data dimension is i-type traffic data, n and m are positive integers, and i is an integer greater than 1;
the first factor matrix is an m multiplied by Q matrix, the second factor matrix is an n multiplied by Q matrix, the third factor matrix is an i multiplied by Q matrix, and Q is a positive integer;
the determination submodule includes:
the first acquisition unit is used for acquiring Q groups of traffic data sequences according to the first factor matrix, the second factor matrix and the third factor matrix;
and the second acquisition unit is used for acquiring Q traffic modes corresponding to the Q groups of traffic data sequences.
Optionally, the length of the traffic data dimension of the third tensor is i-type traffic data, and i is an integer greater than 1;
in case the target information comprises traffic laws, the first determining module 603 is configured to at least one of:
acquiring an average value of kth-class traffic data of the target road network in a target time length;
acquiring an average value of kth-class traffic data of a target space of the target road network in the first time period;
wherein the time unit of the target duration comprises at least one of: day; week; year; the spatial unit of the target space includes at least one of: a lane; a road segment; a road network; k ranges from 1 to i.
Optionally, in case the target information comprises a future traffic state, the first determining module 603 is configured to:
inputting the third tensor into a trained first model, and predicting to obtain a fourth tensor, wherein the fourth tensor is used for reflecting the traffic state of the target road network in a second time length, and the second time length is the next time length of the first time length;
wherein a first hidden state when the first model predicts the fourth tensor is determined by the third tensor and a historical hidden state, the historical hidden state is determined by a hidden state of the first model in a third duration, and the third duration is a duration previous to the first duration; the hidden state of the first model is used to determine the output of the first model.
Optionally, the traffic analysis apparatus 600 further comprises:
a second obtaining module, configured to obtain a sample tensor;
a third obtaining module, configured to input the sample tensor into an untrained first model to obtain a fifth tensor;
a second determining module, configured to determine a loss function corresponding to the first model according to the sample tensor and the fifth tensor;
and the adjusting module is used for adjusting the weight value of the first model according to the loss function to obtain the trained first model.
The traffic analysis device 600 can implement each process of the method embodiment in fig. 1 in the embodiment of the present application, and achieve the same beneficial effects, and is not described herein again to avoid repetition.
The embodiment of the application also provides the electronic equipment. Referring to fig. 7, the electronic device may include a processor 701, a memory 702, and a program 7021 stored in the memory 702 and operable on the processor 701, and when the program 7021 is executed by the processor 701, any step in the method embodiment corresponding to fig. 1 may be implemented and the same advantageous effect may be achieved, which is not described herein again.
Those skilled in the art will appreciate that all or part of the steps of the method according to the above embodiments may be implemented by hardware associated with program instructions, and the program may be stored in a readable medium. An embodiment of the present application further provides a readable storage medium, where a computer program is stored on the readable storage medium, and when the computer program is executed by a processor, any step in the method embodiment corresponding to fig. 1 may be implemented, and the same technical effect may be achieved, and in order to avoid repetition, details are not repeated here.
The storage medium may be a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk.
While the foregoing is directed to the preferred embodiment of the present application, it will be understood by those skilled in the art that various changes and modifications may be made without departing from the principles of the disclosure, and it is intended that such changes and modifications be considered as within the scope of the disclosure.
Claims (13)
1. A traffic analysis method, characterized in that the method comprises:
generating a first tensor according to traffic data detected by each detector in a target road network within a first time length, wherein the length of a space dimension of the first tensor is S road sections in the target road network, the length of a time dimension is T time points in the first time length, the length of a traffic data dimension is R-type traffic data detected by the detectors, S and T are positive integers, and R is an integer greater than 1;
fusing the space dimensionality and the traffic data dimensionality in the first tensor by using an attention mechanism to obtain a second tensor;
determining target information of the target road network according to the third tensor;
wherein the third tensor is the second tensor or determined based on the second tensor; the target information includes at least one of: traffic patterns, traffic laws, and future traffic conditions.
2. The method of claim 1, wherein generating the first vector according to the traffic data detected by each detector in the target road network within the first time period comprises:
generating a time sequence corresponding to T time points detected by each detector in a target road network and a space sequence corresponding to S road sections in the target road network;
and splicing the value of the R-type traffic data corresponding to each time point in the time sequence and the value of the R-type traffic data corresponding to each road section in the space sequence to obtain a first quantity.
3. The method of claim 1, wherein the fusing the spatial dimension and the traffic data dimension in the first tensor using the attention mechanism to obtain a second tensor comprises:
extracting the regional characteristics of the space dimensionality and the traffic data dimensionality in the first tensor to obtain an intermediate tensor;
inputting the intermediate tensor into an attention fusion network to obtain the weight value of each regional characteristic in the intermediate tensor;
and multiplying the weight value of each regional characteristic by the first tensor to obtain a second tensor.
4. The method according to claim 1, wherein after the fusion of the spatial dimension and the traffic data dimension in the first tensor by using the attention mechanism to obtain a second tensor, and before the determination of the target information of the target road network according to a third tensor, the method further comprises:
detecting whether the second tensor has an empty value;
and under the condition that the second tensor has the vacancy value, utilizing an interpolation method based on a weight value to fill the vacancy value to obtain a third tensor.
5. The method of claim 1, wherein in the case that the target information includes a traffic pattern, the determining the target information of the target road network according to the third tensor comprises:
carrying out tensor decomposition on the third tensor to obtain a first factor matrix corresponding to a time dimension, a second factor matrix corresponding to a space dimension and a third factor matrix corresponding to the traffic data;
and determining the traffic mode of the target road network according to the first factor matrix, the second factor matrix and the third factor matrix.
6. The method of claim 5, wherein said determining traffic patterns of said target road network based on said first factor matrix, said second factor matrix and said third factor matrix comprises:
constructing an objective function, wherein the objective function is used for reflecting a difference between a first value and a second value, the first value is a value of the third tensor, and the second value is calculated based on the first factor matrix, the second factor matrix and the third factor matrix;
optimizing the first factor matrix, the second factor matrix and the third factor matrix according to the objective function;
and determining the traffic mode of the target road network according to the optimized first factor matrix, the optimized second factor matrix and the optimized third factor matrix.
7. The method of claim 5, wherein the third tensor has a spatial dimension of n road segments, a temporal dimension of m time points, a traffic data dimension of i traffic data, n and m being positive integers, i being an integer greater than 1;
the first factor matrix is an m multiplied by Q matrix, the second factor matrix is an n multiplied by Q matrix, the third factor matrix is an i multiplied by Q matrix, and Q is a positive integer;
the determining the traffic mode of the target road network according to the first factor matrix, the second factor matrix and the third factor matrix comprises:
obtaining Q groups of traffic data sequences according to the first factor matrix, the second factor matrix and the third factor matrix;
and acquiring Q traffic modes corresponding to the Q groups of traffic data sequences.
8. The method of claim 1, wherein the length of the traffic data dimension of the third tensor is i-class traffic data, i being an integer greater than 1;
under the condition that the target information comprises a traffic rule, determining the target information of the target road network according to the third tensor, wherein the target information comprises at least one of the following items:
acquiring an average value of kth-class traffic data of the target road network in a target time length;
acquiring an average value of kth-class traffic data of a target space of the target road network in the first time period;
wherein the time unit of the target duration comprises at least one of: day; week; year; the spatial unit of the target space includes at least one of: a lane; a road segment; a road network; k ranges from 1 to i.
9. The method of claim 1, wherein in the case that the target information includes a future traffic state, the determining the target information of the target road network according to the third tensor comprises:
inputting the third tensor into a trained first model, and predicting to obtain a fourth tensor, wherein the fourth tensor is used for reflecting the traffic state of the target road network in a second time length, and the second time length is the next time length of the first time length;
wherein a first hidden state when the first model predicts the fourth tensor is determined by the third tensor and a historical hidden state, the historical hidden state is determined by a hidden state of the first model in a third duration, and the third duration is a duration previous to the first duration; the hidden state of the first model is used to determine the output of the first model.
10. The method of claim 9, wherein the third tensor is input into the trained first model and the method further comprises, before predicting the fourth tensor:
acquiring a sample tensor;
inputting the sample tensor into an untrained first model to obtain a fifth tensor;
determining a loss function corresponding to the first model according to the sample tensor and the fifth tensor;
and adjusting the weight value of the first model according to the loss function to obtain the trained first model.
11. A traffic analysis apparatus, comprising:
the system comprises a generating module, a calculating module and a processing module, wherein the generating module is used for generating a first tensor according to traffic data detected by each detector in a target road network within a first time length, the length of a space dimension of the first tensor is S road sections in the target road network, the length of a time dimension is T time points in the first time length, the length of a traffic data dimension is R-type traffic data detected by the detector, S and T are positive integers, and R is an integer greater than 1;
the fusion module is used for fusing the space dimensionality and the traffic data dimensionality in the first tensor by using an attention mechanism to obtain a second tensor;
a first determining module, configured to determine, according to a third tensor, target information of the target road network;
wherein the third tensor is the second tensor or determined based on the second tensor; the target information includes at least one of: traffic patterns, traffic laws, and future traffic conditions.
12. An electronic device, comprising: a transceiver, a memory, a processor, and a program stored on the memory and executable on the processor; characterized by a processor for reading a program in a memory implementing the traffic analysis method according to any one of claims 1 to 10.
13. A readable storage medium storing a program, wherein the program, when executed by a processor, implements the traffic analysis method according to any one of claims 1 to 10.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110727130.5A CN113256986B (en) | 2021-06-29 | 2021-06-29 | Traffic analysis method, related device and readable storage medium |
| PCT/CN2022/095206 WO2023273724A1 (en) | 2021-06-29 | 2022-05-26 | Traffic analysis method, related device, and readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110727130.5A CN113256986B (en) | 2021-06-29 | 2021-06-29 | Traffic analysis method, related device and readable storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113256986A true CN113256986A (en) | 2021-08-13 |
| CN113256986B CN113256986B (en) | 2021-10-26 |
Family
ID=77190268
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110727130.5A Active CN113256986B (en) | 2021-06-29 | 2021-06-29 | Traffic analysis method, related device and readable storage medium |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN113256986B (en) |
| WO (1) | WO2023273724A1 (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2023273724A1 (en) * | 2021-06-29 | 2023-01-05 | 中移(上海)信息通信科技有限公司 | Traffic analysis method, related device, and readable storage medium |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2017220000A (en) * | 2016-06-07 | 2017-12-14 | 日本電信電話株式会社 | Flow rate prediction device, pattern estimation device, flow rate prediction method, pattern estimation method and program |
| JP2018195244A (en) * | 2017-05-22 | 2018-12-06 | 日本電信電話株式会社 | Od traffic prediction device, method, and program |
| CN109215372A (en) * | 2018-10-15 | 2019-01-15 | 百度在线网络技术（北京）有限公司 | Road network information update method, device and equipment |
| CN110428614A (en) * | 2019-07-11 | 2019-11-08 | 银江股份有限公司 | A kind of traffic congestion temperature spatio-temporal prediction method based on non-negative tensor resolution |
| CN110766942A (en) * | 2019-10-18 | 2020-02-07 | 北京大学 | Traffic network congestion prediction method based on convolution long-term and short-term memory network |
| CN110827546A (en) * | 2019-11-21 | 2020-02-21 | 银江股份有限公司 | Signalized intersection road section short-term speed prediction method |
| US10853720B1 (en) * | 2017-04-26 | 2020-12-01 | EMC IP Holding Company LLC | Traffic condition forecasting using matrix compression and deep neural networks |
| CN112734100A (en) * | 2020-12-31 | 2021-04-30 | 北京航空航天大学 | Road network travel time prediction method based on tensor neural network |
| CN112988760A (en) * | 2021-04-27 | 2021-06-18 | 北京航空航天大学 | Tensor decomposition-based traffic space-time big data missing completion method |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113256986B (en) * | 2021-06-29 | 2021-10-26 | 中移(上海)信息通信科技有限公司 | Traffic analysis method, related device and readable storage medium |
-
2021
- 2021-06-29 CN CN202110727130.5A patent/CN113256986B/en active Active
-
2022
- 2022-05-26 WO PCT/CN2022/095206 patent/WO2023273724A1/en unknown
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2017220000A (en) * | 2016-06-07 | 2017-12-14 | 日本電信電話株式会社 | Flow rate prediction device, pattern estimation device, flow rate prediction method, pattern estimation method and program |
| US10853720B1 (en) * | 2017-04-26 | 2020-12-01 | EMC IP Holding Company LLC | Traffic condition forecasting using matrix compression and deep neural networks |
| JP2018195244A (en) * | 2017-05-22 | 2018-12-06 | 日本電信電話株式会社 | Od traffic prediction device, method, and program |
| CN109215372A (en) * | 2018-10-15 | 2019-01-15 | 百度在线网络技术（北京）有限公司 | Road network information update method, device and equipment |
| CN110428614A (en) * | 2019-07-11 | 2019-11-08 | 银江股份有限公司 | A kind of traffic congestion temperature spatio-temporal prediction method based on non-negative tensor resolution |
| CN110766942A (en) * | 2019-10-18 | 2020-02-07 | 北京大学 | Traffic network congestion prediction method based on convolution long-term and short-term memory network |
| CN110827546A (en) * | 2019-11-21 | 2020-02-21 | 银江股份有限公司 | Signalized intersection road section short-term speed prediction method |
| CN112734100A (en) * | 2020-12-31 | 2021-04-30 | 北京航空航天大学 | Road network travel time prediction method based on tensor neural network |
| CN112988760A (en) * | 2021-04-27 | 2021-06-18 | 北京航空航天大学 | Tensor decomposition-based traffic space-time big data missing completion method |
Non-Patent Citations (3)
| Title |
|---|
| MUHAMMAD TAYYAS ASIF ET AL.: "Matrix and Tensor Based Methods for Missing Data Estimation in Large Traffic Networks", 《 IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS》 * |
| 程川云: "基于张量模型的短时交通流量预测技术研究", 《万方学位论文全文数据库》 * |
| 胡雪等: "张量表达下的多模态交通缺失数据补全算法", 《小型微型计算机系统》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2023273724A1 (en) * | 2021-06-29 | 2023-01-05 | 中移(上海)信息通信科技有限公司 | Traffic analysis method, related device, and readable storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113256986B (en) | 2021-10-26 |
| WO2023273724A1 (en) | 2023-01-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Sameen et al. | Application of convolutional neural networks featuring Bayesian optimization for landslide susceptibility assessment | |
| Kumar et al. | A comparative study of supervised machine learning algorithms for stock market trend prediction | |
| CN110164128B (en) | City-level intelligent traffic simulation system | |
| CN110675623B (en) | Short-term traffic flow prediction method, system and device based on hybrid deep learning | |
| CN113496104B (en) | Precipitation prediction correction method and system based on deep learning | |
| CN111612243A (en) | Traffic speed prediction method, system and storage medium | |
| CN111582559B (en) | Arrival time estimation method and device | |
| CN112508173A (en) | Traffic space-time sequence multi-step prediction method, system and storage medium | |
| CN103366602A (en) | Method of determining parking lot occupancy from digital camera images | |
| CN113591380B (en) | Traffic flow prediction method, medium and equipment based on graph Gaussian process | |
| CN114664091A (en) | Early warning method and system based on holiday traffic prediction algorithm | |
| CN111047078B (en) | Traffic characteristic prediction method, system and storage medium | |
| CN114241053B (en) | Multi-category tracking method based on improved attention mechanism FairMOT | |
| CN104156943A (en) | Multi-target fuzzy cluster image variance detecting method based on non-control-neighborhood immune algorithm | |
| CN113537462A (en) | Data processing method, neural network quantization method and related device | |
| CN113256986B (en) | Traffic analysis method, related device and readable storage medium | |
| CN110163294A (en) | Remote Sensing Imagery Change method for detecting area based on dimensionality reduction operation and convolutional network | |
| CN116311091A (en) | Vehicle counting method based on pyramid density perception attention network | |
| Sunindyo et al. | Traffic congestion prediction using multi-layer perceptrons and long short-term memory | |
| CN116580322A (en) | Unmanned aerial vehicle infrared small target detection method under ground background | |
| CN115995042A (en) | Video SAR moving target detection method and device | |
| CN114842370A (en) | Computer-implemented method for training a computer vision model | |
| CN114120635A (en) | Tensor decomposition-based urban road network linear missing flow estimation method and system | |
| Xu et al. | Dynamic spatiotemporal graph wavelet network for traffic flow prediction | |
| CN117131991A (en) | Urban rainfall prediction method and platform based on hybrid neural network |
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 |