A disturbance-oriented resiliency-oriented signal control sub-zone division method and system

Abstract

A method and system for resilient guided signal control sub-region partitioning under disturbance conditions is disclosed, belonging to the field of traffic control system technology. This invention addresses the problems of existing technologies, such as lack of resilient response expression, insufficient prior guidance for key nodes, unstable clustering boundaries, and insufficient feasibility in traffic engineering under disturbance scenarios. The method involves constructing a directed graph of the urban road network, setting disturbance scenarios and recovery parameters, and calculating the global efficiency of the urban road network during the disturbance process. Based on the global efficiency, a node-level resilience index is calculated for each node, and resilient key control nodes are identified based on these indices. Necessary connection constraints for the influence domains corresponding to the resilient key control nodes are generated, and a constraint similarity matrix is ​​constructed. Constraint spectrum clustering is performed based on the constraint similarity matrix, and the clustering results are boundary-corrected to generate the signal control sub-region partitioning results. This invention is used to implement resilient guided signal control sub-region partitioning.

Description

Technical Field

[0001] This invention relates to the field of traffic control system technology, and more specifically to a method for dividing resilient steering signal control sub-regions in response to disturbances. Background Technology

[0002] With the expansion of urban road networks, the continuous growth of traffic demand, and the frequent occurrence of emergencies such as extreme weather, traffic accidents, temporary construction, and infrastructure failures, urban road networks may experience problems such as decreased traffic capacity, reduced connectivity efficiency, accelerated congestion propagation, and delayed recovery processes in a short period of time. Traditional traffic signal control usually relies on average flow rate under normal traffic conditions, spatial proximity, or conventional congestion characteristics for regional coordination, which is insufficient to fully characterize the performance degradation, functional maintenance, and recovery processes after disturbances occur.

[0003] Signal control sub-zone division is the foundation for implementing regional coordinated control. Its goal is to divide a large-scale signalized road network into several mutually coordinated control units, so that intersections within the same sub-zone have strong traffic correlation, high control coordination, and relatively consistent operating status. Existing methods mostly use spatial distance, traffic flow correlation, road segment length, turning flow, saturation, phase difference, or graph theory clustering indicators as the division criteria, which have certain application value in routine traffic organization.

[0004] However, under disturbance conditions, the responses of different intersections to the same disturbance exhibit significant heterogeneity. Some intersections experience a large performance degradation in the initial stage of the disturbance but recover quickly, while others experience a smaller degradation but remain inefficient for a long period. Furthermore, some nodes, although having low local traffic volume, are located at critical passages or the boundaries of multiple potential control sub-regions, significantly impacting disturbance propagation and recovery coordination. Dividing intersections solely based on static topology, spatial proximity, or average traffic correlation easily leads to the grouping of intersections with significantly different degradation-recovery behaviors into the same sub-region. Traditional sub-region division methods lack resilient response representation under disturbance scenarios, suffer from insufficient prior guidance for key nodes, unstable cluster boundaries, and inadequate traffic engineering feasibility, resulting in dispersed recovery resource allocation, unstable boundary control, and difficulty in suppressing disturbance propagation.

[0005] Therefore, there is an urgent need to propose a technical solution that can directly transform traffic network resilience theory into signal control sub-region division constraints and similarity measurements, so that the sub-region boundaries not only meet the requirements of spatial continuity and traffic flow coupling, but also serve disturbance isolation, function preservation and recovery coordination, thereby improving the service continuity of urban road networks under disturbance conditions. Summary of the Invention

[0006] This invention addresses the problems of existing technologies, such as lack of resilient response expression under disturbance scenarios, insufficient prior guidance for key nodes, unstable cluster boundaries, and insufficient feasibility in traffic engineering.

[0007] The present invention provides a method for dividing a resilient steering signal control sub-region to handle disturbances, comprising the following steps: Step 1: Construct a directed graph of the urban road network, set the disturbance scenario and recovery parameters, and calculate the global efficiency of the urban road network during the disturbance process. Step 2: Calculate node-level resilience indices for each node based on the overall efficiency of the road network, and identify key resilience control nodes based on the node-level resilience indices. Step 3: Generate the necessary connection constraints of the influence domain corresponding to the toughness key control nodes, and construct a toughness-oriented similarity matrix embedded with the necessary connection constraints as the constraint similarity matrix; Step 4: Perform constrained spectrum clustering based on the constrained similarity matrix, perform boundary correction on the clustering results, and generate signal control sub-region partitioning results.

[0008] Furthermore, in one embodiment of the present invention, in step 2, a node-level resilience index is calculated for each node based on the overall efficiency of the road network. The node-level resilience index includes absorption capacity, resilience loss area, and recovery rate. The absorption capacity is: ; The area of ​​the toughness loss is: ; The recovery rate is: ; in, For absorption capacity, For the area of ​​toughness loss, For recovery rate, This represents the lowest performance after the disturbance. The moment when performance reaches its lowest value. Performance before disturbance. To restore performance to a preset threshold The first moment, The moment when the disturbance begins. For the overall efficiency of the road network, In order to be in The global efficiency value of the road network at time t. For any time, To restore the threshold coefficient.

[0009] Furthermore, in one embodiment of the present invention, step 2, which identifies critical resilience control nodes based on node-level resilience indicators, specifically involves: The entropy weight TOPSIS method is adopted, and the node-level resilience index of each node is used as the evaluation index to calculate the relative proximity of the corresponding nodes. Each node is sorted from high to low relative proximity, and the top-ranked nodes are identified as resilience key control nodes.

[0010] Furthermore, in one embodiment of the present invention, the absorption capacity and recovery rate in the node-level toughness index are normalized according to the benefit-type index, and the toughness loss area in the node-level toughness index is normalized according to the cost-type index.

[0011] Furthermore, in one embodiment of the present invention, the mandatory connection constraints for generating the influence domain corresponding to the toughness key control node in step 3 are specifically as follows: ; in, For nodes With key nodes The pairs of constraint indicator values ​​must be connected. For nodes With key nodes The undirected adjacency indicator value, For nodes With key nodes Spatial distance between them To influence the distance threshold, and They are nodes and key nodes The resilience feature vector, This represents the threshold for the difference in resilience response.

[0012] Furthermore, in one embodiment of the present invention, the construction of a resilience-guided similarity matrix embedding the necessary connection constraints in step 3, as the constraint similarity matrix, specifically involves: Calculate the resilience-guided similarity between different node pairs : ; Resilience-oriented similarity of all node pairs Constructing a resilience-oriented similarity matrix Embedded constraints must be connected. The constraint similarity matrix is ​​obtained as follows: ; in, For nodes With nodes The similarity in resilience between them For nodes With nodes The undirected adjacency indicator value, This is the spatial distance attenuation coefficient. For nodes With nodes Geographical or network distance between them The average distance between adjacent road segments. For the synchronous sensitivity coefficient of toughness response, For nodes The resilience feature vector, To constrain the similarity matrix, For resilience-oriented similarity matrices, For the required connection constraint strength, This is a constraint that must be connected.

[0013] Furthermore, in one embodiment of the present invention, the constrained spectral clustering in step 4 specifically includes: Constructing a degree matrix based on constraint similarity matrices And the Laplace matrix Solving the generalized eigenvalue problem: ; in, For generalized eigenvectors, These are generalized eigenvalues; Before selection The eigenvectors corresponding to the smallest non-zero eigenvalues ​​form a low-dimensional spectral embedding matrix. The low-dimensional spectral embedding matrix is ​​row-normalized, and the cluster centers are initialized using resilient key control nodes. The clustering results are obtained based on the weighted K-means method.

[0014] Furthermore, in one embodiment of the present invention, the boundary correction of the clustering results in step 4 specifically involves: compute nodes With cluster center Traffic flow coupling strength between: ; Determine nodes based on traffic flow coupling strength. The final sub-region assignment is determined, and boundary correction is completed. ; in, For nodes With cluster center The average traffic flow correlation strength between them For nodes To the cluster center The length of the road segment or path between them This refers to phase difference, coordination offset, or its fluctuation. To avoid positive numbers with a denominator of zero, For nodes The final sub-region ownership label, For nodes The row-normalized spectral embedding vector, For the first Cluster centers, These are the weighting coefficients for the traffic flow coupling term.

[0015] Furthermore, in one embodiment of the present invention, the objective function of the method is: ; in, The objective function value, This is the result of the signal control sub-region division. For the number of sub-regions, For the first Sub-districts, and They are different nodes, For nodes With nodes Traffic flow coupling strength between them The penalty coefficient is... This is a penalty item.

[0016] The present invention discloses a resilient steering signal control sub-region partitioning system oriented towards disturbances, comprising the following modules: The calculation module constructs a directed graph of the urban road network, sets disturbance scenarios and recovery parameters, and calculates the global efficiency of the urban road network during the disturbance process. The identification module calculates node-level resilience indicators for each node based on the overall efficiency of the road network, and identifies key resilience control nodes based on the node-level resilience indicators. The module generates the necessary connection constraints for the influence domains corresponding to the key control nodes of resilience, and constructs a resilience-oriented similarity matrix embedded with the necessary connection constraints as the constraint similarity matrix; The partitioning module performs constrained spectrum clustering based on the constrained similarity matrix, corrects the boundaries of the clustering results, and generates signal control sub-region partitioning results.

[0017] This invention addresses the problems of existing technologies, such as lack of resilient response expression under perturbation scenarios, insufficient prior guidance for key nodes, unstable cluster boundaries, and insufficient feasibility in traffic engineering. Specific beneficial effects of this invention include: This invention proposes a resilient guidance signal control sub-region partitioning method for disturbances. Compared with existing sub-region partitioning methods based on static spatial proximity, average traffic correlation, or conventional unsupervised clustering, this invention directly transforms dynamic resilience indicators into sub-region partitioning features. This enables the partitioning results to reflect the differences of intersections during disturbance occurrence, performance degradation, and recovery. It avoids the problems of forcibly classifying nodes with significantly different degradation depths, cumulative losses, and recovery speeds into the same control sub-region, which leads to a lack of resilient response expression, insufficient prior guidance for key nodes, unstable cluster boundaries, and insufficient feasibility of traffic engineering under disturbance scenarios.

[0018] This invention belongs to the technical fields of intelligent transportation systems, urban road traffic engineering, traffic signal coordination and control, and traffic network resilience assessment. Specifically, it relates to a method for dividing urban road network signal control sub-regions for non-frequent disturbance scenarios such as traffic accidents, road capacity reduction, road construction, infrastructure failures, and severe weather. In particular, it relates to a method for generating signal control sub-regions based on dynamic resilience measurement, entropy weight TOPSIS (superior-inferior solution distance method) key control node identification, influence domain constraint spectrum clustering, and traffic flow coupling boundary correction. Attached Figure Description

[0019] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 This is a flowchart of the toughness-guided signal control sub-region division method described in Implementation Method 1; Figure 2 This is a schematic diagram of the disturbance-recovery toughness curve, absorption capacity, toughness loss area, and recovery rate as described in Implementation Method 3; Figure 3 This is a schematic diagram of the process for identifying resilience key control nodes and generating influence domain constraints using the entropy weight TOPSIS described in Implementation Method 3. Figure 4 It is the comprehensive resilience index of the different sub-region partitioning methods described in Implementation Method 3 relative to Traditional SC (traditional spectral clustering method). Diagram showing the improvement process. Detailed Implementation

[0020] Various embodiments of the present invention will now be clearly and completely described with reference to the accompanying drawings. The embodiments described with reference to the drawings are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention.

[0021] Implementation Method 1: Existing sub-region division methods are insufficient in expressing the resilience process of road networks. Traditional divisions typically focus on traffic efficiency or traffic homogeneity under normal conditions, lacking a dynamic description of network performance curves and node-level resilience indicators, and failing to reflect the service continuity contribution of intersections throughout the disturbance-recovery process.

[0022] To address the aforementioned technical problems, this embodiment provides a resilient guided signal control sub-region partitioning method for ensuring the service continuity of urban road networks under disturbances. This method uses the overall efficiency of the road network as a performance function and node-level resilience indices as dynamic features to identify key resilience control nodes. These nodes are then embedded as prior control cores in a constraint spectrum clustering process. Boundary corrections are applied to the clustering results to generate the signal control sub-region partitioning results. For example... Figure 1 As shown, the specific steps include: Step 1: Construct a directed graph of the urban road network, set the disturbance scenario and recovery parameters, and calculate the global efficiency of the urban road network during the disturbance process. First, a disturbance-recovery capacity model is established. In one implementation, the disturbance process is divided into a normal phase, a degradation phase, and a recovery phase. Affected road sections. Capacity during the recovery phase Recover using a linear function: (1) in, For road section The original passage capacity, For the perturbation scene The capacity degradation coefficient below, The moment when performance reaches its lowest value. This is the first moment when performance recovers to a preset threshold. Greater than At that time, the affected road sections were restored to their original traffic capacity.

[0023] In random failure scenarios, affected road segments can be randomly selected from the set of directed road segments according to a given disturbance ratio. In targeted attack scenarios, important road segments can be selected as affected targets based on edge betweenness centrality, traffic load, or critical channel identification results. The same disturbance input can be used to compare the resilience and operational performance of different sub-region partitioning methods.

[0024] Then, dynamic network resilience is measured. For any given time... Adopting global road network efficiency Indicates the overall functional connectivity of the road network under disturbance conditions: (2) in, The total number of nodes. For a moment node To the node The shortest path distance between nodes, if nodes From node If unreachable, then the corresponding The value is 0. Global road network efficiency reflects the overall reachability and service continuity of the road network during capacity degradation, link failures, or recovery processes.

[0025] Step 2: Calculate node-level resilience indices for each node based on the overall efficiency of the road network, and identify key resilience control nodes based on the node-level resilience indices. Step 3: Generate the necessary connection constraints of the influence domain corresponding to the toughness key control nodes, and construct a toughness-oriented similarity matrix embedded with the necessary connection constraints as the constraint similarity matrix; Step 4: Perform constrained spectrum clustering based on the constrained similarity matrix, perform boundary correction on the clustering results, and generate signal control sub-region partitioning results.

[0026] Output sub-zone division results for collaborative control. For intersections within the same sub-zone, implement unified cycle, phase difference optimization, green light ratio allocation, or restore priority configuration; for boundaries between adjacent sub-zones with high traffic flow coupling intensity, implement boundary coordination, channel protection, or restore resource linkage.

[0027] This implementation incorporates node-level resilience indices into the sub-region division calculation, simultaneously considering spatial distance attenuation and resilience response synchronicity during the sub-region division process. This strengthens the aggregation relationship between nodes with similar degradation-recovery behaviors while maintaining geographical continuity, improves resilience homogeneity within sub-regions, and avoids classifying intersections with excessively different resilience behaviors into the same sub-region, effectively enhancing the service continuity of the urban road network under disturbance conditions.

[0028] Implementation Method Two: The difference between this implementation method and Implementation Method One is that the objective of the signal control sub-region division can be expressed as maximizing the resilient guidance similarity and traffic flow coupling strength within the same sub-region while satisfying constraints, and penalizing discontinuous sub-regions and violations of prior constraints. Its objective function is: (3) in, The objective function value, This is the result of the signal control sub-region division. For the number of sub-regions, For the first Sub-districts, and They are different nodes, For nodes With nodes Traffic flow coupling strength between them The penalty coefficient is... This is a penalty term for non-contiguous sub-regions or violations of prior constraints.

[0029] This implementation does not directly solve the combinatorial optimization problem, but transforms the above requirements into a process of similarity matrix construction, Laplace transformation, prior constraint embedding, and cluster boundary correction.

[0030] Implementation Method 3: The difference between this implementation method and Implementation Method 1 is that existing key node identification methods often focus on topological centrality, traffic volume, or empirical judgment, making it difficult to simultaneously integrate a node's ability to maintain functionality under disturbances, loss area, and recovery speed. If key nodes are not embedded as control cores in the sub-region partitioning process, problems may arise such as key corridors being artificially cut, excessively large recovery coordination radii, or key node influence domains being dispersed across regions.

[0031] To address the aforementioned technical problems, this embodiment proposes a method for identifying critical resilience control nodes based on node-level resilience indices. Specifically, it includes the following: Based on the resilience triangle concept, let... Performance before disturbance. This represents the lowest performance after the disturbance. The moment when the disturbance begins. The moment when performance reaches its lowest value. To restore performance to a preset threshold The first moment, then the absorption capacity Area of ​​toughness loss and recovery rate The calculations are as follows: (4) (5) (6) in, In order to be in The global efficiency value of the road network at time _____. For any time, To restore the threshold coefficient. For example... Figure 2 As shown, The larger the value, the stronger the ability of the road network to maintain its original function when disturbances occur; The smaller the value, the smaller the cumulative functional loss caused by the disturbance; The larger the value, the faster the system recovers from its lowest performance state to the preset threshold.

[0032] For node-level resilience metrics, select nodes The incident road segments and their associated incident road segments are considered as affected local components. During the disturbance phase, the capacity of the incident road segments is reduced according to a set capacity degradation rule. The global road network efficiency of the entire network is recalculated at each time step, resulting in the efficiency of the nodes. Triggered network-wide performance trajectory Then, the nodes are calculated according to equations (4) to (6). of , and .

[0033] To facilitate unified comparisons between different methods or different disturbance scenarios, a comprehensive resilience index is established. .because and It belongs to the category of benefit-type indicators. This is a cost-related indicator, so we've made it dimensionless: (7) (8) (9) (10) In the formula, To avoid dividing by zero, the smallest positive number, , and These are the comprehensive evaluation weights for different indicators. In the evaluation of network-level collaborative control effectiveness, all three can be given equal weights; in the ranking of node-level criticality, the entropy weight method is used to objectively determine the weights.

[0034] This implementation method can use average delay, queue length, traffic volume, resilience loss area, absorption capacity, recovery rate, and comprehensive resilience index as operational evaluation indicators to determine whether the division results balance functional loss reduction and traffic operation efficiency improvement under disturbance conditions. Figure 4 The figure shows the improvement of the overall resilience index (CRI) of different sub-region division methods compared to the traditional spectral clustering method.

[0035] like Figure 3 As shown, the entropy weight TOPSIS method is used, with the node-level resilience index of each node as the evaluation index, to calculate the relative proximity of the corresponding nodes. Each node is sorted from high to low relative proximity, and the top-ranked nodes are identified as key resilience control nodes. The absorption capacity and recovery rate in the node-level resilience index are normalized according to the benefit-type index, and the resilience loss area in the node-level resilience index is normalized according to the cost-type index.

[0036] For each node Construct an original decision matrix with absorption capacity, toughness loss conversion value, and recovery rate as indicators. For benefit-type indicators, the following normalization method is used: (11) For the area of ​​toughness loss Cost-related indicators are normalized using the following benefit-based approach: (12) Calculate the first normalized matrix The feature weight of each node under the j-th indicator: (13) Calculate the information entropy of the j-th indicator. and corresponding objective weights : (14) (15) According to weight Constructing a weighted standardization matrix And determine the positive ideal solution respectively. and negative ideal solution .node The Euclidean distances to the positive and negative ideal solutions are: (16) (17) node Relative closeness The calculation is as follows: (18) The larger the value, the more likely it is to be a node. The closer a node is to the ideal resilience response characteristics, the stronger its ability to retain function, the smaller its cumulative functional loss, and the faster its recovery speed. In this implementation, the nodes ranked highest are selected as the set of key resilience control nodes. This is used as the prior control core for subsequent constraint spectral clustering.

[0037] This implementation uses entropy-weighted TOPSIS to objectively integrate node-level absorption capacity, resilience loss area, and recovery rate. It objectively determines the weights based on the dispersion of each indicator data, avoiding reliance on expert experience or single topological centrality to identify key nodes. It can identify control sensitive nodes from the perspective of dynamic traffic function response, providing a basis for priority allocation of recovery resources and protection of key channels.

[0038] Implementation Method Four: The difference between this implementation method and Implementation Method One is that conventional spectral clustering, K-means clustering, fuzzy C-means, and traditional traffic association segmentation methods generally adopt unsupervised clustering frameworks, lacking prior constraints for traffic engineering implementation. Especially in weighted directed road networks, simply relying on adjacency matrices or similarity matrices can lead to inconsistencies between the segmentation results and the actual traffic flow coupling relationship, signal coordination boundaries, and disturbance propagation paths.

[0039] To address the aforementioned technical problems, this embodiment proposes a priori constraint method considering implementation in traffic engineering. Specifically, it includes the following: Construct a resilience-oriented similarity matrix. Since urban road networks are typically directed graphs, first construct the directed adjacency matrix. Perform symmetry processing, if or ,but ,otherwise .node The resilience feature vector is defined as: (19) in, This represents the normalized toughness loss characteristic. Node With nodes The similarity between the resilience-oriented spaces is defined as: (20) According to Equation (20), only spatially connectable intersections with similar degradation-recovery responses have a high degree of similarity.

[0040] The similarity constructed in this implementation includes both spatial distance attenuation and resilience response synchronization, which can strengthen the aggregation relationship between nodes with similar degradation-recovery behaviors while maintaining geographical continuity, thereby improving the resilience homogeneity within sub-regions.

[0041] Construct the necessary connectivity constraints for the influence domain of resilience-critical control nodes. For resilience-critical control nodes... If its adjacent nodes If the physical connectivity, distance threshold, and resilience feature difference threshold are all satisfied simultaneously, then the node will be... Included nodes Influence domain : ;(twenty one) Represents a node With nodes There is a mandatory connection constraint. This constraint can prevent critical control nodes and their recovery influence domains from being unnecessarily split during the clustering process.

[0042] The connection must embed a similarity matrix to obtain a constrained similarity matrix: ;(twenty two) in, For nodes With key nodes The pairs of constraint indicator values ​​must be connected. For nodes With key nodes The undirected adjacency indicator value, For nodes With key nodes Spatial distance between them To influence the distance threshold, and They are nodes and key nodes The resilience feature vector, The threshold for the difference in toughness response. For nodes With nodes The similarity in resilience between them For nodes With nodes The undirected adjacency indicator value, This is the spatial distance attenuation coefficient. For nodes With nodes Geographical or network distance between them The average distance between adjacent road segments. For the synchronous sensitivity coefficient of toughness response, For nodes The resilience feature vector, To constrain the similarity matrix, For resilience-oriented similarity matrices, For the required connection constraint strength, This is a constraint that must be connected.

[0043] Based on the constraint similarity matrix Construct degree matrix And the Laplace matrix : ;(twenty three) ;(twenty four) Solving the generalized eigenvalue problem yields the low-dimensional spectral embedding matrix: (25) in, For generalized eigenvectors, These are generalized eigenvalues; Before selection The eigenvectors corresponding to the smallest non-zero generalized eigenvalues ​​form a low-dimensional spectral embedding matrix. And in accordance with the Ng-Jordan-Weiss criterion (NJW criterion) Unitize each row: (26) in, This is the row-normalized spectral embedding vector. For matrix The OK.

[0044] Weighted K-means clustering is initialized based on resilient critical control nodes. The spectral embedding coordinates corresponding to the resilient critical control nodes are preferentially used as cluster centers to reduce the partitioning fluctuations caused by random initialization and to ensure that sub-regions form around the critical control core.

[0045] This implementation introduces a constraint that the influence domain of key nodes must be connected in spectral clustering, so that key control nodes and their resilient similar neighborhoods are preserved during the partitioning process. This is conducive to forming recovery collaborative units around key nodes and reducing the risk of key corridors being fragmented.

[0046] Implementation Method 5: The difference between this implementation method and Implementation Method 1 is that existing boundary correction processes typically do not simultaneously consider spatial continuity, resilience response synchronization, the influence domain of key nodes, and traffic flow coupling strength. Even if adjacent intersections are spatially close, they should not be simply classified into the same control sub-area if their absorption capacity, recovery speed, or cumulative losses differ significantly. Conversely, nodes located on the same key pathway, even if far apart, may need to maintain coordination during the recovery phase.

[0047] To address the aforementioned technical challenges and ensure that the spectral space partitioning results meet the requirements of traffic engineering, a Whitson-type traffic flow coupling term is introduced. (Node) With cluster center The traffic flow coupling strength between them is defined as: (27) The larger the value, the more likely it is to be a node. The greater the need for traffic transfer and coordination with this control core.

[0048] node The final cluster labels are determined by the following formula: (28) in, For nodes With cluster center The average traffic flow correlation strength between them For nodes To the cluster center The length of the road segment or path between them This refers to phase difference, coordination offset, or its fluctuation. To avoid positive numbers with a denominator of zero, For nodes The final sub-region ownership label, For nodes The row-normalized spectral embedding vector, For the first Cluster centers, These are the weighting coefficients for the traffic flow coupling term. Used to balance the spectral embedding distance and traffic flow coupling strength.

[0049] Iteratively update node assignments and cluster centers until the clustering results no longer change or the change in centers is less than a preset threshold, thus obtaining the signal control sub-region. .

[0050] This implementation uses a Whitson-type traffic flow coupling term to engineer the spectral clustering boundary, taking into account average traffic flow correlation, path length, and phase offset. This helps ensure that the final sub-region is not only mathematically reasonable but also facilitates the implementation of actual signal coordination control.

[0051] In summary, this embodiment provides a resilient steering signal control sub-region partitioning method for ensuring service continuity of urban road networks under disturbances. This method represents the urban road network as a directed graph. ,in, A set of signal-controlled intersections. Let be a set of directed road connections; the signal control sub-region to be divided is represented as Under disturbance scenarios, the partitioning results should simultaneously meet the requirements of spatial continuity, strong traffic flow coupling, consistent resilience response within sub-regions, and complete coverage of the influence domain of key control nodes.

[0052] In an experimental verification, a comparative analysis was conducted using the Sioux Falls baseline network and the Berlin-Tiergarten network. The results show that the method described in this embodiment can reduce the resilience loss area and improve operational indicators such as average delay, queue length, and traffic volume under both random failure and targeted attack scenarios. This indicates that the method is beneficial for improving the disturbance suppression, service continuity, and recovery coordination capabilities of urban road networks.

[0053] Implementation Method Six: This implementation method provides a systematic architecture that can be deployed in traffic signal control platforms, traffic simulation systems, or urban traffic operation monitoring platforms. It enables traffic managers to reconstruct signal control sub-zones with recovery and coordination capabilities based on real-time or near-real-time traffic conditions in scenarios such as accidents, capacity reduction, road construction, severe weather, or damage to important routes. It includes the following modules: The calculation module constructs a directed graph of the urban road network, sets disturbance scenarios and recovery parameters, and calculates the global efficiency of the urban road network during the disturbance process. The identification module calculates node-level resilience indicators for each node based on the overall efficiency of the road network, and identifies key resilience control nodes based on the node-level resilience indicators. The module generates the necessary connection constraints for the influence domains corresponding to the key control nodes of resilience, and constructs a resilience-oriented similarity matrix embedded with the necessary connection constraints as the constraint similarity matrix; The partitioning module performs constrained spectrum clustering based on the constrained similarity matrix, corrects the boundaries of the clustering results, and generates signal control sub-region partitioning results.

[0054] Implementation Method Seven: This implementation method is a specific embodiment of the above-described method.

[0055] Example 1 A resilient guide signal control sub-region division method for ensuring service continuity of urban road networks under disturbances.

[0056] This embodiment can be deployed in urban traffic signal control platforms, regional traffic simulation platforms, or traffic operation status monitoring platforms to dynamically form signal control sub-areas before and after non-frequent disturbances.

[0057] In practice, the following process is included: S101, Traffic Network Data Acquisition and Preprocessing. This involves collecting data such as intersection numbers, node coordinates, road segment directions, road segment lengths, traffic capacity, free-flow speeds, number of lanes, traffic demand, real-time traffic flow, queue lengths, signal cycles, phase differences, and green light ratios; and constructing a directed graph. And generate a directed adjacency matrix based on the adjacency relationship. .

[0058] The collected traffic operation data were then processed for outlier removal, time granularity unification, and normalization. For missing data, interpolation by adjacent detectors, historical mean completion, or short-term prediction model completion were used. For indicators with dimensional differences, Min-Max normalization or standardization methods were used to convert them into comparable variables.

[0059] S102, Disturbance Scenario Construction. Based on application requirements, define disturbance types such as random failures, targeted attacks, construction obstruction, accidental closures, or capacity reduction due to extreme weather, and determine the start time of the disturbance. The moment when minimum performance is reached , Restore completion time Capacity reduction coefficient and recovery threshold In random failure scenarios, road segments are selected from the set according to the disturbance ratio. Affected road segments are randomly selected from the data. In the target attack scenario, road segments are ranked according to edge betweenness centrality, traffic load, or critical passage indicators, and the top-ranked road segments are selected as affected road segments. For each disturbance scenario, different partitioning methods are compared using the same disturbance input.

[0060] S103, Calculation of network-level and node-level resilience metrics. First, global efficiency is calculated during the normalization, degradation, and recovery phases. The network level is obtained according to equations (4) to (6). , and Then, for each node Perform a local incident road segment degradation simulation and obtain And calculate node level , and In one implementation, when the node When the capacity of the incident path set decreases, not only the nodes are calculated. Local traffic capacity changes, and the shortest path and global efficiency are recalculated across the entire network to evaluate nodes. Impact on the continuity of services across the entire network.

[0061] S104, Resilience-critical control node identification. This involves identifying all nodes... , and A decision matrix is ​​formed, and benefit-type and cost-type indicators are standardized separately; the objective weights of the indicators are calculated using the entropy weight method; and the relative proximity of each node is calculated using TOPSIS. , and according to Sort by high to low and select the first Each node serves as a critical control node for resilience. In determining... When setting a value, the number of control sub-zones can be preset. The settings include key nodes' proximity to abrupt changes, the amount of traffic control resources, or the scale of the road network. For example, when it is necessary to divide... When selecting a sub-region, priority can be given to selecting no less than [number missing]. A high proximity node was selected as an initial control core candidate.

[0062] S105, Resilience-oriented similarity and influence domain constraint construction. For each pair of adjacent nodes... , Calculate spatial distance Differences in toughness characteristics and similarity For each critical control node The influence domain is determined according to equation (21). and will and The nodes in the configuration are set to mandatory connection constraints. For nodes that are not adjacent or exceed the distance threshold, their basic spatial similarity can be set to 0 to ensure the spatial continuity of the sub-region partitioning results. For adjacent nodes with large differences in resilience characteristics, even if they are geographically close, their similarity is reduced exponentially to avoid the forced aggregation of resilient heterogeneous nodes.

[0063] S106, Solving by constraint spectral clustering. Based on the constraint similarity matrix. Construct degree matrix and Laplace matrix Solving the generalized eigenvalue problem yields Spectral embedding; row normalization of the spectral embedding matrix; initialization using resilience key control nodes as cluster centers, and execution of weighted K-means iteration. During the iteration process, for each node... Simultaneously considering spectral spatial distance and traffic flow coupling terms. If the node If a node has a strong traffic flow correlation with a cluster center, a short path length, and a stable phase difference, then the node is more likely to be assigned to the control sub-region corresponding to that cluster center.

[0064] S107, Sub-region Boundary Correction and Output. The initial partitioning results undergo connectivity checks, isolated node correction, and boundary traffic flow coupling checks. If a boundary node is related to the core of an adjacent sub-region... If the value is significantly higher than the current sub-region, then boundary adjustments will be made while satisfying the necessary connection constraints and spatial continuity.

[0065] The output includes the sub-section number of each intersection, the key control core of each sub-section, the influence domain of key nodes, the sub-section boundary connection relationship, the high-coupling boundary across sub-sections, and the recovery priority of each sub-section. The traffic control system can then configure a unified cycle, coordinated phase difference, green light ratio, and boundary control strategies based on this information.

[0066] Example 2 This embodiment discloses a specific parameterized application process based on embodiment 1.

[0067] Taking an urban road network containing 24 signalized intersections and 76 directed road segments as an example, we obtain the node coordinates, road segment connections, road lengths, road capacity, and traffic demand data for this road network. We set the number of signal control sub-zones to be divided, K, to be 4.

[0068] This embodiment sets up two types of disturbance scenarios.

[0069] The first category is random failure scenarios. In random failure scenarios, a portion of road segments are randomly selected from the set of directed road segments as the disturbed road segments, and the proportion of road segments disturbed is... Take 0.105263 as the capacity reduction coefficient for the disturbed road section. Take 0.82.

[0070] The second category is targeted attack scenarios. In targeted attack scenarios, the edge betweenness number or traffic carrying capacity importance of each road segment is first calculated, and then the top-ranked road segments are selected as the disturbed road segments, with the road segment disturbance ratio... Take 0.105263 as the capacity reduction coefficient for the disturbed road section. Take 1.

[0071] In practice, the following process is included: S201, construct a directed graph according to the road connection relationships. The shortest path distance between nodes before the disturbance is calculated based on the road segment length and capacity.

[0072] S202, in a random failure scenario, the selected disturbed road segment is handled according to... Reduce traffic capacity; in targeted attack scenarios, target selected critical road segments according to... Reduce traffic capacity.

[0073] S203, in to Phased simulation of road network performance degradation, in to The phased simulation linearly restores the road segment capacity and calculates the global efficiency of the road network at each time step. .

[0074] S204, perform disturbance simulations on the associated road segments for each node to obtain the corresponding data for each node. And calculate the values ​​of each node. , and .

[0075] S205, , and Input the entropy-weighted TOPSIS model to obtain the relative proximity of each node. .according to Nodes are sorted from largest to smallest, and the nodes with the highest ranking are selected as resilience critical control nodes. For example, in this embodiment, nodes 9, 2, 1, 14, 23, and 21 can be selected as candidate resilience critical control nodes.

[0076] S206, using the candidate resilience key control nodes as the control core, construct the key node influence domain. If a neighboring node has a direct connection with a key control node, and the spatial distance is less than a preset spatial threshold, and the difference in resilience response feature vectors is less than a preset resilience threshold, then the neighboring node is included in the influence domain of the corresponding key control node.

[0077] S207, based on the influence domain constraint matrix, corrects the resilience-guided space similarity matrix so that key control nodes and their influence domain nodes are preferentially kept in the same signal control sub-region during spectral clustering.

[0078] S208 solves for the eigenvectors corresponding to the Laplacian matrix of the constraint graph to obtain the low-dimensional spectral embedding results, and uses the toughness key control nodes to initialize the cluster centers.

[0079] S209, based on traffic flow coupling strength Sub-region boundaries are modified. For intersections located near the boundaries of two sub-regions, if the traffic flow coupling strength between the intersection and the core node of the adjacent sub-region is higher, it is assigned to the sub-region with stronger traffic flow coupling; if the intersection has higher resilience response consistency with the nodes in the current sub-region, its original sub-region affiliation is maintained.

[0080] S210 outputs the final signal control sub-region division result.

[0081] In this embodiment, the final output includes four signal control sub-zones, each of which includes at least one resilient critical control node or an intersection with a strong influence domain relationship to the resilient critical control node. This division ensures that disturbance-sensitive road sections, critical control nodes, and their adjacent recovery coordination nodes are located within the same control sub-zone as much as possible, thereby facilitating regional signal coordination in the event of accidents, road capacity reduction, or temporary traffic control.

[0082] Example 3 An electronic device includes a processor, a memory, and a computer program stored in the memory and executable by the processor. When the processor executes the computer program, it implements the resilient steering signal control sub-zone division method for ensuring the continuity of urban road network services under disturbances as described in Embodiment 1 or Embodiment 2.

[0083] The above provides a detailed description of the perturbation-oriented resilient guided signal control sub-region partitioning method and system proposed in this invention. Specific examples have been used to illustrate the principles and implementation methods of this invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of this invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this invention. Therefore, the content of this specification should not be construed as a limitation of this invention.

Claims

1. A method for disturbance-oriented resilience-oriented signal control sub-zone division, characterized in that, Includes the following steps: Step 1: Construct a directed graph of the urban road network, set the disturbance scenario and recovery parameters, and calculate the global efficiency of the urban road network during the disturbance process. Step 2: Calculate node-level resilience indices for each node based on the overall efficiency of the road network, and identify key resilience control nodes based on the node-level resilience indices. Step 3: Generate the necessary connection constraints of the influence domain corresponding to the toughness key control nodes, and construct a toughness-oriented similarity matrix embedded with the necessary connection constraints as the constraint similarity matrix; Step 4: Perform constrained spectrum clustering based on the constrained similarity matrix, perform boundary correction on the clustering results, and generate signal control sub-region partitioning results.

2. The perturbation-oriented, resilience-oriented signal control sub-zone division method according to claim 1, characterized in that, In step 2, based on the overall efficiency of the road network, a node-level resilience index is calculated for each node. This node-level resilience index includes absorption capacity, resilience loss area, and recovery rate. The absorption capacity is: ； The area of ​​the toughness loss is: ； The recovery rate is: ； in, For absorption capacity, For the area of ​​toughness loss, For recovery rate, This represents the lowest performance after the disturbance. The moment when performance reaches its lowest value. Performance before disturbance. To restore performance to a preset threshold The first moment, The start time of the disturbance. For the overall efficiency of the road network, In order to be in The global efficiency value of the road network at time t. For any time, To restore the threshold coefficient.

3. The method for dividing a resilient steering signal control sub-region for disturbance-oriented applications according to claim 1, characterized in that, Step 2, which identifies critical resilience control nodes based on node-level resilience indicators, specifically involves: The entropy weight TOPSIS method is adopted, and the node-level resilience index of each node is used as the evaluation index to calculate the relative proximity of the corresponding nodes. Each node is sorted from high to low relative proximity, and the top-ranked nodes are identified as resilience key control nodes.

4. The method for dividing a resilient steering signal control sub-region for disturbance-oriented applications according to claim 3, characterized in that, The absorption capacity and recovery rate in the node-level toughness indices are normalized using benefit-type indices, and the toughness loss area in the node-level toughness indices is normalized using cost-type indices.

5. The method for dividing a resilient steering signal control sub-region for disturbance-oriented applications according to claim 1, characterized in that, In step 3, the necessary connection constraints for generating the influence domain corresponding to the toughness key control nodes are specifically as follows: ； in, For nodes With key nodes The pairs of constraint indicator values ​​must be connected. For nodes With key nodes The undirected adjacency indicator value, For nodes With key nodes Spatial distance between them To influence the distance threshold, and They are nodes and key nodes The resilience feature vector, This represents the threshold for the difference in resilience response.

6. The method for dividing a resilient steering signal control sub-region for disturbance-oriented applications according to claim 1, characterized in that, In step 3, a resilience-guided similarity matrix is ​​constructed that embeds the constraints that must be connected, serving as the constraint similarity matrix. Specifically: Calculate the resilience-guided similarity between different node pairs : ； Resilience-oriented similarity of all node pairs Constructing a resilience-oriented similarity matrix Embedded constraints must be connected. The constraint similarity matrix is ​​obtained as follows: ； in, For nodes With nodes The similarity in resilience between them For nodes With nodes The undirected adjacency indicator value, This is the spatial distance attenuation coefficient. For nodes With nodes Geographical or network distance between them The average distance between adjacent road segments. For the synchronous sensitivity coefficient of toughness response, For nodes The resilience feature vector, To constrain the similarity matrix, For resilience-oriented similarity matrices, For the required connection constraint strength, This is a constraint that must be connected.

7. The method for dividing a resilient steering signal control sub-region for disturbance-oriented applications according to claim 1, characterized in that, The constrained spectral clustering in step 4 specifically includes: Constructing a degree matrix based on constraint similarity matrices And the Laplace matrix Solving the generalized eigenvalue problem: ； in, For generalized eigenvectors, These are generalized eigenvalues; Before selection The eigenvectors corresponding to the smallest non-zero eigenvalues ​​form a low-dimensional spectral embedding matrix. The low-dimensional spectral embedding matrix is ​​row-normalized, and the cluster centers are initialized using resilient key control nodes. The clustering results are obtained based on the weighted K-means method.

8. The method for dividing a resilient steering signal control sub-region for disturbance-oriented applications according to claim 1, characterized in that, Step 4 involves boundary correction of the clustering results, specifically as follows: compute nodes With cluster center Traffic flow coupling strength between: ； Determine nodes based on traffic flow coupling strength. The final sub-region assignment is determined, and boundary correction is completed. ； in, For nodes With cluster center The average traffic flow correlation strength between them For nodes To the cluster center The length of the road segment or path between them This refers to phase difference, coordination offset, or its fluctuation. To avoid positive numbers with a denominator of zero, For nodes The final sub-region ownership label, For nodes The row-normalized spectral embedding vector, For the first Cluster centers, These are the weighting coefficients for the traffic flow coupling term.

9. The method for dividing a resilient steering signal control sub-region for disturbance-oriented applications according to claim 1, characterized in that, The objective function of the method is: ； in, The objective function value, This is the result of the signal control sub-region division. For the number of sub-regions, For the first Sub-districts, and They are different nodes, For nodes With nodes Traffic flow coupling strength between them The penalty coefficient is... This is a penalty item.

10. A resilient steering signal control sub-region partitioning system oriented towards disturbances, characterized in that, Includes the following modules: The calculation module constructs a directed graph of the urban road network, sets disturbance scenarios and recovery parameters, and calculates the global efficiency of the urban road network during the disturbance process. The identification module calculates node-level resilience indicators for each node based on the overall efficiency of the road network, and identifies key resilience control nodes based on the node-level resilience indicators. The module generates the necessary connection constraints for the influence domains corresponding to the key control nodes of resilience, and constructs a resilience-oriented similarity matrix embedded with the necessary connection constraints as the constraint similarity matrix; The partitioning module performs constrained spectrum clustering based on the constrained similarity matrix, corrects the boundaries of the clustering results, and generates signal control sub-region partitioning results.

Images