Highway network traffic flow prediction method and system based on multiple forward-backward tracing

Abstract

The application relates to the technical field of intelligent traffic, and provides a highway network traffic flow prediction method and system based on multiple positive and reverse tracing, which comprises the following steps: constructing a road network directed topology model, obtaining vehicle trajectory data and projecting the vehicle trajectory data into the road network directed topology model to generate an edge flow matrix and an edge travel matrix; based on the edge flow matrix and the edge travel matrix, calibrating an average travel time matrix, a positive flow transfer matrix and a reverse tracing contribution matrix; constructing a time sequence alignment kernel function based on the average travel time matrix; based on the positive flow transfer matrix, the reverse tracing contribution matrix and the time sequence alignment kernel function, establishing a physically-oriented prediction model, deducing multi-source upstream traffic flow, and outputting traffic flow in a target period. The application improves the accuracy and interpretability of traffic flow prediction through multiple upstream correlation domains, a positive flow transfer matrix, a reverse tracing contribution matrix and a time sequence alignment function.

Description

Technical Field

[0001] This invention relates to the field of intelligent transportation technology, and in particular to a method and system for predicting traffic flow in highway networks based on multiple positive and negative tracing sources. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] With the continuous growth of highway mileage and the rapid increase in car ownership, highway traffic congestion has become increasingly serious, becoming a prominent bottleneck restricting travel quality. Intelligent transportation systems have received widespread attention as an effective means to alleviate traffic congestion and improve road network efficiency.

[0004] Currently, short-term traffic flow prediction methods for highway networks mainly fall into two categories. One category is based on classical physical models, such as the kinematic wave theory, which has clear physical meaning but suffers from serious idealization assumptions and makes parameter calibration difficult. The other category is based on pure data-driven deep learning methods, which, although achieving high fitting accuracy, are black-box models lacking mechanistic explanations of traffic congestion causes. Moreover, existing data-driven models typically assume synchronous time steps, but in the physical world, traffic flow propagation exhibits significant physical time lags, and due to the heterogeneity of driving behavior, convoy dispersion effects can occur. For example, if the physical time from upstream point A to downstream point B is 10 minutes, and the prediction step is 5 minutes, then the state of point A at time t mainly affects the state of point B at time t+2, not at time t+1. This temporal mismatch causes existing models to break the causal chain when facing long-distance road sections or congestion, resulting in a significant decrease in prediction accuracy. Summary of the Invention

[0005] To overcome the shortcomings of the prior art, this invention provides a method and system for predicting traffic flow in high-speed networks based on multiple positive and negative tracing sources, aiming to solve the problems of low accuracy and poor interpretability of traffic flow prediction caused by spatiotemporal nonstationarity and asynchronous propagation in the prior art.

[0006] To achieve the above objectives, the technical solution provided by the present invention is as follows: In a first aspect, the present invention provides a method for predicting traffic flow in a high-speed road network based on multiple positive and negative source tracing, including: Construct a directed topology model of the road network, acquire vehicle trajectory data and project it onto the directed topology model of the road network to generate an edge flow matrix and an edge travel matrix; Based on the edge flow matrix and edge travel matrix, the average travel time matrix, forward flow transfer matrix, and reverse source tracing contribution matrix are calibrated; a time-series alignment kernel function is constructed based on the average travel time matrix. A physical-guided prediction model is established based on the forward flow transfer matrix, the reverse source tracing contribution matrix, and the time-series alignment kernel function to extrapolate the upstream traffic flow from multiple sources and output the traffic flow for the target time period.

[0007] As a further technical solution, the elements in the average travel time matrix For all vehicles to reach the downstream node Time and vehicle departure from upstream node The average of the sum of the time differences is expressed as: ; in, Indicates the observation period from the upstream node Drive to the downstream node The total number of vehicle trajectory samples; Indicates the first The vehicle leaves the upstream node Time; Indicates the first The vehicle arrives at the downstream node The time.

[0008] As a further technical solution, the elements in the positive flow transfer matrix upstream node Flowing to downstream nodes Traffic and upstream nodes The ratio of total outflow is expressed as: ; in, Indicates upstream node Flowing to downstream nodes Traffic; Indicates upstream node Total outflow.

[0009] As a further technical solution, the elements in the reverse source tracing contribution matrix via upstream node Reaching downstream nodes Weighted traffic and all upstream neighbor nodes Reaching downstream nodes The weighted total flow ratio is expressed as: ; Where K represents the number of nodes from the upstream node within the historical observation period. Drive to the downstream node The total number of vehicle trajectory samples. Indicates the indicator function, when the first... The trajectory passes through the upstream node Then reach the downstream node The value is 1 if the condition is met, and 0 otherwise. Indicates upstream node In the Observed flow rate for each time period; The index variable representing the upstream neighbor node; Indicates downstream node The set of upstream neighbor nodes; Indicates upstream neighbor nodes In the The observed flow rate for each time period.

[0010] As a further technical solution, the time-aligned kernel function constructed based on the average travel time matrix is ​​expressed as follows: ; in, Represents the maximum value function. Indicates upstream node to downstream nodes The average travel time corresponds to the scrolling interval of the sliding window; Indicates the time step to be predicted; This represents the prediction time lag steps, and takes values ​​within the sliding window range. M represents the maximum step size of the sliding window.

[0011] As a further technical solution, a physical guidance prediction model is established based on the forward flow transfer matrix, the reverse source tracing contribution matrix, and the time-series alignment kernel function to extrapolate multi-source upstream traffic flow and output the traffic flow for the target time period, expressed as: ; in, Indicates downstream node In the future Predicted flow for a given time period Indicates downstream node The set of upstream neighbor nodes, Represents the upstream nodes in the reverse source tracing contribution matrix to downstream nodes The element value, Indicates upstream node to downstream nodes Predicting time lag steps The timing alignment kernel function, Indicates the upstream node in the positive flow transfer matrix Flowing to downstream nodes The element value, Indicates upstream node In history Observed flow rate during the time period Indicates a time period index. Indicates the time step to be predicted. This represents the prediction time lag steps, and takes values ​​within the sliding window range. M represents the maximum step size of the sliding window.

[0012] As a further technical solution, it also includes: dynamically adjusting the reverse source tracing contribution matrix by using the weighted average absolute percentage error between the predicted flow and the actual observed flow, wherein the error is expressed as: ; in, Indicates downstream node The later Actual flow rate observed in each time period Indicates downstream node The later Traffic forecast for each time period This indicates taking the absolute value.

[0013] Secondly, this invention provides a high-speed road network traffic flow prediction system based on multiple-positive-negative source tracing, including: Road network construction and data acquisition module: used to construct a directed topology model of the road network, acquire vehicle trajectory data and project it onto the directed topology model of the road network, and generate edge flow matrix and edge travel matrix; Prediction module: Based on the edge flow matrix and edge travel matrix, calibrate the average travel time matrix, forward flow transfer matrix, and reverse source tracing contribution matrix; construct a time-series alignment kernel function based on the average travel time matrix; A physical-guided prediction model is established based on the forward flow transfer matrix, the reverse source tracing contribution matrix, and the time-series alignment kernel function to extrapolate the upstream traffic flow from multiple sources and output the traffic flow for the target time period.

[0014] Thirdly, the present invention provides an electronic device including a processor and a memory, wherein the memory is used to store a computer program, the computer program being loaded and executed by the processor to implement the steps in the high-speed road network traffic flow prediction method based on multiple-positive-negative tracing as described in the first aspect above.

[0015] Fourthly, the present invention provides a computer-readable storage medium for storing a computer program; wherein the computer program, when executed by a processor, implements the steps of the high-speed road network traffic flow prediction method based on multiple-positive-negative tracing as described in the first aspect above.

[0016] As can be seen from the above technical solutions, the present invention has the following beneficial effects: The above-mentioned technical solution of the present invention provides a method and system for predicting traffic flow in highway networks based on multiple forward and reverse source tracing. By using multiple upstream correlation domains, the prediction perspective is extended outward to capture early propagation signs of traffic waves at distant points. The forward flow transfer matrix describes the physical path and diversion probability of traffic flow from upstream to downstream from a cause-and-effect perspective, while the reverse source tracing contribution matrix quantifies the dependence weight of downstream congestion on different upstream sources in a cause-and-effect manner, enhancing the interpretability of the model. The time-series alignment function accurately calculates the nonlinear matching relationship between the model lag step number and the actual physical travel time, effectively restoring the dispersive physical characteristics of traffic flow during propagation within the data-driven model. Thus, the accuracy and interpretability of traffic flow prediction are improved. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 A schematic diagram of the process of a highway network traffic flow prediction method based on multiple positive and negative source tracing provided in an embodiment of the present invention; Figure 2 This is a schematic diagram illustrating the construction of a multi-positive and negative tracing mechanism matrix provided in an embodiment of the present invention; Figure 3 This is a schematic diagram illustrating the working mechanism of the forward flow matrix, reverse source tracing matrix, and time-series alignment function provided in an embodiment of the present invention. Figure 3 In the diagram, 'a' represents the principle of upstream flow lag propagation. Figure 3 In the diagram, b is a schematic diagram of traffic transfer source tracing calculation. Figure 3 In the middle, c is a schematic diagram of the fusion of weighted prediction results from multiple upstream nodes; Figure 4 This is a schematic diagram of model prediction error analysis under the National Day high saturation flow scenario provided in an embodiment of the present invention; Figure 5 A schematic diagram illustrating the comparative analysis of prediction errors of different models in a high-saturation flow scenario provided by an embodiment of the present invention; Figure 6 This is a schematic diagram illustrating the comparative analysis of prediction errors of different module models in a full-day traffic scenario provided by an embodiment of the present invention. Detailed Implementation

[0019] It should be noted that the following detailed description is exemplary and intended to further illustrate the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0020] It should be noted that the terminology used herein is for the purpose of describing particular implementations only and is not intended to limit the exemplary implementations of the present invention.

[0021] Where there is no conflict, the embodiments and features in the embodiments of the present invention can be combined with each other.

[0022] Example 1 like Figure 1 As shown, this embodiment provides a highway network traffic flow prediction method based on multiple-positive-negative source tracing, including: Step 1: Construct a directed topology model of the road network, obtain vehicle trajectory data and project it into the directed topology model of the road network to generate the edge flow matrix and the edge travel matrix; Step 2: Based on the edge flow matrix and edge travel matrix, calibrate the average travel time matrix, forward flow transfer matrix, and reverse source tracing contribution matrix; construct a time-series alignment kernel function based on the average travel time matrix; Step 3: Based on the forward flow transfer matrix, the reverse source tracing contribution matrix, and the time-series alignment kernel function, establish a physical-guided prediction model to extrapolate the upstream traffic flow from multiple sources and output the traffic flow for the target time period.

[0023] The following content will provide a more detailed introduction to the highway network traffic flow prediction method based on multiple positive and negative tracing provided in this embodiment.

[0024] In step 1, a directed topology model of the road network is constructed, vehicle trajectory data is obtained and projected onto the directed topology model of the road network, and edge flow matrix and edge travel matrix are generated.

[0025] Specifically, a directed topology graph G=(V,E) is constructed using highway ETC gantries as nodes and the driving directions between adjacent upstream and downstream gantries as directed edges. Here, V is the set of gantry nodes and E is the set of directed edges. Vehicle trajectory data is mapped to each gantry node in chronological order. The traffic flow through each node and the travel time of vehicles between adjacent nodes are counted in each time window to generate an edge flow matrix and an edge travel matrix.

[0026] In step 2, such as Figure 2 As shown, based on the edge flow matrix and the edge travel matrix, the average travel time matrix, the forward flow transfer matrix, and the reverse source tracing contribution matrix are calibrated; and a time-series alignment kernel function is constructed based on the average travel time matrix.

[0027] In one specific implementation, the elements in the average travel time matrix For all vehicles to reach the downstream node Time and vehicle departure from upstream node The average of the sum of the time differences is expressed as: ; in, Indicates the observation period from the upstream node Drive to the downstream node The total number of vehicle trajectory samples; Indicates the first The vehicle leaves the upstream node Time; Indicates the first The vehicle arrives at the downstream node The time.

[0028] The aforementioned average travel time matrix is ​​used to characterize the physical impedance between road segments and serves as a benchmark for subsequent spatiotemporal alignment.

[0029] In one specific implementation, the elements in the positive flow transfer matrix upstream node Flowing to downstream nodes Traffic and upstream nodes The ratio of total outflow is expressed as: ; in, Indicates upstream node Flowing to downstream nodes Traffic; Indicates upstream node Total outflow.

[0030] Specifically, the positive traffic transfer matrix answers the question of "where did the upstream vehicles go?" and reflects the diversion characteristics of the highway network.

[0031] The aforementioned positive traffic transfer matrix is ​​used to characterize the probability of vehicle route selection.

[0032] In one specific implementation, the elements in the reverse source tracing contribution matrix via upstream node Reaching downstream nodes Weighted traffic and all upstream neighbor nodes Reaching downstream nodes The weighted total flow ratio is expressed as: ; Where K represents the number of nodes from the upstream node within the historical observation period. Drive to the downstream node The total number of vehicle trajectory samples. Indicates the indicator function, when the first... The trajectory passes through the upstream node Then reach the downstream node The value is 1 if the condition is met, and 0 otherwise. Indicates upstream node In the Observed flow rate for each time period; The index variable representing the upstream neighbor node; Indicates downstream node The set of upstream neighbor nodes; Indicates upstream neighbor nodes In the The observed flow rate for each time period.

[0033] Furthermore, for the target downstream node downstream nodes upstream neighbor node set Extending outwards to order K, it is represented as: ; ; in, Indicates downstream node The set of first-order upstream neighbor nodes, Indicates downstream node of Upstream neighbor group, Indicates downstream node of A set of neighbors on the same level.

[0034] Understandably, That is, all nodes that have a directed edge pointing directly to i; That is, it includes all of them. The upstream node of downstream node i can be reached within a step; when When =1, it degenerates into a traditional first-order neighborhood. When the value is ≥2, the predicted field of view expands outward layer by layer to capture the propagation signal of distant traffic waves.

[0035] Specifically, the reverse source tracing contribution matrix answers the question of "where do downstream vehicles mainly come from?" If upstream A diverts 50% of its traffic to downstream B, but if A itself has very little traffic, A's total contribution to B is still very low. Therefore, the final composite result needs to be weighted by the reverse source tracing contribution matrix.

[0036] The aforementioned reverse tracing contribution matrix is ​​used to characterize the influence weight of the source node.

[0037] In one specific implementation, to address the dispersion phenomenon during high-speed driving, this invention constructs a time-aligned kernel function based on the average travel time matrix, expressed as: ; in, Represents the maximum value function. Indicates upstream node to downstream nodes The average travel time corresponds to the scrolling interval of the sliding window; Indicates the time step to be predicted; This represents the prediction time lag steps, and takes values ​​within the sliding window range. M represents the maximum step size of the sliding window.

[0038] Specifically, the time alignment function enables the model to extract upstream features at the correct time lag; for example, if the upstream is 30 minutes away from the downstream, the model will automatically assign the highest weight in the 30-minute time window, rather than at the current time.

[0039] In step 3, such as Figure 3 As shown, a physical-guided prediction model is established based on the forward flow transfer matrix, the reverse source tracing contribution matrix, and the time-series alignment kernel function to extrapolate the upstream traffic flow from multiple sources and output the traffic flow for the target time period.

[0040] In one specific implementation, a physical guidance prediction model is established based on the forward flow transfer matrix, the reverse source tracing contribution matrix, and the time-series alignment kernel function to extrapolate multi-source upstream traffic flow and output the traffic flow for the target time period, expressed as: ; in, Indicates downstream node In the future Predicted flow for a given time period Indicates downstream node The set of upstream neighbor nodes, Represents the upstream nodes in the reverse source tracing contribution matrix to downstream nodes The element value, Indicates upstream node to downstream nodes Predicting time lag steps The timing alignment kernel function, Indicates the upstream node in the positive flow transfer matrix Flowing to downstream nodes The element value, Indicates upstream node In history Observed flow rate during the time period Indicates a time period index. Indicates the time step to be predicted. This represents the prediction time lag steps, and takes values ​​within the sliding window range. M represents the maximum step size of the sliding window.

[0041] In one specific implementation, the method further includes: dynamically adjusting the reverse source tracing contribution matrix using the weighted average absolute percentage error (WAPE) between the predicted flow and the actual observed flow, wherein the error is expressed as: ; in, Indicates downstream node The later Actual flow rate observed in each time period Indicates downstream node The later Traffic forecast for each time period This indicates taking the absolute value.

[0042] Specifically, based on the error The gradient direction contributes to the reverse tracing matrix. Perform backpropagation updates.

[0043] To verify the effectiveness of this invention, tests were conducted on a highway network holiday dataset, and the test results are as follows: Figures 4 to 6 As shown. Specifically, as Figure 4 As shown, the model of this invention performs daily under the high-saturation flow scenario of the National Day holiday from October 1st to 7th. Value; such as Figure 5 As shown, in a congested scenario with a high saturation flow rate > 800 pcu / h, the model of this invention... The value was 0.224, which was 20.6%, 39.3%, and 10.0% lower than that of the historical average method HA (0.282), the differential integrated moving average autoregressive model ARIMA (0.369), and the long short-term memory network LSTM (0.249), respectively. Figure 6As shown, the comparative analysis of prediction errors of models with different modules under the scenario of all-day traffic flow further verified the effectiveness of each module: after removing the positive traffic transfer matrix, the WAPE increased to 0.291 (+4.7%); after removing the reverse source tracing contribution matrix, the WAPE increased to 0.367 (+32.0%); and after removing the time-series alignment kernel function, the WAPE increased to 0.674 (+142.4%), among which the time-series alignment kernel function made the most significant contribution to accuracy. Thanks to the capture of distant traffic fluctuations by multiple upstream correlation domains, the accurate modeling of traffic flow propagation paths and sources by positive and negative source tracing, and the calibration of propagation lag by time-series alignment, the model can capture the propagation of congestion waves several kilometers away in advance, which is not only accurate in prediction but also fast in response.

[0044] Example 2 This embodiment provides a high-speed road network traffic flow prediction system based on multiple-source forward and reverse tracing, including: Road network construction and data acquisition module: used to construct a directed topology model of the road network, acquire vehicle trajectory data and project it onto the directed topology model of the road network, and generate edge flow matrix and edge travel matrix; Prediction module: Based on the edge flow matrix and edge travel matrix, calibrate the average travel time matrix, forward flow transfer matrix, and reverse source tracing contribution matrix; construct a time-series alignment kernel function based on the average travel time matrix; A physical-guided prediction model is established based on the forward flow transfer matrix, the reverse source tracing contribution matrix, and the time-series alignment kernel function to extrapolate the upstream traffic flow from multiple sources and output the traffic flow for the target time period.

[0045] Example 3 This embodiment provides an electronic device, including a processor and a memory, wherein the memory is used to store a computer program, which is loaded and executed by the processor to implement the steps in the high-speed road network traffic flow prediction method based on multiple-positive-negative tracing as described in Embodiment 1.

[0046] It should be understood that in this embodiment, the processor can be a central processing unit (CPU), or it can be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), off-the-shelf programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor, etc.

[0047] Memory may include read-only memory and random access memory, and provides instructions and data to the processor. A portion of memory may also include non-volatile random access memory. For example, memory may also store information about the device type.

[0048] In the implementation process, each step of the above method can be completed by the integrated logic circuits in the processor hardware or by software instructions.

[0049] The method in Embodiment 1 can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor. The software modules can reside in readily available storage media in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, or registers. This storage medium is located in memory; the processor reads information from the memory and, in conjunction with its hardware, completes the steps of the above method. To avoid repetition, a detailed description is not provided here.

[0050] Those skilled in the art will recognize that the units and algorithm steps described in conjunction with the embodiments herein can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0051] Example 4 This embodiment provides a computer-readable storage medium for storing a computer program; wherein the computer program, when executed by a processor, implements the steps in the high-speed road network traffic flow prediction method based on multiple-positive-negative tracing as described in Embodiment 1.

[0052] While the specific embodiments of the present invention have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of the present invention are still within the scope of protection of the present invention.

Claims

1. A method for predicting traffic flow in a high-speed road network based on multiple positive and negative source tracing, characterized in that, include: Construct a directed topology model of the road network, acquire vehicle trajectory data and project it onto the directed topology model of the road network to generate an edge flow matrix and an edge travel matrix; Based on the edge flow matrix and edge travel matrix, the average travel time matrix, forward flow transfer matrix, and reverse source tracing contribution matrix are calibrated; a time-series alignment kernel function is constructed based on the average travel time matrix. A physical-guided prediction model is established based on the forward flow transfer matrix, the reverse source tracing contribution matrix, and the time-series alignment kernel function to extrapolate the upstream traffic flow from multiple sources and output the traffic flow for the target time period.

2. The highway network traffic flow prediction method based on multiple positive and negative source tracing as described in claim 1, characterized in that, Elements in the average travel time matrix For all vehicles to reach the downstream node Time and vehicle departure from upstream node The average of the sum of the time differences is expressed as: ； in, Indicates the observation period from the upstream node Drive to the downstream node The total number of vehicle trajectory samples; Indicates the first The vehicle leaves the upstream node Time; Indicates the first The vehicle arrives at the downstream node The time.

3. The highway network traffic flow prediction method based on multiple positive and negative source tracing as described in claim 1, characterized in that, Elements in the positive flow transfer matrix upstream node Flowing to downstream nodes Traffic and upstream nodes The ratio of total outflow is expressed as: ； in, Indicates upstream node Flowing to downstream nodes Traffic; Indicates upstream node Total outflow.

4. The highway network traffic flow prediction method based on multiple positive and negative source tracing as described in claim 1, characterized in that, Elements in the reverse source tracing contribution matrix via upstream node Reaching downstream nodes Weighted traffic and all upstream neighbor nodes Reaching downstream nodes The weighted total flow ratio is expressed as: ； Where K represents the number of nodes from the upstream node within the historical observation period. Drive to the downstream node The total number of vehicle trajectory samples. Indicates the indicator function, when the first The trajectory passes through the upstream node Then reach the downstream node The value is 1 if the condition is met, and 0 otherwise. Indicates upstream node In the Observed flow rate for each time period; The index variable representing the upstream neighbor node; Indicates downstream node The set of upstream neighbor nodes; Indicates upstream neighbor nodes In the The observed flow rate for each time period.

5. The highway network traffic flow prediction method based on multiple positive and negative source tracing as described in claim 1, characterized in that, The time-aligned kernel function constructed based on the average travel time matrix is ​​expressed as follows: ； in, Represents the maximum value function. Indicates upstream node to downstream nodes The average travel time corresponds to the scrolling interval of the sliding window; Indicates the time step to be predicted; This represents the prediction time lag steps, and takes values ​​within the sliding window range. M represents the maximum step size of the sliding window.

6. The highway network traffic flow prediction method based on multiple positive and negative source tracing as described in claim 1, characterized in that, A physical guidance prediction model is established based on the forward flow transfer matrix, the reverse source tracing contribution matrix, and the time-series alignment kernel function. This model extrapolates multi-source upstream traffic flow and outputs the traffic flow for the target time period, expressed as: ； in, Indicates downstream node In the future Predicted flow for a given time period Indicates downstream node The set of upstream neighbor nodes, Represents the upstream nodes in the reverse source tracing contribution matrix to downstream nodes The element value, Indicates upstream node to downstream nodes Predicting time lag steps The timing alignment kernel function, Indicates the upstream node in the positive flow transfer matrix Flowing to downstream nodes The element value, Indicates upstream node In history Observed flow rate during the time period Indicates time period index, Indicates the time step to be predicted. This represents the prediction time lag steps, and takes values ​​within the sliding window range. M represents the maximum step size of the sliding window.

7. The highway network traffic flow prediction method based on multiple positive and negative source tracing as described in claim 1, characterized in that, Also includes: The reverse source tracing contribution matrix is ​​dynamically adjusted by the weighted average absolute percentage error between the predicted flow and the actual observed flow, whereby the error is expressed as: ； in, Indicates downstream node The later Actual flow rate observed in each time period Indicates downstream node The later Traffic forecast for each time period This indicates taking the absolute value.

8. A high-speed road network traffic flow prediction system based on multiple positive and negative source tracing, characterized in that, include: Road network construction and data acquisition module: used to construct a directed topology model of the road network, acquire vehicle trajectory data and project it onto the directed topology model of the road network, and generate edge flow matrix and edge travel matrix; Prediction module: Based on the edge flow matrix and edge travel matrix, calibrate the average travel time matrix, forward flow transfer matrix, and reverse source tracing contribution matrix; construct a time-series alignment kernel function based on the average travel time matrix; A physical-guided prediction model is established based on the forward flow transfer matrix, the reverse source tracing contribution matrix, and the time-series alignment kernel function to extrapolate the upstream traffic flow from multiple sources and output the traffic flow for the target time period.

9. An electronic device, characterized in that, It includes a processor and a memory, wherein the memory is used to store a computer program, which is loaded and executed by the processor to implement the steps in the high-speed road network traffic flow prediction method based on multiple-positive-negative tracing as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, Used to store computer programs; wherein the computer programs, when executed by a processor, implement the steps in the high-speed road network traffic flow prediction method based on multiple-positive-negative tracing as described in any one of claims 1-7.

Images