CN113469425A - Deep traffic jam prediction method - Google Patents
Deep traffic jam prediction method Download PDFInfo
- Publication number
- CN113469425A CN113469425A CN202110697865.8A CN202110697865A CN113469425A CN 113469425 A CN113469425 A CN 113469425A CN 202110697865 A CN202110697865 A CN 202110697865A CN 113469425 A CN113469425 A CN 113469425A
- Authority
- CN
- China
- Prior art keywords
- road
- data
- road section
- data set
- traffic
- 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
- 238000000034 method Methods 0.000 title claims abstract description 35
- 238000004891 communication Methods 0.000 claims abstract description 18
- 238000013135 deep learning Methods 0.000 claims abstract description 17
- 239000013598 vector Substances 0.000 claims abstract description 4
- 239000011159 matrix material Substances 0.000 claims description 19
- 230000005540 biological transmission Effects 0.000 claims description 18
- 230000008569 process Effects 0.000 claims description 10
- 230000000295 complement effect Effects 0.000 claims description 9
- 238000012360 testing method Methods 0.000 claims description 8
- 238000004364 calculation method Methods 0.000 claims description 6
- 230000008859 change Effects 0.000 claims description 4
- 238000011056 performance test Methods 0.000 claims description 3
- 239000004576 sand Substances 0.000 claims description 3
- 230000002123 temporal effect Effects 0.000 claims description 3
- 238000012549 training Methods 0.000 description 4
- 230000007423 decrease Effects 0.000 description 2
- 238000013136 deep learning model Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 239000000126 substance Substances 0.000 description 2
- 241001522296 Erithacus rubecula Species 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 230000002457 bidirectional effect Effects 0.000 description 1
- 239000002131 composite material Substances 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000002950 deficient Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
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
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- 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
- G06Q10/067—Enterprise or organisation modelling
-
- 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
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/26—Government or public services
-
- 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
Abstract
The invention discloses a deep traffic jam prediction method, which comprises the following steps: step 1, a network server stores road traffic information, establishes a sample set of road section flow vectors and obtains a historical data set; step 2, converting the traffic network structure information into a topological graph structure; step 3, obtaining the degree information of the nodes and selecting selected road section data in the historical data set according to the degree information of the nodes to obtain a missing data set; step 4, complementing the missing road data to obtain a supplemented data set; and 5, inputting the completion data set into a deep learning flow prediction model to predict the road congestion. The prediction method has high prediction accuracy and reduces communication overhead; the amount of data transmitted is reduced and the impact of missing data on the predictive performance is reduced.
Description
Technical Field
The invention relates to the technical field of traffic congestion prediction, in particular to a deep traffic congestion prediction method.
Background
The modern urban road traffic has a serious congestion problem, and congestion prediction becomes an effective means for relieving traffic pressure and avoiding congestion. A large amount of traffic data is sent to a network server in real time through various terminal devices (such as vehicle-mounted units or road side units), and the server analyzes and processes vehicle congestion data of different time and different road sections through a big data and deep learning algorithm, extracts features and explores rules to predict the congestion situation of a future road.
Although the current traffic jam prediction technology is relatively mature, a deep learning model generally needs a large amount of data to support so as to train a model with relatively stable and good predictive performance, but meanwhile, in the process of sending a large amount of data from a terminal to a network server, the cost and resource occupation caused by the deep learning model cause huge pressure on communication transmission, and how to reduce the communication cost as much as possible and ensure that the accuracy of traffic jam prediction is relatively deficient in the current research.
Based on the technical defects in the prior art, the invention provides a deep traffic jam prediction method.
Disclosure of Invention
In order to solve the technical problems of the existing system, the invention provides a deep traffic jam prediction method, which carries out real-time traffic jam prediction by converting an actual traffic road into a topological network, selecting road sections based on node degrees, complementing missing road section data, establishing and training a deep prediction model, determining the optimal number of transmitting road sections according to a test result and carrying out the real-time traffic jam prediction.
The invention adopts the following technical scheme:
a method of deep traffic congestion prediction, comprising:
step 4, complementing the missing road data to obtain a supplemented data set;
and 5, inputting the completion data set into a deep learning flow prediction model to predict the road congestion.
Further, in step 1, the road traffic information includes traffic network structure information and historical traffic data information.
Further, in step 2, the step of converting the traffic network structure information into a topological graph structure includes: defining a directed graph network G (V, E) according to the traffic roads, and taking the roads as nodes V ═ V in the directed graph network1,v2,…,vNN represents the number of road segments, and the road segment intersection is taken as the edge E of the node in the directed graph network, i.e. { v ═ v }ivj},vivjIndicating a road viCan directly reach the road vj。
Further, in step 3, according to the directed graph network G (V, E), the adjacency matrix a E R of the directed graph network G is obtained by calculationN×N,A∈RN×NRepresenting the connection relationship between roads, the matrix contains 1 and 0 elements, wherein 1 represents directly reachable, and 0 represents not directly reachable, i.e. the following formula (1):
calculating the outgoing degree matrix of the directed graph network G by the A, namely the following formula (2):
D=∑jΑij……(2),
obtaining the node degree according to the formula (1) and the formula (2), selecting and selecting the road section data, and setting the number of the finally selected road sections as NtThen set the selected road section set VsAnd non-selected road section set VnDividing all road sections, including:
selecting, namely, selecting the road section with the maximum node degree and putting the road section into the selection set VsIn the method, considering that information redundancy exists between adjacent road sections, the neighbor road section which can be directly reached by the largest road section is temporarily put into a non-selected set VnRepeatedly executing the process until only isolated road sections without neighbors are left;
put back the non-selected set VnThe road sections in the road section set are put back into the whole road section set again;
repeating the selecting and replacing steps;
selecting road section set V at any timesThe number of segments in (1) is NtWhen so, the process ends.
Further, in step 4, setting the matrix P to indicate whether the road segment data is missing, the corresponding value of the selected road segment is 1, the corresponding value of the non-selected road segment is 0, and the server calculates the missing data set X from the complete historical data set X to obtain the missing data set X1Namely, the following formula (3):
X1=X*P……(3),
in the above formula (3), the expression matrix is multiplied by a dot, that is, the corresponding position elements are multiplied, the historical traffic data set X is divided into seven days per week, the congestion indexes of a certain road section per week X one day are added by week and then averaged to obtain the historical average value of the congestion indexes of week X, the congestion condition of the road section can be affected by the congestion condition of the neighboring road section, and the historical change rule of the road section is corrected according to the information of the neighboring road section, and the calculation method is as follows:
defining missing segment completion values consists of two parts, namely the following equation (4):
in the above formula (4), vnIndicating missing segments, i.e. segments that need to be filled,representing a section of road vnOf { v } vgDenotes a section vnAll the road sections selected in the neighborhood are selected,representing the average value of the real value of the neighbor selected road section, wherein the proportional coefficient alpha + beta of the road section historical information and the neighbor real information is 1;
calculating the mean value of the real values of the neighbor selected road sections, as shown in the following formula (5):
calculating selected road section { v) in neighbor road sectionsgThe ratio of (6):
in the above formula (6), { vcDenotes a section vnAll neighbors of (1), including the selected road segment { v }gAnd missing road section, coefficientTo adjust the value of β;
the beta value is dynamically adjusted according to the number of the surrounding selected neighbor segments to determine the influence of neighbor information on a compensation value, and aiming at different selected segment numbers NtAccording to the completed data set X2Separately calculating to minimize the mean absolute error of the complement and the accurate dataThe following formula (7):
further, in step 5, inputting the completion data set into a deep learning flow prediction model, training the model and performing performance test, and obtaining the optimal number of sending segments according to the test result.
Further, in step 5, the road congestion prediction includes:
the network server obtains the optimal number of the sending road sections, determines whether the optimal number of the sending road sections is sent to the terminal equipment or not, and feeds back the optimal number information of the sending road sections to the vehicle or the road side unit;
the vehicle or the road side unit selectively sends the current road data according to the indication of the network server, the network server obtains a missing data set, and completes the data which is not sent to obtain a real-time completed data set;
and the network server inputs the completion data set into a deep learning flow prediction model to predict the road congestion.
Further, in step 5, the deep learning traffic prediction model includes a graph convolution network GCN for capturing data spatial correlation and a gating loop unit GRU for capturing data temporal correlation.
Further, in step 5, a complementary data set X is used2Inputting a deep learning flow prediction model and testing, defining the communication learning efficiency eta as a ratio of a square root of prediction congestion index useful power to transmission data overhead, wherein the communication learning efficiency eta is higher in value when the transmission data overhead is smaller as the useful power is larger, and is calculated according to the following formula (8):
in the above-mentioned formula (8),a predicted value representing a congestion index is displayed,representing absolute error between true and predicted values, the overhead of transmitting data being the number of transmit segments NtA coefficient u represents a variable cost in relation to the number of transmission links, a coefficient v represents a fixed cost, and u and v are variable according to a communication system model, that is, the number of transmission links for which communication learning efficiency is the maximum is the optimal number, as shown in the following equation (9):
No=argmax{η(Nt)}……(9)。
1. according to the deep traffic jam prediction method, missing road data are filled through a completion algorithm, so that the transmitted data volume is greatly reduced, and the influence of the missing data on the prediction performance is reduced;
2. according to the deep traffic jam prediction method, the spatial characteristics and the time characteristics of the data can be respectively extracted through the composite structure of the GCN and the GRU, and the time-space dynamic change rule of the data is explored, so that the prediction accuracy rate is high;
3. the deep traffic jam prediction method provided by the invention can reduce the influence on the prediction performance as much as possible while reducing the data sending quantity by using the road section selection and completion algorithm, and can effectively reduce the communication overhead.
Drawings
FIG. 1 is a schematic flow chart of a deep traffic congestion prediction method according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of a road segment week X congestion index in an embodiment of the present invention;
FIG. 3 is a TGCN network structure diagram in an embodiment of the present invention;
fig. 4 is a graph showing the variation of the average absolute error and the communication learning efficiency with the number of transmission links in the embodiment of the present invention.
Detailed Description
So that the manner in which the above recited objects, features and advantages of the present invention can be understood in detail, a more particular description of the invention, briefly summarized above, may be had by reference to the embodiments thereof which are illustrated in the appended drawings. It should be noted that the embodiments of the present invention and features of the embodiments may be combined with each other without conflict.
Example 1
As shown in fig. 1, the method for predicting deep traffic congestion includes:
step 4, complementing the missing road data to obtain a supplemented data set;
and 5, inputting the completion data set into a deep learning flow prediction model, and performing road congestion prediction to the graph network.
In step 1, the road traffic information includes traffic network structure information and historical traffic data information.
In the above embodiment, the historical traffic data information includes historical traffic timing data based on links within the road network, including flow values, collection times, road attributes, and the like.
In step 2, the step of converting the traffic network structure information into a topological graph structure comprises the following steps: a directed graph network G (V, E) is defined from traffic roads, and since an actual traffic network includes bidirectional lanes, the road network G defined here is a directed graph network in which roads are taken as nodes V ═ V in the directed graph network1,v2,…,vNN represents the number of road segments, and the road segment intersection is taken as the edge E of the node in the directed graph network, i.e. { v ═ v }ivj},vivjIndicating a road viCan directly reach the road vj;
In step 3, according to the directed graph network G (V, E), calculating to obtain an adjacency matrix A epsilon R of the directed graph network GN×N,A∈RN×NRepresenting the connection relation between roads, the matrix contains 1 and 0 elements, wherein 1 represents directly reachable, and 0 represents not directly reachable, that is:
and calculating to obtain an outgoing matrix of the directed graph network G by A:
D=∑jΑij……(2),
the larger the degree of the node is, the node isThe more information contained, the more important the congestion index prediction, the road section selection algorithm based on the node degree is provided by selecting the selected road section according to the node degree, the node degree is obtained according to the formula 1-2, and the number of the road sections which are finally required to be selected in the selected road section data is set as NtThen set the selected road section set VsAnd non-selected road section set VnDividing all road sections, including:
selecting, namely, selecting the road section with the maximum node degree and putting the road section into the selection set VsIn the method, considering that information redundancy exists between adjacent road sections, the neighbor road section which can be directly reached by the largest road section is temporarily put into a non-selected set VnRepeatedly executing the process until only isolated road sections without neighbors are left;
put back the non-selected set VnThe road sections in the road section set are put back into the whole road section set again;
repeating the selecting and replacing steps;
selecting road section set V at any timesThe number of segments in (1) is NtWhen so, the process ends.
In step 4, setting a matrix P to indicate whether the road section data is missing, selecting a road section corresponding value to be 1, and a non-selected road section corresponding value to be 0, and calculating by the server end from the complete historical data set X to obtain a missing data set X1Namely:
X1=X*P……(3),
the historical traffic data set X is divided into seven days per week, the congestion indexes of a certain road section per week X (X is { one, two, three, four, five, six, day }) per day are added according to the week and then averaged to obtain the historical average value of the congestion indexes of the week X, as shown in FIG. 2, the abscissa represents the time of 24 hours per day, the ordinate represents the congestion indexes, the historical average values of all N road sections per week X are calculated in the same way, and finally the historical average value data M is spliced according to the format of original data to obtain the historical average value data M epsilon RN×P;
Because the congestion condition of the road section can be influenced by the congestion condition of the adjacent road section, the historical change rule of the road section is corrected according to the information of the adjacent road section, and the calculation method comprises the following steps:
defining missing segment completion values consists of two parts, namely:
wherein v isnIndicating missing segments, i.e. segments that need to be filled,representing a section of road vnHistorical mean data of (i.e., v in M)nCorresponding row) { v }gDenotes a section vnAll the road sections selected in the neighborhood are selected,representing the average value of the real value of the neighbor selected road section, wherein the proportional coefficient alpha + beta of the road section historical information and the neighbor real information is 1;
calculating the mean value of the real values of the neighbor selected road sections, namely:
because part of road section data is selectively sent, a certain neighbor road section needing to be completed may also be missing, and the road section { v ] is selected from the neighbor road sections in consideration of calculationgThe proportion occupied by the components is as follows:
wherein, { vcDenotes a section vnAll neighbors of (1), including the selected road segment { v }gAnd missing road section, coefficientTo adjust the value of β;
the beta value is dynamically adjusted according to the number of the surrounding selected neighbor segments to determine the influence of neighbor information on a compensation value, and aiming at different selected segment numbers NtAccording to the completed data set X2Separately calculating to minimize the mean absolute error of the complement and the accurate dataNamely:
in step 5, the road congestion prediction includes:
the network server obtains the optimal number of the sending road sections, determines whether the optimal number of the sending road sections is sent to the terminal equipment or not, and feeds back the optimal number information of the sending road sections to the vehicle or the road side unit;
the vehicle or the road side unit selectively sends the current road data according to the indication of the network server, the network server obtains a missing data set, and completes the data which is not sent to obtain a real-time completed data set;
the network server inputs the completion data set into a deep learning traffic prediction model to predict road congestion;
in step 5, a complementary data set X is used2Inputting a deep learning flow prediction model and testing, defining communication learning efficiency eta as a ratio of a square root of prediction congestion index useful power to transmission data overhead, wherein when the useful power is larger (namely the prediction error is smaller), the transmission data overhead is smaller, the communication learning efficiency is higher, namely:
wherein the content of the first and second substances,a predicted value representing a congestion index is displayed,representing absolute error between true and predicted values, the overhead of transmitting data being the number of transmit segments NtA coefficient u represents a variable cost, which is related to the number of transmission links, a coefficient v represents a fixed cost, and u and v are variable according to a communication system model, so that the number of transmission links that maximizes communication learning efficiency, that is, the optimal number of transmission links:
No=argmax{η(Nt)}……(9);
as shown in fig. 4, the solid line indicates the average absolute error, the dotted line indicates u of 0.1, and v of 80, it can be seen from the figure that the average absolute error decreases with the increase of the number of transmission segments, since the image is a monotonically decreasing curve, the optimal point at which both the average absolute error and the number of transmission segments are as small as possible cannot be determined, and the observation communication learning efficiency increases first and then decreases with the increase of the number of transmission segments, and a maximum value exists, so that the optimal point at which the prediction accuracy can be ensured and the amount of transmission data can be as small as possible can be obtained.
In the above embodiment, the deep learning traffic prediction model employs a combined time-graph convolution network, including a graph convolution network GCN for capturing data spatial correlation and a gated round robin unit GRU for capturing data temporal correlation;
from the complement data set X2∈RN×PAnd the adjacency matrix A ∈ RN×NThe GCN captures the spatial features between nodes by the first-order neighbors of the nodesINIs a matrix of units, and is,is a degree matrix, and in the above embodiment 2-layer GCN is used for captureSpatial dependence, the final result can be expressed as:
wherein the content of the first and second substances,is to pairNormalization is carried out to prevent unstable numerical value, W, in operation0∈RP×HIs a weight matrix from an input layer to a hidden layer, P is the length of a historical time series of congestion indexes, H is the number of hidden units, W1∈RH×TIs a weight matrix from a hidden layer to an output layer, ReLU (-) represents a modified linear unit, σ (-) is a sigmoid nonlinear activation function, f (X)2,A)∈RN×TThe congestion index is predicted and output, and therefore the GCN can extract information of first-order neighbors of each node and capture the dependency of a central road section and adjacent road sections of the central road section in the dimension of space.
The result of GCN is input into GRU to form TGCN network used in the above embodiment, the structure of TGCN network is shown in FIG. 3, where ht-1Is the output at time t-1, htIs the output at time t, rtAnd ztReset gate and update gate respectively representing GRU, and the figure is a volume representation of 2-layer GCN networkThe complementary dataset representing the current time t is combined as an input to the graph convolution, and the output of the graph convolution is then expressed as:
the GCN module is used for capturing the spatial dependency among roads in a traffic network, the GRU module is used for capturing the time-varying dependency of the road traffic congestion index, and the dynamic acquisition of the macroscopic traffic law is comprehensively realized to predict the congestion index of the future road.
After the model training and testing process is completed, the network server side can obtain the optimal number of the transmitting road sections, when the road real-time congestion is predicted, the server indicates whether each vehicle or road side unit transmits the road section data or not according to the optimal number of the transmitting road sections, and the vehicle or road side unit selectively transmits the road section data. And the server completes the received missing data and finally inputs the completed data into the model for congestion index prediction.
The present invention is not limited to the above-described embodiments, which are described in the specification and illustrated only for illustrating the principle of the present invention, but various changes and modifications may be made within the scope of the present invention as claimed without departing from the spirit and scope of the present invention. The scope of the invention is defined by the appended claims.
Claims (9)
1. A method for predicting deep traffic congestion, comprising:
step 1, a network server stores road traffic information, establishes a sample set of road section flow vectors and obtains a historical data set;
step 2, converting the traffic network structure information into a topological graph structure;
step 3, obtaining the degree information of the nodes and selecting selected road section data in the historical data set according to the degree information of the nodes to obtain a missing data set;
step 4, complementing the missing road data to obtain a supplemented data set;
and 5, inputting the completion data set into a deep learning flow prediction model to predict the road congestion.
2. The method according to claim 1, wherein in step 1, the road traffic information comprises traffic network structure information and historical traffic data information.
3. The method of claim 1, wherein the step 2 of converting the traffic network structure information into the topological graph structure comprises: defining a directed graph network G (V, E) according to the traffic roads, and taking the roads as nodes V ═ V in the directed graph network1,v2,…,vNN represents the number of road segments, and the road segment intersection is taken as the edge E of the node in the directed graph network, i.e. { v ═ v }ivj},vivjIndicating a road viCan directly reach the road vj。
4. The method according to claim 3, wherein in step 3, the adjacency matrix A E R of the directed graph network G is obtained by calculation according to the directed graph network G (V, E)N×N,A∈RN×NRepresenting the connection relationship between roads, the matrix contains 1 and 0 elements, wherein 1 represents directly reachable, and 0 represents not directly reachable, i.e. the following formula (1):
calculating the outgoing degree matrix of the directed graph network G by the A, namely the following formula (2):
D=∑jΑij……(2),
obtaining the node degree according to the formula (1) and the formula (2), selecting and selecting the road section data, and setting the number of the finally selected road sections as NtThen set the selected road section set VsAnd non-selected road section set VnDividing all road sections, including:
selecting, namely, selecting the road section with the maximum node degree and putting the road section into the selection set VsIn the method, considering that information redundancy exists between adjacent road sections, the neighbor road section which can be directly reached by the largest road section is temporarily put into a non-selected set VnRepeatedly executing the process until only isolated road sections without neighbors are left;
put back the non-selected set VnThe road sections in the road section set are put back into the whole road section set again;
repeating the selecting and replacing steps;
selecting road section set V at any timesThe number of segments in (1) is NtWhen so, the process ends.
5. The method according to claim 1, wherein in step 4, the matrix P is set to indicate whether the road section data is missing, the corresponding value of the selected road section is 1, the corresponding value of the non-selected road section is 0, and the server calculates the missing data set X from the complete historical data set X to obtain the missing data set X1Namely, the following formula (3):
X1=X*P……(3),
in the above formula (3), the expression matrix is multiplied by a dot, that is, the corresponding position elements are multiplied, the historical traffic data set X is divided into seven days per week, the congestion indexes of a certain road section per week X one day are added by week and then averaged to obtain the historical average value of the congestion indexes of week X, the congestion condition of the road section can be affected by the congestion condition of the neighboring road section, and the historical change rule of the road section is corrected according to the information of the neighboring road section, and the calculation method is as follows:
defining missing segment completion values consists of two parts, namely the following equation (4):
the above formula (4), vnIndicating missing segments, i.e. segments that need to be filled,representing a section of road vnOf { v } vgDenotes a section vnAll the road sections selected in the neighborhood are selected,indicating neighbor selectionTaking the average value of the real values of the road sections, wherein the proportional coefficient alpha + beta of the road section historical information and the neighbor real information is 1;
calculating the mean value of the real values of the neighbor selected road sections, as shown in the following formula (5):
calculating selected road section { v) in neighbor road sectionsgThe ratio of (6):
in the above formula (6), { vcDenotes a section vnAll neighbors of (1), including the selected road segment { v }gAnd missing road section, coefficientTo adjust the value of β;
the beta value is dynamically adjusted according to the number of the surrounding selected neighbor segments to determine the influence of neighbor information on a compensation value, and aiming at different selected segment numbers NtAccording to the completed data set X2Separately calculating to minimize the mean absolute error of the complement and the accurate dataThe following formula (7):
6. the method according to claim 1, wherein in step 5, the complementary data set is input into a deep learning traffic prediction model, the model is trained and a performance test is performed, and the optimal number of transmission segments is obtained according to a test result.
7. The method according to claim 6, wherein the step 5 of predicting the road congestion comprises:
the network server obtains the optimal number of the sending road sections, determines whether the optimal number of the sending road sections is sent to the terminal equipment or not, and feeds back the optimal number information of the sending road sections to the vehicle or the road side unit;
the vehicle or the road side unit selectively sends the current road data according to the indication of the network server, the network server obtains a missing data set, and completes the data which is not sent to obtain a real-time completed data set;
and the network server inputs the completion data set into a deep learning flow prediction model to predict the road congestion.
8. The method according to claim 1, wherein in step 5, the deep learning traffic prediction model comprises a Graph Convolution Network (GCN) for capturing data spatial correlation and a gating cycle unit (GRU) for capturing data temporal correlation.
9. The method according to claim 7, wherein in step 5, a complementary data set X is used2Inputting a deep learning flow prediction model and testing, defining communication learning efficiency eta as a ratio of a root of prediction congestion index useful power to sending data cost, and calculating the communication learning efficiency eta according to the following formula (8):
in the above-mentioned formula (8),a predicted value representing a congestion index is displayed,representing absolute error between true and predicted values, the overhead of transmitting data being the number of transmit segments NtA coefficient u represents a variable cost in relation to the number of transmission links, a coefficient v represents a fixed cost, and u and v are variable according to a communication system model, that is, the number of transmission links for which communication learning efficiency is the maximum is the optimal number, as shown in the following equation (9):
No=argmax{η(Nt)}……(9)。
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110697865.8A CN113469425B (en) | 2021-06-23 | 2021-06-23 | Deep traffic jam prediction method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110697865.8A CN113469425B (en) | 2021-06-23 | 2021-06-23 | Deep traffic jam prediction method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN113469425A true CN113469425A (en) | 2021-10-01 |
CN113469425B CN113469425B (en) | 2024-02-13 |
Family
ID=77869230
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110697865.8A Active CN113469425B (en) | 2021-06-23 | 2021-06-23 | Deep traffic jam prediction method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113469425B (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113936462A (en) * | 2021-10-18 | 2022-01-14 | 广州交信投科技股份有限公司 | Bus road condition prediction method and system based on ASTGCN algorithm |
CN115063054A (en) * | 2022-08-17 | 2022-09-16 | 深圳市索菱实业股份有限公司 | Internet of vehicles big data analysis method and server |
CN115620524A (en) * | 2022-12-15 | 2023-01-17 | 中南大学 | Traffic jam prediction method, system, equipment and storage medium |
Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2007103180A2 (en) * | 2006-03-03 | 2007-09-13 | Inrix, Inc. | Assessing road traffic conditions using data from mobile data sources |
CN104091081A (en) * | 2014-07-15 | 2014-10-08 | 中国科学院自动化研究所 | Traffic data make-up method |
CN109584553A (en) * | 2018-11-29 | 2019-04-05 | 中电海康集团有限公司 | A kind of section degree of association missing complementing method based on space time information |
US20190180612A1 (en) * | 2016-08-15 | 2019-06-13 | University Of Southern California | Latent space model for road networks to predict time-varying traffic |
CN110555018A (en) * | 2019-07-29 | 2019-12-10 | 浙江工业大学 | traffic flow completion and prediction method |
CN111694830A (en) * | 2020-06-12 | 2020-09-22 | 复旦大学 | Missing data completion method based on deep ensemble learning |
CN111724601A (en) * | 2020-06-30 | 2020-09-29 | 西安电子科技大学 | Motor vehicle traffic flow prediction method suitable for data loss |
CN111899510A (en) * | 2020-07-28 | 2020-11-06 | 南京工程学院 | Intelligent traffic system flow short-term prediction method and system based on divergent convolution and GAT |
CN112085947A (en) * | 2020-07-31 | 2020-12-15 | 浙江工业大学 | Traffic jam prediction method based on deep learning and fuzzy clustering |
-
2021
- 2021-06-23 CN CN202110697865.8A patent/CN113469425B/en active Active
Patent Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2007103180A2 (en) * | 2006-03-03 | 2007-09-13 | Inrix, Inc. | Assessing road traffic conditions using data from mobile data sources |
CN104091081A (en) * | 2014-07-15 | 2014-10-08 | 中国科学院自动化研究所 | Traffic data make-up method |
US20190180612A1 (en) * | 2016-08-15 | 2019-06-13 | University Of Southern California | Latent space model for road networks to predict time-varying traffic |
CN109584553A (en) * | 2018-11-29 | 2019-04-05 | 中电海康集团有限公司 | A kind of section degree of association missing complementing method based on space time information |
CN110555018A (en) * | 2019-07-29 | 2019-12-10 | 浙江工业大学 | traffic flow completion and prediction method |
CN111694830A (en) * | 2020-06-12 | 2020-09-22 | 复旦大学 | Missing data completion method based on deep ensemble learning |
CN111724601A (en) * | 2020-06-30 | 2020-09-29 | 西安电子科技大学 | Motor vehicle traffic flow prediction method suitable for data loss |
CN111899510A (en) * | 2020-07-28 | 2020-11-06 | 南京工程学院 | Intelligent traffic system flow short-term prediction method and system based on divergent convolution and GAT |
CN112085947A (en) * | 2020-07-31 | 2020-12-15 | 浙江工业大学 | Traffic jam prediction method based on deep learning and fuzzy clustering |
Non-Patent Citations (4)
Title |
---|
DONG GUAN: "Intelligent Prediction of Mobile Vehicle Trajectory Based on Space-Time Information", IEEE 89TH VEHICULAR TECHNOLOGY CONFERENCE * |
LINGYI HAN: "Short-Term Traffic Prediction Based on DeepCluster in Large-Scale Road Networks", IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, vol. 68, no. 12, XP011762192, DOI: 10.1109/TVT.2019.2947080 * |
于德新;邱实;周户星;王卓睿;: "基于GRU-RNN模型的交叉口短时交通流预测研究", 公路工程, no. 04 * |
刘安琪: "城市实时交通流预测研究及应用", 中国优秀硕士学位论文全文数据库 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113936462A (en) * | 2021-10-18 | 2022-01-14 | 广州交信投科技股份有限公司 | Bus road condition prediction method and system based on ASTGCN algorithm |
CN115063054A (en) * | 2022-08-17 | 2022-09-16 | 深圳市索菱实业股份有限公司 | Internet of vehicles big data analysis method and server |
CN115063054B (en) * | 2022-08-17 | 2022-11-29 | 深圳市索菱实业股份有限公司 | Internet of vehicles big data analysis method and server |
CN115620524A (en) * | 2022-12-15 | 2023-01-17 | 中南大学 | Traffic jam prediction method, system, equipment and storage medium |
Also Published As
Publication number | Publication date |
---|---|
CN113469425B (en) | 2024-02-13 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111161535B (en) | Attention mechanism-based graph neural network traffic flow prediction method and system | |
US11270579B2 (en) | Transportation network speed foreeasting method using deep capsule networks with nested LSTM models | |
CN108197739B (en) | Urban rail transit passenger flow prediction method | |
CN113469425A (en) | Deep traffic jam prediction method | |
CN109285346B (en) | Urban road network traffic state prediction method based on key road sections | |
CN111210633B (en) | Short-term traffic flow prediction method based on deep learning | |
CN110570651B (en) | Road network traffic situation prediction method and system based on deep learning | |
Nie et al. | Network traffic prediction based on deep belief network and spatiotemporal compressive sensing in wireless mesh backbone networks | |
CN111832814A (en) | Air pollutant concentration prediction method based on graph attention machine mechanism | |
CN110517482B (en) | Short-term traffic flow prediction method based on 3D convolutional neural network | |
US20220092418A1 (en) | Training method for air quality prediction model, prediction method and apparatus, device, program, and medium | |
CN112071062B (en) | Driving time estimation method based on graph convolution network and graph attention network | |
CN112598180A (en) | Distributed regional wind power prediction method | |
CN110689183B (en) | Cluster photovoltaic power probability prediction method, system, medium and electronic device | |
CN112037539B (en) | Method and system for recommending signal control scheme for saturated urban traffic network | |
CN112149887A (en) | PM2.5 concentration prediction method based on data space-time characteristics | |
CN115376317B (en) | Traffic flow prediction method based on dynamic graph convolution and time sequence convolution network | |
CN113808396A (en) | Traffic speed prediction method and system based on traffic flow data fusion | |
CN112232543A (en) | Multi-site prediction method based on graph convolution network | |
CN112329997A (en) | Power demand load prediction method and system, electronic device, and storage medium | |
CN114925836A (en) | Urban traffic flow reasoning method based on dynamic multi-view graph neural network | |
CN111341109A (en) | City-level signal recommendation system based on space-time similarity | |
CN115966083A (en) | Distributed traffic flow prediction method and system based on time-space decoupling | |
CN115099328A (en) | Traffic flow prediction method, system, device and storage medium based on countermeasure network | |
CN114884828A (en) | Space-based networking efficiency evaluation system and method based on analytic hierarchy process |
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 |