A traffic network robustness evaluation method and system based on multi-source data fusion

Abstract

This invention relates to the field of intelligent transportation technology, specifically to a method and system for robustness assessment of traffic networks based on multi-source data fusion. The method involves: obtaining a planar axis map based on acquired research maps and road data, and marking the acquired data information; importing the planar axis map into depthmapX to construct a traffic network topology model and performing spatial syntactic analysis; exporting the traffic network topology model and spatial syntactic analysis results, and integrating real-time data from the acquired data; mapping the traffic network topology model to the road network spatial coordinate system using a spatiotemporal alignment algorithm; constructing a nonlinear coupling model to calculate the robustness weights of each road segment, and optimizing the parameters in the model using a loss function and gradient descent method; combining the optimized robustness weights of each road segment with a disaster probability factor to output a comprehensive assessment result. This invention solves the problems of data isolation and delayed assessment in traditional methods and is applicable to urban road network optimization and emergency management.

Description

Technical Field

[0001] This invention relates to the field of intelligent transportation technology, and in particular to a method and system for evaluating the robustness of traffic networks based on multi-source data fusion. Background Technology

[0002] The stability of road traffic networks is directly related to the normal operation of urban socio-economic activities. Studying the robustness of road traffic networks helps improve the anti-interference ability of urban transportation systems, thereby effectively reducing losses when facing emergencies such as natural disasters and traffic congestion.

[0003] Space syntax explores the importance of a space to people, using data-driven methods to represent subjective human imagery. It analyzes the relationships between spaces using quantitative indicators such as integration and selectivity, thereby describing the impact of architectural and urban spaces on people. Sheng Qiang (2018) conducted an empirical study on cross-sectional passenger flow between subway stations in Beijing, Tianjin, and Chongqing, using integration and selectivity as analytical factors. Yu Yang (2023) analyzed the spatial layout of the Kazanchi Folk Culture Street using road connectivity, road control, road selectivity, and pedestrian simulation. In complex networks, a crucial node is called a key node. In the study of complex networks, scholars have found that these key nodes determine the structure and function of the entire network.

[0004] Research on the robustness of road traffic networks not only helps improve the stability and anti-interference capabilities of urban transportation systems, but also provides a scientific basis for traffic planning and management, thereby promoting the sustainable development of urban socio-economic activities. Traditional road network design typically focuses on optimizing network efficiency and capacity. However, with the acceleration of urbanization, road networks face increasing challenges. Therefore, enhancing the anti-interference capabilities of road networks by improving their robustness has become a major research objective. Complex network theory provides a new perspective on this problem, abstracting road networks into topological structures to study the relationship between their centrality, vulnerability, and robustness.

[0005] Therefore, this invention proposes a method and system for evaluating the robustness of traffic networks based on multi-source data fusion to solve the above problems. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention develops a robustness assessment method and system for traffic road networks based on multi-source data fusion. Through a space-syntax-based assessment model, this invention can provide a more comprehensive understanding of the performance of road networks under disturbances. Furthermore, by combining complex network theory and space-syntax analysis methods, it can provide a deeper understanding of the stability and adaptability of urban road networks in the face of various disturbances.

[0007] On the one hand, the technical solution of this invention to solve the technical problem is a robustness assessment method for traffic network based on multi-source data fusion, which includes the following steps:

[0008] Data acquisition: Acquire research maps and road data, import the acquired data into AutoCAD software for processing to obtain a planar axis diagram, and mark the acquired data information on the planar axis diagram;

[0009] Spatial syntactic analysis: Import the planar axis map into depthmapX to construct a traffic network topology model and perform spatial syntactic analysis, which includes integration degree analysis, connectivity value analysis, and selectivity degree analysis.

[0010] Traffic network topology model mapping: Export the traffic network topology model and spatial syntactic analysis results, and integrate real-time data from the acquired data. Map the traffic network topology model to the road network spatial coordinate system through a spatiotemporal alignment algorithm.

[0011] Model construction and optimization: A nonlinear coupled model is constructed to calculate the robustness weights of each road segment, and the parameters in the model are optimized by using the loss function and gradient descent method;

[0012] Evaluation results output: The robustness weights of each road segment after parameter optimization are combined with the disaster probability factor to output a comprehensive evaluation result.

[0013] In a specific implementation, the data acquisition is as follows:

[0014] The acquired data includes road axes, intersection nodes, road attribute information, real-time data, and historical accident records. The road axes, intersection nodes, and road attribute information are then marked on the planar axis map.

[0015] Among them, road axis, intersection nodes and road attribute information are obtained by parsing publicly available geographic information databases. The road attribute information includes design speed, regional maximum speed limit, road grade and number of lanes.

[0016] Real-time data includes real-time vehicle speed, congestion index, real-time probability of heavy rain and probability of earthquake. Real-time vehicle speed and congestion index are obtained through API access to network maps, while real-time probability of heavy rain and probability of earthquake are obtained through meteorological bureau API.

[0017] Historical accident records are obtained from the government data open platform and include the accident location, recovery time, and scope of impact.

[0018] In a specific implementation, space syntactic analysis is as follows:

[0019] The planar axis diagram is imported into the DepthmapX spatial parsing software for spatial parsing. The analysis process is as follows:

[0020] (1) Construct a traffic network topology model based on the planar axis diagram and marked data, convert the actual roads in the planar axis diagram into line segments, each line segment represents a road segment, make the connection relationship between intersections and road segments consistent with the traffic network topology model, and correct the adjacency relationship according to the actual traffic rules. For complex intersections, perform topology simplification and retain only the main traffic paths.

[0021] (2) Integration analysis: In the DepthmapX spatial syntax analysis software, select the integration analysis module, calculate the topological depth between each road segment based on the shortest path algorithm, and generate a global integration value, which represents the integration degree of the first road segment in the traffic network topology model. The degree of clustering of a line segment in the entire traffic network topology model represents the total number of road segments in the traffic network topology model;

[0022] (3) Connection value analysis: In the DepthmapX spatial syntax analysis software, the number of road segments directly connected to each road segment in the traffic network topology model is counted to obtain the connection value corresponding to each road segment;

[0023] (4) Selection analysis: Select the selection analysis module in the DepthmapX spatial syntax analysis software to count the frequency of each road segment in all shortest paths. This frequency is the selection value corresponding to each road segment. In the entire traffic network topology model, the road segment with a high selection value is the key road segment with high traffic pressure.

[0024] In a specific implementation, the traffic network topology model mapping is as follows:

[0025] By combining a spatiotemporal alignment algorithm with spatial syntactic analysis results, the traffic network topology model is mapped to the road network spatial coordinate system, unifying the data in time and space. The time window of the road network spatiotemporal coordinate system is defined as... Data within a time window is considered to be at the same moment. For high-frequency data in the traffic network topology model, the last frame is taken. For low-frequency data in the traffic network topology model, alignment and interpolation are performed. Then, spatial registration is performed through an implicit neural representation network. The coordinate points of the road network space are verified based on the known calibration points. The alignment results of the spatiotemporal alignment algorithm are trained. The calibration points are the data information marked on the planar axis diagram.

[0026] In a specific implementation, the nonlinear coupling model is as follows:

[0027] The connectivity value of each road segment is calculated using a nonlinear coupling model. S-shaped function gain term, selectivity Power-law correction term and integration degree With vehicle speed The robustness weight of each road segment in the traffic network topology model is calculated using a ratio attenuation term, and the calculation formula is as follows:

[0028] ,

[0029] in, In the traffic network topology model, the first... Robustness weights for each road segment The gain term of the sigmoid function representing the connection value. Indicates the gain steepness coefficient. The power-law correction term represents the degree of selection. Indicates the selectivity intensity coefficient. This represents a non-linear index of selectivity. This indicates the selectivity bias term. This represents the attenuation term in the ratio of integration degree to vehicle speed. Indicates the first Real-time speed of each road segment Indicates the first The maximum speed limit for each road section This represents the overflow prevention constant.

[0030] In a specific implementation, the loss function is as follows:

[0031] The loss function includes a prediction error term and a regularization term, and its calculation formula is as follows:

[0032] ,

[0033] in, The actual robustness score is derived from historical accident data. This represents the regularization coefficient used to suppress parameter overfitting. ;

[0034] The actual robustness score is the accident recovery time obtained by the traffic management department. The calculation is derived by reverse engineering, and the formula is as follows:

[0035] ,

[0036] in, The traffic management department retrieved the first Accident recovery time for each road section, Indicates the first The maximum tolerance time is preset for each road segment.

[0037] In a specific implementation, the gradient descent method is as follows:

[0038] set up The initial parameters are , , express The set, cyclically adjusted The parameters in the formula are calculated until convergence, and the formula is as follows:

[0039] ,

[0040] ,

[0041] in, This represents the absolute difference in loss values ​​between two consecutive iterations. This indicates the updated parameters. This represents the current value that has not been updated. The initial value is , This represents the partial derivative of the loss function with respect to the parameters. This represents the learning rate. The dynamic decay law is as follows , This represents the initial value of the learning rate. Indicates the number of iterations;

[0042] If the loss function does not decrease after a preset number of iterations, training is terminated and the optimal parameters are saved.

[0043] In a specific implementation, the evaluation results are output as follows:

[0044] The disaster probability factor for each road segment is calculated based on the acquired data. The calculation formula is as follows:

[0045] ,

[0046] in, Indicates the first The probability of heavy rain occurring on each road section Indicates the first The probability of earthquake occurrence for each road section. Indicates the first Road density of each section Indicates the first Number of lanes in each road segment This indicates the cumulative effect of multiple disasters. This indicates the degree to which road network redundancy mitigates disasters;

[0047] The robustness weight of each road segment is combined with the disaster probability factor to obtain a comprehensive evaluation result. The calculation formula is as follows:

[0048] ,

[0049] in, This represents the disaster weighting coefficient. This indicates the congestion index.

[0050] On the other hand, the present invention also provides a traffic network robustness assessment system based on multi-source data fusion, including a module for executing processing instructions for each step in the traffic network robustness assessment method based on multi-source data fusion.

[0051] The effects described in the invention are merely those of the embodiments, and not all the effects of the invention. The above technical solutions have the following advantages or beneficial effects:

[0052] This invention discloses a method and system for evaluating the robustness of traffic networks based on multi-source data fusion. The evaluation model based on space syntax not only considers the topological structure of the road network but also incorporates multiple morphological variables in space syntax, such as integration and selectivity. These variables help to more comprehensively understand the performance of the road network under disturbances. By combining complex network theory and space syntax analysis methods, the traffic network robustness evaluation model based on space syntax can provide a deeper understanding of the stability and adaptability of urban road networks in the face of various disturbances. Through multi-source data fusion and nonlinear coupling models, this invention comprehensively and accurately characterizes the road network state. Combined with gradient descent to quickly optimize parameters, it significantly improves the model training efficiency. The design of the loss function ensures the convergence and generalization ability of the model, while the dynamic adjustment of disaster factors enables real-time identification and emergency response to high-risk areas.

[0053] Therefore, this invention can not only efficiently evaluate the robustness of road networks and optimize existing road network designs, but also provide reliable decision support for traffic management and disaster emergency response, and provide a scientific basis for the planning of future urban transportation systems, thus possessing certain engineering practical value. Attached Figure Description

[0054] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof.

[0055] Figure 1 This is a schematic diagram of the method flow of the present invention.

[0056] Figure 2 The main travel paths are retained after topology simplification.

[0057] Figure 3 This is a schematic diagram of the experimental results of the present invention. Detailed Implementation

[0058] To clearly illustrate the technical features of this solution, the invention will be described in detail below through specific embodiments and in conjunction with the accompanying drawings. The following disclosure provides many different embodiments or examples for implementing different structures of the invention. To simplify the disclosure of the invention, the components and arrangements of specific examples are described below.

[0059] Example 1

[0060] The robustness assessment method for traffic networks based on multi-source data fusion from this invention is used to conduct a robustness assessment of the traffic network within a one-kilometer radius of Xuejiabozi Metro Station in the West Coast New Area of ​​Qingdao City, as detailed below:

[0061] Step 1: Data Acquisition

[0062] The study obtained research maps and road data within a one-kilometer radius of Xuejiabozi Metro Station in the West Coast New Area of ​​Qingdao City. The data obtained included road alignment, intersection nodes, road attribute information, real-time data, and historical accident records.

[0063] Road alignment, intersection nodes, and road attribute information are obtained by parsing publicly available geographic information databases. Road attribute information includes design speed, regional maximum speed limit, road grade, and number of lanes.

[0064] Real-time data includes real-time vehicle speed, congestion index, probability of heavy rain, and probability of earthquake. Real-time vehicle speed and congestion index are obtained through API access to network maps, while real-time probability of heavy rain and probability of earthquake are obtained through meteorological bureau API.

[0065] Historical accident records are obtained from the government data open platform and include the accident location, recovery time, and scope of impact.

[0066] The acquired data is imported into AutoCAD software for processing to obtain a planar axis diagram, and the acquired data information is marked on the planar axis diagram;

[0067] Step 2, Spatial Syntactic Analysis:

[0068] (1) Import the acquired road alignment data into AutoCAD software for processing to obtain a planar axis diagram. Mark the intersection nodes and road attribute information on the planar axis diagram. Construct a traffic network topology model based on the planar axis diagram and the marked data. Convert the actual roads in the planar axis diagram into line segments, with each line segment representing a road segment. Ensure that the connection relationship between intersections and road segments is consistent with the traffic network topology model. Correct the adjacency relationship according to actual traffic rules. For complex intersections, perform topology simplification, retaining only the main traffic paths. The retained main traffic paths are as follows: Figure 2 As shown;

[0069] (2) Integration analysis: In the DepthmapX spatial syntax analysis software, select the integration analysis module, calculate the topological depth between each road segment based on the shortest path algorithm, and generate a global integration value, which represents the integration degree of the first road segment in the traffic network topology model. The degree of clustering of a line segment in the entire traffic network topology model represents the total number of road segments in the traffic network topology model;

[0070] (3) Connection value analysis: In the DepthmapX spatial syntax analysis software, the number of road segments directly connected to each road segment in the traffic network topology model is counted to obtain the connection value corresponding to each road segment;

[0071] (4) Selection analysis: Select the selection analysis module in the DepthmapX spatial syntax analysis software to count the frequency of each road segment in all shortest paths. This frequency is the selection value corresponding to each road segment. In the entire traffic network topology model, the road segment with a high selection value is the key road segment with high traffic pressure.

[0072] Step 3: Mapping the traffic network topology model:

[0073] Export the traffic network topology model generated by spatial syntactic analysis, along with the results of integration, connectivity, and selectivity analysis.

[0074] Connect to real-time data sources and define the time window of the road network spatiotemporal coordinate system. =5 minutes, and the data within the time window are considered to be at the same moment; the last frame data of the time window is extracted for high-frequency data, and a linear interpolation algorithm is used to align the time of low-frequency data;

[0075] Spatial registration of a traffic network topology model and a road network spatial coordinate system is performed using an implicit neural representation network. The implicit neural representation network consists of a multilayer perceptron. The input is the node coordinates of the traffic network topology model. High-frequency spatial features are captured through a sinusoidal activation function and position encoding. The output is the registered coordinates. The network parameters are trained with supervision using the ground truth coordinates of the calibration points. The loss function is the mean squared error. The structure of the implicit neural representation network includes a position encoding module and a multilayer perceptron, which are used to map the topology model nodes to the road network spatial coordinate system. The implicit neural representation network includes the following modules connected in sequence.

[0076] Input module: Receives the node coordinates (x, y) and timestamp t of the traffic network topology model, and normalizes them to the interval [−1, 1].

[0077] Position encoding module: Performs Fourier feature mapping on the input coordinates to generate high-frequency feature vectors. Where p is the input coordinate, L For frequency levels, the specific encoding parameters are:L =10, covering spatial scales from 10 meters to 0.1 meters;

[0078] Multilayer perceptron: It contains 4 hidden layers, each with 256 neurons, and the activation function is a sine function (Sin), which is used to fit the nonlinear relationship of coordinate mapping;

[0079] Output module: The linear layer outputs the registered coordinates, and the loss function is the mean square error (MSE) of the calibration point coordinates.

[0080] The network training process is performed simultaneously with gradient descent, using a dynamic learning rate. according to attenuation;

[0081] Based on the pre-marked calibration points (including intersection nodes and accident locations) in the planar axis diagram, verify the coordinate registration accuracy to ensure that the error is below the preset threshold.

[0082] Using calibration points as training samples, the parameters of the spatiotemporal alignment algorithm are optimized; by iteratively adjusting the weights of the implicit neural representation network, the spatial position deviation of the calibration points after mapping is minimized until the convergence condition is met.

[0083] Output a spatiotemporally aligned traffic network topology model to ensure complete consistency with real-time data in the spatiotemporal coordinate system.

[0084] Step 4: Model Construction and Optimization

[0085] The connectivity value of each road segment is calculated using a nonlinear coupling model. S-shaped function gain term, selectivity Power-law correction term and integration degree With vehicle speed The robustness weight of each road segment in the traffic network topology model is calculated using a ratio attenuation term, and the calculation formula is as follows:

[0086] ,

[0087] in, In the traffic network topology model, the first... Robustness weights for each road segment The gain term of the sigmoid function representing the connection value. Indicates the gain steepness coefficient. The power-law correction term represents the degree of selection. Indicates the selectivity intensity coefficient. This represents a non-linear index of selectivity. This indicates the selectivity bias term. This represents the attenuation term in the ratio of integration degree to vehicle speed. Indicates the first Real-time speed of each road segment Indicates the first The maximum speed limit for each road section This represents the overflow prevention constant.

[0088] The parameters in the model are optimized using a loss function and gradient descent.

[0089] Obtain accident recovery time from traffic management department Calculate the actual robustness score:

[0090] ,

[0091] in, The traffic management department retrieved the first Accident recovery time for each road section, Indicates the first The maximum tolerance time is preset for each road segment.

[0092] The loss function formula is as follows:

[0093] ,

[0094] in, The first one is derived from historical accident data. Actual robustness score for each road segment This represents the regularization coefficient used to suppress parameter overfitting. ;

[0095] The gradient descent method is as follows:

[0096] set up The initial parameters are =0.5, , express The set, cyclically adjusted The parameters in the formula are calculated until convergence, and the formula is as follows:

[0097] ,

[0098] ,

[0099] in, This represents the absolute difference in loss values ​​between two consecutive iterations. 。 This indicates the updated parameters. This represents the current value that has not been updated. The initial value is , This represents the partial derivative of the loss function with respect to the parameters. This represents the learning rate. The dynamic decay law is as follows , Indicates the number of iterations, and sets the initial value of the learning rate. =0.1.

[0100] Step 5: Output of Evaluation Results:

[0101] The disaster probability factor for each road segment is calculated based on the acquired data. The calculation formula is as follows:

[0102] ,

[0103] in, Indicates the first The probability of heavy rain occurring on each road section Indicates the first Earthquake probability for each road segment Indicates the first Road density of each section Indicates the first Number of lanes in each road segment This indicates the cumulative effect of multiple disasters. This indicates the degree to which road network redundancy mitigates disasters;

[0104] The robustness weight of each road segment is combined with the disaster probability factor to obtain a comprehensive evaluation result. The calculation formula is as follows:

[0105] ,

[0106] in, This represents the disaster weighting coefficient. , This indicates the congestion index.

[0107] Finally, the data was connected to each road segment in ArcGIS, and risk nodes at high, medium, and low levels were displayed hierarchically. The road network robustness assessment results are as follows: Figure 3 As shown.

[0108] Although the specific embodiments of the invention have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of the invention. Based on the technical solutions of the invention, various modifications or variations that can be made by those skilled in the art without creative effort are still within the scope of protection of the invention.

Claims

1. A robustness assessment method for traffic network based on multi-source data fusion, characterized in that, Includes the following steps: Data acquisition: Acquire research maps and road data, import the acquired data into AutoCAD software for processing to obtain a planar axis diagram, and mark the acquired data information on the planar axis diagram; Spatial syntactic analysis: Import the planar axis map into depthmapX to construct a traffic network topology model and perform spatial syntactic analysis, which includes integration degree analysis, connectivity value analysis, and selectivity degree analysis. Traffic network topology model mapping: Export the traffic network topology model and spatial syntactic analysis results, and integrate real-time data from the acquired data. Map the traffic network topology model to the road network spatial coordinate system through a spatiotemporal alignment algorithm. Model construction and optimization: A nonlinear coupled model is constructed to calculate the robustness weights of each road segment, and the parameters in the model are optimized by using the loss function and gradient descent method; Evaluation results output: The robustness weights of each road segment after parameter optimization are combined with the disaster probability factor to output a comprehensive evaluation result.

2. The robustness assessment method for traffic network based on multi-source data fusion according to claim 1, characterized in that, The specific data acquisition process is as follows: The acquired data includes road axes, intersection nodes, road attribute information, real-time data, and historical accident records. The road axes, intersection nodes, and road attribute information are then marked on the planar axis map. Among them, road axis, intersection nodes and road attribute information are obtained by parsing publicly available geographic information databases. The road attribute information includes design speed, regional maximum speed limit, road grade and number of lanes. Real-time data includes real-time vehicle speed, congestion index, real-time probability of heavy rain and probability of earthquake. Real-time vehicle speed and congestion index are obtained through API access to network maps, while real-time probability of heavy rain and probability of earthquake are obtained through meteorological bureau API. Historical accident records are obtained from the government data open platform and include the accident location, recovery time, and scope of impact.

3. The robustness assessment method for traffic network based on multi-source data fusion according to claim 2, characterized in that, spatial... The syntactic analysis is as follows: The planar axis diagram is imported into the DepthmapX spatial parsing software for spatial parsing. The analysis process is as follows: (1) Construct a traffic network topology model based on the planar axis diagram and marked data, convert the actual roads in the planar axis diagram into line segments, each line segment represents a road segment, make the connection relationship between intersections and road segments consistent with the traffic network topology model, and correct the adjacency relationship according to the actual traffic rules. For complex intersections, perform topology simplification and retain only the main traffic paths. (2) Integration analysis: Select the integration analysis module in the DepthmapX spatial parsing software, calculate the topological depth between each road segment based on the shortest path algorithm, and generate a global integration value. , In the traffic network topology model, the first... The degree of clustering of line segments in the overall traffic network topology model. , This represents the total number of road segments in the traffic network topology model; (3) Connection value analysis: In the DepthmapX spatial syntax analysis software, the number of road segments directly connected to each road segment in the traffic network topology model is counted to obtain the connection value corresponding to each road segment; (4) Selection analysis: Select the selection analysis module in the DepthmapX spatial syntax analysis software to count the frequency of each road segment in all shortest paths. This frequency is the selection value corresponding to each road segment. In the entire traffic network topology model, the road segment with a high selection value is the key road segment with high traffic pressure.

4. The robustness assessment method for traffic network based on multi-source data fusion according to claim 3, characterized in that, The specific mapping of the traffic network topology model is as follows: By combining a spatiotemporal alignment algorithm with spatial syntactic analysis results, the traffic network topology model is mapped to the road network spatial coordinate system, unifying the data in time and space. The time window of the road network spatiotemporal coordinate system is defined as... Data within a time window is considered to be at the same moment. For high-frequency data in the traffic network topology model, the last frame is taken. For low-frequency data in the traffic network topology model, alignment and interpolation are performed. Then, spatial registration is performed through an implicit neural representation network. The coordinate points of the road network space are verified based on the known calibration points. The alignment results of the spatiotemporal alignment algorithm are trained. The calibration points are the data information marked on the planar axis diagram.

5. The robustness assessment method for traffic network based on multi-source data fusion according to claim 4, characterized in that, The nonlinear coupling model is as follows: The connectivity value of each road segment is calculated using a nonlinear coupling model. S-shaped function gain term, selectivity Power-law correction term and integration degree With vehicle speed The robustness weight of each road segment in the traffic network topology model is calculated using a ratio attenuation term, and the calculation formula is as follows: ， in, In the traffic network topology model, the first... Robustness weights for each road segment The gain term of the sigmoid function representing the connection value. Indicates the gain steepness coefficient. The power-law correction term represents the degree of selection. Indicates the selectivity intensity coefficient. This represents a non-linear index of selectivity. This indicates the selectivity bias term. This represents the attenuation term in the ratio of integration degree to vehicle speed. Indicates the first Real-time speed of each road segment Indicates the first The maximum speed limit for each road section This represents the overflow prevention constant.

6. The method for evaluating the robustness of a traffic network based on multi-source data fusion according to claim 5, characterized in that, The loss function is as follows: The loss function includes a prediction error term and a regularization term, and its calculation formula is as follows: ， in, The actual robustness score is derived from historical accident data. This represents the regularization coefficient used to suppress parameter overfitting. ; The actual robustness score is the accident recovery time obtained by the traffic management department. The calculation is derived by reverse engineering, and the formula is as follows: ， in, The traffic management department retrieved the first Accident recovery time for each road section, Indicates the first The maximum tolerance time is preset for each road segment.

7. The robustness assessment method for traffic network based on multi-source data fusion according to claim 6, characterized in that, The gradient descent method is as follows: set up The initial parameters are , , express The set, cyclically adjusted The parameters in the formula are calculated until convergence, and the formula is as follows: ， ， in, This represents the absolute difference in loss values ​​between two consecutive iterations. This indicates the updated parameters. This represents the current value that has not been updated. The initial value is , This represents the partial derivative of the loss function with respect to the parameters. This represents the learning rate. The dynamic decay law is as follows , This represents the initial value of the learning rate. Indicates the number of iterations; If the loss function does not decrease after a preset number of iterations, training is terminated and the optimal parameters are saved.

8. The robustness assessment method for traffic network based on multi-source data fusion according to claim 7, characterized in that, The specific output of the evaluation results is as follows: The disaster probability factor for each road segment is calculated based on the acquired data. The calculation formula is as follows: ， in, Indicates the first The probability of heavy rain occurring on each road section Indicates the first The probability of earthquake occurrence for each road section. Indicates the first Road density of each section Indicates the first Number of lanes in each road segment This indicates the cumulative effect of multiple disasters. This indicates the degree to which road network redundancy mitigates disasters; The robustness weight of each road segment is combined with the disaster probability factor to obtain a comprehensive evaluation result. The calculation formula is as follows: ， in, This represents the disaster weighting coefficient. This indicates the congestion index.

9. A robustness evaluation system for traffic networks based on multi-source data fusion, characterized in that: Includes those used to execute the claims The module for processing instructions in each step of the traffic network robustness assessment method based on multi-source data fusion as described in any one of the claims.

Images