A Traffic Diversion Method for Road Networks Based on Dynamic Topology Reconstruction

Abstract

This invention discloses a traffic diversion method for road networks based on dynamic topology reconstruction. The method identifies abnormal events and determines their impact range, accurately locating the abnormal event's position and the set of affected nodes. By dynamically updating node representations, event characteristics are propagated to nodes within the affected range, with dynamic attenuation based on the distance between nodes and core nodes and the severity of the event. Furthermore, by reducing the connection weights of affected areas and enhancing the connection weights of detour paths, a dynamic adjacency matrix is ​​reconstructed. Finally, based on the reconstructed adjacency matrix and combined with a path planning algorithm, the traffic efficiency and congestion level of each path are comprehensively evaluated to generate the optimal diversion path and allocate traffic proportionally. This invention, through dynamic network reconstruction and intelligent path planning, effectively alleviates traffic congestion caused by abnormal events and improves the emergency response capability and overall traffic efficiency of the transportation network.

Description

Technical Field

[0001] This invention belongs to the field of intelligent traffic management technology, specifically relating to a traffic diversion method for road networks based on dynamic topology reconstruction. Background Technology

[0002] With the acceleration of urbanization and the continuous growth of motor vehicle ownership, the operational pressure on urban transportation systems has increased significantly, especially when emergencies (such as traffic accidents, road construction and natural disasters) occur. These events often have a serious impact on the capacity and stability of the transportation network. According to the research in the literature [1], abnormal events are the main reason for the reduction of transportation network efficiency. The main mechanisms of their impact are the decline in local capacity, redistribution of traffic flow and chain congestion effects.

[0003] Under the influence of abnormal events, the traffic network faces two major problems. First, the capacity of local nodes decreases. The capacity of a road is closely related to the speed and density of traffic flow, and a sudden event can cause the speed of traffic on a local road segment to decrease significantly or even be completely interrupted[2]. This phenomenon weakens the connectivity of the traffic network, causing the original route planning to fail, and further leading to a decrease in the efficiency of network operation. Second, traffic flow is redistributed and congestion spreads. After an abnormal event occurs, vehicles usually choose alternative routes by taking detours on their own or under the guidance of traffic management departments. However, this redistribution of traffic flow is often uneven and can easily put additional pressure on surrounding road segments, forming new congestion points, and thus exacerbating the instability of the traffic system[3]. This kind of diffusion effect has nonlinear characteristics and is very likely to cause local congestion to spread rapidly to the entire traffic network.

[0004] To address these issues, existing traffic diversion methods have significant technical limitations, which manifest in two aspects. First, they lack adaptability to dynamic network changes. Most existing methods rely on fixed-weight path planning models, such as the classic Dijkstra algorithm [4] and A* algorithm [5]. These methods assume that the topology of the traffic network remains unchanged during the event, making it difficult to cope with complex scenarios of dynamic changes in road segment capacity, resulting in a lack of flexibility and accuracy in diversion schemes. Second, they cannot fully capture the spread effect of accidents. Although some methods introduce real-time updates, they usually only make simple adjustments to local road segments and fail to systematically re-evaluate the topology of the entire traffic network. For example, the shortest path optimization algorithm proposed in reference [6] only considers the core area of ​​the accident and ignores the spread effect of peripheral road segments, resulting in the inability to effectively alleviate secondary congestion problems.

[0005] Given the significant shortcomings of existing technologies, there is an urgent need for a novel traffic diversion method that can flexibly adapt to dynamic network changes and comprehensively capture the spillover effects of accidents. This invention addresses this need by proposing a road network traffic diversion method based on dynamic topology reconstruction. This method first scientifically reflects changes in the connection strength of affected nodes and road segments through dynamic adjacency matrix updates, while simultaneously enhancing the connection weights between potential detour paths and unaffected nodes, thereby achieving adaptive adjustment of the network topology. Compared to traditional fixed topology methods, this invention can more accurately capture dynamic changes in the traffic network, significantly improving the flexibility and reliability of diversion strategies. Furthermore, by comprehensively quantifying the accident impact area through data-driven approaches, it provides data support for topology reconstruction, not only more accurately reflecting the actual impact of accidents but also laying a scientific foundation for dynamic diversion path planning. In addition, based on the updated dynamic topology, combined with optimal path algorithms and network capacity constraints, a scientifically sound diversion scheme is generated. Compared to existing shortest path algorithms, this invention fully considers the dynamic characteristics of traffic flow and network capacity limitations, effectively avoiding the generation of new congestion points while maximizing the overall operational efficiency of the traffic system. In summary, this invention innovatively combines dynamic topology reconstruction with data-driven accurate evaluation, overcoming the shortcomings of existing methods in terms of adaptability and accuracy. By dynamically updating the adjacency matrix and implementing efficient path planning, this invention can provide traffic management departments with flexible and accurate diversion strategies, effectively addressing congestion problems caused by abnormal events and improving the adaptability and emergency response efficiency of road networks.

[0006] [1]Smith J, et al.Impact of Incidents on Urban TrafficFlow.Transportation Research Part A, 2020.

[0007] [2] Zhao X, et al. Modeling Road Capacity under DisruptiveEvents. Journal of Traffic Engineering, 2021.

[0008] [3]Liu Y, et al.Traffic Flow Redistribution Mechanisms.IEEE Transactions on Intelligent Transportation Systems, 2019.

[0009] [4]Dijkstra EA Note on Two Problems in Connexion withGraphs.Numerische Mathematik, 1959.

[0010] [5]Hart PE, et al. The A*Algorithm for Shortest PathSearch. Communications of the ACM, 1968.

[0011] [6]Wang Y, et al. Optimization of Traffic Routing under Disruption. Journal of Applied Transportation Research, 2022. Summary of the Invention

[0012] The purpose of this invention is to provide a traffic diversion method for road networks based on dynamic topology reconstruction, aiming to address the problems of decreased capacity of road network nodes and redistribution of traffic caused by abnormal events. It effectively solves the problems of static diversion strategies, inaccurate impact assessment, and low diversion efficiency in existing technologies, providing traffic management departments with flexible and precise diversion solutions, and significantly improving the adaptive capability and emergency response efficiency of road networks.

[0013] The specific plan is as follows:

[0014] A traffic diversion method for road networks based on dynamic topology reconstruction first identifies the location and impact range of abnormal events, dynamically updates the feature representations of network nodes, and adjusts node features based on geographical distance, topological relationships, and event severity. Then, it dynamically updates the network adjacency matrix, reducing the connection weights between nodes in affected areas and enhancing the connection weights of potential detour paths to reflect changes in capacity and traffic redistribution demands caused by the event. Finally, it generates a set of candidate diversion paths through a path planning algorithm, comprehensively evaluates path capacity utilization, travel time, and path cost, selects the optimal diversion path, and allocates traffic proportionally, dynamically adjusting path weights to form an efficient diversion strategy. This achieves scientific redistribution of traffic flow and improvement of network efficiency under the influence of abnormal events. Specifically, it includes the following steps:

[0015] Step S1: Identification of Abnormal Events and Determination of Impact Scope

[0016] When an anomaly occurs in a traffic network, timely and accurate identification of the event's location and impact range is fundamental to developing diversion strategies. The main objective of this step is to dynamically determine the specific location of the event, identify the affected road segments and their node sets, and quantify the event's impact on the traffic network through traffic data analysis. This process provides data support for subsequent topology reconstruction and diversion path optimization. By accurately assessing the spatial extent and capacity changes of the anomaly, the adjacency matrix elements in the traffic network can be scientifically adjusted to reflect the event's disturbance and impact on the overall road network structure. Specifically, it includes the following sub-steps:

[0017] Sub-step S11: Traffic data input and representation

[0018] The input data for this method mainly includes three parts: a weighted traffic network map, historical traffic data, and abnormal event data.

[0019] The weighted graph structure of a traffic network is defined as G = (V, E, A). Here, V is the set of nodes, representing key nodes in the road network; E is the set of edges, representing road segments between nodes; and A is the adjacency matrix, describing the static connection relationships between nodes.

[0020]

[0021] Among them W ij For edge e ij The weight represents the throughput capacity.

[0022] Historical traffic data Including each edge e ij Historical traffic (Unit: Number of vehicles / Time unit), Historical speed (Unit: km / h), Road segment occupancy rate (Reflects vehicle density)

[0023] Abnormal event data is defined as a quadruple E = {L, S, T, D}, where L represents the location of the event, S represents the severity of the event, which reflects the percentage decrease in traffic capacity, T represents the time of the event, and D represents the duration of the event.

[0024] Sub-step S12: Determining the location of the abnormal event

[0025] Abnormal events are divided into explicit input events and potential abnormal events affected by input events (which require implicit identification).

[0026] For events that display input, simply use event e. k Position l k Mapped to network node v i ∈V:

[0027]

[0028] Where d(l) k ,v i ) indicates the event location l k to node v i The standardized geographical distance, cos(θ(l) k ,v i The cosine of the angle between the event propagation direction and the road segment direction is used to measure directional consistency. d and w θ These are the learnable weight coefficients for geographical distance and direction factors, respectively.

[0029] For events requiring implicit identification, the location of potential anomalies is dynamically identified by analyzing the differences between historical traffic data and input flow, specifically including the flow change rate ΔQ. ij Traffic condition degradation rate R i Calculation of (t):

[0030]

[0031]

[0032] where Q ij (t), V ij (t) represents the current flow rate and speed. For historical flow and speed. If |ΔQ ij |Exceeds the threshold and R i If (t) is less than the threshold and the state duration exceeds a certain value, then the road segment is determined to be an abnormal road segment, and the nodes included in the road segment are recorded in the affected node set v. affected middle.

[0033] Sub-step S13: Identification of the scope of impact of abnormal events

[0034] An attention mechanism model is used to quantify the weights of affected nodes and determine the scope of the event's influence. For each pair of nodes (v... i ,v j Attention weight α ij Represents node v j For node v i Importance:

[0035]

[0036] Among them W a F is a trainable weight matrix. i F j Let || be the feature representation of the node, and || be the vector concatenation operation.

[0037] Based on attention weights, the set of affected nodes is further filtered:

[0038] v affected ={v i |α ic >T,d(v i ,v c )≤φ(s k )}

[0039] Wherein, φ(s) k ) is based on the severity of the event. k The mapping function, α ic T represents the attention weights between the events and the nodes that influence them, and is the weight selection threshold.

[0040] Step S2: Dynamically update the node representation

[0041] After an anomaly occurs, to accurately reflect its direct and indirect impacts on traffic network nodes, it is necessary to dynamically update the node feature representations. By propagating event features from core nodes to surrounding nodes, the spatial scope and temporal decay effect of the anomaly can be described.

[0042] Based on step S12, the core node v is obtained. c By directly injecting event characteristics into its representation, the node's sensitivity to abnormal events is enhanced. The node's representation is updated as follows:

[0043]

[0044] in, For node v c The original feature representation, F e As a characteristic of the event, W e Inject weights into learnable event features.

[0045] For nodes affected by the core node, the node representation is decayed based on its distance to the core node and the severity of the event. The node representation is updated as follows:

[0046]

[0047] Where, η i The attenuation coefficient is defined as follows:

[0048]

[0049] Wherein d(v i ,v c ) represents node v i To the core node v c Geographical distance, hop(v i,v c ) indicates that in the original road network topology, from v i To the core node v c The number of hops on the shortest path (i.e., topological distance). λ1 and λ2 are learnable decay rates controlling the two distances. t k and d k It indicates the time and duration of the event, reflecting the process of the event's impact decaying over time.

[0050] Step S3: Update the dynamic adjacency matrix

[0051] Based on the updated node representation, the network connectivity needs to be dynamically adjusted to reflect the reduction in traffic capacity and traffic redistribution caused by the event. This includes the following sub-steps:

[0052] Sub-step S31: Weight update of affected road segments

[0053] For the set of nodes v affected by the event affected The connection strength between them needs to be reduced to reflect the weakening of traffic capacity caused by abnormal events. The updated dynamic adjacency matrix... Defined as:

[0054]

[0055] in, This is the original adjacency matrix. α represents the deviation rate between the node's dynamic representation and the original features, while α is a connection strength adjustment parameter that controls the extent of weight weakening.

[0056] Sub-step S32: Potential detour path diversion update

[0057] For nodes other than those affected, their connection weights with unaffected nodes need to be increased to encourage traffic redistribution to alternative paths. The updated dynamic adjacency matrix... Defined as:

[0058]

[0059] in, The original adjacency matrix is ​​given, β is the detour path weight adjustment parameter, and d is the detour path weight adjustment parameter. max To determine the maximum permissible detour distance, d(v) is obtained based on the statistical characteristics of actual traffic patterns in the network. i ,v k () represents the geographical distance between nodes.

[0060] Sub-step S33: Overall road network connectivity adjustment

[0061] Based on the adjacency matrix of the affected region obtained in step S31 above. The adjacency matrix of potential bypass paths obtained from S32 The weighted fusion yields the final dynamic adjacency matrix:

[0062]

[0063] Where γ is a learnable weight fusion coefficient. When γ→1, the weakening of the connection in the affected area is given priority, emphasizing the direct impact of the event on the network. When γ→0, the weight enhancement of the detour path is given priority, promoting the diversion of traffic to the unaffected area.

[0064] Step S4: Generation and optimization of routing paths

[0065] After the dynamic adjacency matrix is ​​updated, generating diversion paths is a crucial step in resolving traffic congestion caused by abnormal events. This step uses a path planning algorithm based on the dynamic adjacency matrix to generate a set of candidate diversion paths. It then comprehensively considers factors such as network capacity, detour time, and traffic efficiency to select the optimal path and allocate traffic accordingly.

[0066] The dynamic adjacency matrix obtained using S3 is As input, combined with a shortest path algorithm (such as Dijkstra's algorithm), a candidate branching path set P = {p1, p2, ..., p...} is generated. k}, where p i ={v1,v2,...,v m}, where k is the number of candidate paths. To ensure the practical feasibility of the paths, the length and detour cost of the candidate paths are limited.

[0067] L(p i )≤ψ1·L direct C(p) i )≤ψ2·C direct,

[0068] Wherein, L(p i ) and L direct For path p i The length of the direct path and the length of the direct path, C(p) i ) and C direct For path p i Travel time and direct route travel time, ψ1 and ψ2 are detour cost coefficients.

[0069] For each candidate path p i ∈P, calculate capacity utilization rate where Q ij , C ij These represent the flow and capacity of each edge in the path, respectively. Travel time. Where T ijLet be the travel time for each edge in the path. Considering the trade-off between capacity utilization and travel time, calculate the path cost C(p). i Then, sort the candidate paths according to their path costs, and select the path with the lowest cost as the optimal path p. opt :

[0070] C(p i )=λ·U(p i )+(1-λ)·T(p i )

[0071]

[0072] Where λ∈[0,1] is the weight balance coefficient between capacity and time.

[0073] To avoid overload on a single path, the optimal path p opt Other candidate paths p i ∈P distributes the flow proportionally:

[0074]

[0075] where Q total This represents the total traffic that needs to be diverted.

[0076] The traffic splitting path and traffic allocation ratio are output as the final traffic splitting strategy, and the final output is the optimal path p. opt Path traffic allocation ratio Q pi It provides diversion nodes and guidance information for the implementation of traffic guidance systems. In practice, it can be combined with real-time monitoring of route execution effectiveness, recording data such as traffic flow changes, travel time, and delays. Based on feedback results, route weights are dynamically adjusted to optimize the next round of strategies.

[0077] Through the above technical solutions, the present invention achieves the following beneficial effects:

[0078] The purpose of this invention is to provide a traffic diversion method for road networks based on dynamic topology reconstruction, aiming to address the problems of reduced capacity and traffic redistribution at road network nodes caused by abnormal events (such as traffic accidents). First, through dynamic node representation updates and dynamic adjacency matrix reconstruction, this invention can reflect the direct and indirect impacts of abnormal events on the traffic network in real time. By injecting event features into core nodes and applying distance and time decay processing to indirectly affected nodes, the perturbation impact of events in the spatial and temporal dimensions is accurately quantified, giving the network structure stronger dynamic adaptability and expressive power. Second, by updating the dynamic adjacency matrix, the connection strength between affected nodes is reduced, reflecting the capacity reduction caused by the event, while the weight of detour paths is increased, promoting the redistribution of traffic to unaffected areas. This process effectively balances local congestion and global traffic distribution, alleviating traffic bottlenecks caused by abnormal events and improving the traffic network's efficiency. Finally, by generating and optimizing diversion paths, and combining dynamic adjacency matrices and path planning algorithms, a set of candidate diversion paths is generated. Taking into account multiple dimensions such as network capacity, detour time, and travel cost, the optimal path is selected and traffic is allocated proportionally. This invention provides an efficient and scientific diversion strategy that can dynamically adjust path weights and provide feedback optimization, ensuring the rationality and real-time nature of traffic allocation. In summary, this invention can achieve efficient traffic flow redistribution in the event of abnormal events through accurate impact range identification, dynamic network reconstruction, and intelligent diversion strategies. This significantly improves the emergency response capability, traffic efficiency, and stability of the traffic network, providing practically valuable technical support for intelligent traffic management. Attached Figure Description

[0079] Figure 1 This is a flowchart of a road network traffic diversion method based on dynamic topology reconstruction. Detailed Implementation

[0080] The present invention will be further illustrated below with reference to the accompanying drawings and specific embodiments. It should be understood that the following specific embodiments are for illustrative purposes only and are not intended to limit the scope of the present invention.

[0081] like Figure 1As shown, this invention proposes a traffic diversion method for road networks based on dynamic topology reconstruction. This method dynamically adjusts the connection strength of affected nodes through dynamic adjacency matrix updates, while simultaneously enhancing the connection weights between detour paths and unaffected nodes, thus reflecting the dynamic changes in network topology. Through data-driven impact range assessment, it accurately quantifies the impact area of ​​accidents and changes in traffic capacity, providing a scientific basis for road network topology reconstruction. Through efficient traffic redistribution, combined with optimal path algorithms and network capacity constraints, it generates optimal diversion paths on the dynamic topology structure, achieving scientific traffic allocation. This method effectively solves the problems of static diversion strategies, inaccurate impact range assessment, and low diversion efficiency in existing technologies, providing traffic management departments with flexible and accurate diversion solutions, and significantly improving the adaptive capability and emergency response efficiency of road networks. The process of the traffic diversion method for road networks based on dynamic topology reconstruction is as follows: Figure 1 As shown, the specific steps include:

[0082] Step S1: Identification of Abnormal Events and Determination of Impact Scope

[0083] When an anomaly occurs in a traffic network, timely and accurate identification of the event's location and impact range is fundamental to developing diversion strategies. The main objective of this step is to dynamically determine the specific location of the event, identify the affected road segments and their node sets, and quantify the event's impact on the traffic network through traffic data analysis. This process provides data support for subsequent topology reconstruction and diversion path optimization. By accurately assessing the spatial extent and capacity changes of the anomaly, the adjacency matrix elements in the traffic network can be scientifically adjusted to reflect the event's disturbance and impact on the overall road network structure. Specifically, it includes the following sub-steps:

[0084] Sub-step S11: Traffic data input and representation

[0085] The input data for this method mainly includes three parts: a weighted traffic network map, historical traffic data, and abnormal event data.

[0086] The weighted graph structure of a traffic network is defined as G = (V, E, A). Here, V is the set of nodes, representing key nodes in the road network; E is the set of edges, representing road segments between nodes; and A is the adjacency matrix, describing the static connection relationships between nodes.

[0087]

[0088] Among them W ij For edge e ij The weight represents the throughput capacity.

[0089] Historical traffic data Including each edge e ij Historical traffic (Unit: Number of vehicles / Time unit), Historical speed (Unit: km / h), Road segment occupancy rate (Reflects vehicle density)

[0090] Abnormal event data is defined as a quadruple E = {L, S, T, D}, where L represents the location of the event, S represents the severity of the event, which reflects the percentage decrease in traffic capacity, T represents the time of the event, and D represents the duration of the event.

[0091] Sub-step S12: Determining the location of the abnormal event

[0092] Abnormal events are divided into explicit input events and potential abnormal events affected by input events (which require implicit identification).

[0093] For events that display input, simply use event e. k Position l k Mapped to network node v i ∈V:

[0094]

[0095] Where d(l) k ,v i ) indicates the event location l k to node v i The standardized geographical distance, cos(θ(l) k ,v i The cosine of the angle between the event propagation direction and the road segment direction is used to measure directional consistency. d and w θ These are the learnable weight coefficients for geographical distance and direction factors, respectively.

[0096] For events requiring implicit identification, the location of potential anomalies is dynamically identified by analyzing the differences between historical traffic data and input flow, specifically including the flow change rate ΔQ. ij Traffic condition degradation rate R i Calculation of (t):

[0097]

[0098]

[0099] where Q ij (t), V ij (t) represents the current flow rate and speed. For historical flow and speed. If |ΔQ ij |Exceeds the threshold and R iIf (t) is less than the threshold and the state duration exceeds a certain value, then the road segment is determined to be an abnormal road segment, and the nodes included in the road segment are recorded in the affected node set v. affected middle.

[0100] Sub-step S13: Identification of the scope of impact of abnormal events

[0101] An attention mechanism model is used to quantify the weights of affected nodes and determine the scope of the event's influence. For each pair of nodes (v... i ,v j Attention weight α ij Represents node v j For node v i Importance:

[0102]

[0103] Among them W a F is a trainable weight matrix. i F j Let || be the feature representation of the node, and || be the vector concatenation operation.

[0104] Based on attention weights, the set of affected nodes is further filtered:

[0105] v affected ={v i |α ic >T,d(v i ,v c )≤φ(s k )}

[0106] Wherein, φ(s) k ) is based on the severity of the event. k The mapping function, α ic T represents the attention weights between the events and the nodes that influence them, and is the weight selection threshold.

[0107] Step S2: Dynamically update the node representation

[0108] After an anomaly occurs, to accurately reflect its direct and indirect impacts on traffic network nodes, it is necessary to dynamically update the node feature representations. By propagating event features from core nodes to surrounding nodes, the spatial scope and temporal decay effect of the anomaly can be described.

[0109] Based on step S12, the core node v is obtained. c By directly injecting event characteristics into its representation, the node's sensitivity to abnormal events is enhanced. The node's representation is updated as follows:

[0110]

[0111] in, For node v c The original feature representation, F e As a characteristic of the event, W e Inject weights into learnable event features.

[0112] For nodes affected by the core node, the node representation is decayed based on its distance to the core node and the severity of the event. The node representation is updated as follows:

[0113]

[0114] Where, η i The attenuation coefficient is defined as follows:

[0115]

[0116] Wherein d(v i ,v c ) represents node v i To the core node v c Geographical distance, hop(v i ,v c ) indicates that in the original road network topology, from v i To the core node v c The number of hops on the shortest path (i.e., topological distance). λ1 and λ2 are learnable decay rates controlling the two distances. t k and d k It indicates the time and duration of the event, reflecting the process of the event's impact decaying over time.

[0117] Step S3: Update the dynamic adjacency matrix

[0118] Based on the updated node representation, the network connectivity needs to be dynamically adjusted to reflect the reduction in traffic capacity and traffic redistribution caused by the event. This includes the following sub-steps:

[0119] Sub-step S31: Weight update of affected road segments

[0120] For the set of nodes v affected by the event affected The connection strength between them needs to be reduced to reflect the weakening of traffic capacity caused by abnormal events. The updated dynamic adjacency matrix... Defined as:

[0121]

[0122] in, This is the original adjacency matrix. α represents the deviation rate between the node's dynamic representation and the original features, while α is a connection strength adjustment parameter that controls the extent of weight weakening.

[0123] Sub-step S32: Potential detour path diversion update

[0124] For nodes other than those affected, their connection weights with unaffected nodes need to be increased to encourage traffic redistribution to alternative paths. The updated dynamic adjacency matrix... Defined as:

[0125]

[0126] in, The original adjacency matrix is ​​given, β is the detour path weight adjustment parameter, and d is the detour path weight adjustment parameter. max To determine the maximum permissible detour distance, d(v) is obtained based on the statistical characteristics of actual traffic patterns in the network. i ,v k () represents the geographical distance between nodes.

[0127] Sub-step S33: Overall road network connectivity adjustment

[0128] Based on the adjacency matrix of the affected region obtained in step S31 above. The adjacency matrix of potential bypass paths obtained from S32 The weighted fusion yields the final dynamic adjacency matrix:

[0129]

[0130] Where γ is a learnable weight fusion coefficient. When γ→1, the weakening of the connection in the affected area is given priority, emphasizing the direct impact of the event on the network. When γ→0, the weight enhancement of the detour path is given priority, promoting the diversion of traffic to the unaffected area.

[0131] Step S4: Generation and optimization of routing paths

[0132] After the dynamic adjacency matrix is ​​updated, generating diversion paths is a crucial step in resolving traffic congestion caused by abnormal events. This step uses a path planning algorithm based on the dynamic adjacency matrix to generate a set of candidate diversion paths. It then comprehensively considers factors such as network capacity, detour time, and traffic efficiency to select the optimal path and allocate traffic accordingly.

[0133] The dynamic adjacency matrix obtained using S3 is As input, combined with a shortest path algorithm (such as Dijkstra's algorithm), a candidate branching path set P = {p1, p2, ..., p...} is generated. k}, where p i ={v1,v2,...,v m}, where k is the number of candidate paths. To ensure the practical feasibility of the paths, the length and detour cost of the candidate paths are limited.

[0134] L(p i )≤ψ1·L direct C(p) i )≤ψ2·C direct,

[0135] Wherein, L(p i ) and L direct For path p i The length of the direct path and the length of the direct path, C(p) i ) and C direct For path p i Travel time and direct route travel time, ψ1 and ψ2 are detour cost coefficients.

[0136] For each candidate path p i ∈P, calculate capacity utilization rate where Q ij , C ij These represent the flow and capacity of each edge in the path, respectively. Travel time. Where T ij Let be the travel time for each edge in the path. Considering the trade-off between capacity utilization and travel time, calculate the path cost C(p). i Then, sort the candidate paths according to their path costs, and select the path with the lowest cost as the optimal path p. opt :

[0137] C(p i )=λ·U(p i )+(1-λ)·T(p i )

[0138]

[0139] Where λ∈[0,1] is the weight balance coefficient between capacity and time.

[0140] To avoid overload on a single path, the optimal path p opt Other candidate paths p i ∈P distributes the flow proportionally:

[0141]

[0142] where Q total This represents the total traffic that needs to be diverted.

[0143] The traffic splitting path and traffic allocation ratio are output as the final traffic splitting strategy, and the final output is the optimal path p. opt Path traffic allocation ratio Qpi It provides diversion nodes and guidance information for the implementation of traffic guidance systems. In practice, it can be combined with real-time monitoring of route execution effectiveness, recording data such as traffic flow changes, travel time, and delays. Based on feedback results, route weights are dynamically adjusted to optimize the next round of strategies.

Claims

1. A traffic diversion method for road networks based on dynamic topology reconstruction, characterized in that, First, by identifying the location and impact range of abnormal events, the feature representation of network nodes is dynamically updated, and the node features are adjusted based on the geographical distance between nodes, topological relationships, and event severity. Then, based on this, the network adjacency matrix is ​​dynamically updated to reduce the connection weight between nodes in the affected area and enhance the connection weight of potential detour paths, reflecting the changes in traffic capacity and traffic redistribution needs caused by the event. Finally, a set of candidate diversion paths is generated through a path planning algorithm. The path capacity utilization, travel time and path cost are comprehensively evaluated, the optimal diversion path is selected, and traffic is allocated proportionally. The path weights are dynamically adjusted to form an efficient diversion strategy, realizing the scientific redistribution of traffic flow and the improvement of network traffic efficiency under the influence of abnormal events. The method comprises the following steps: Step S1: Identification of abnormal events and determination of their impact scope; Step S2: Dynamically update the node representation; Step S3: Dynamic adjacency matrix update; Step S4: Generation and optimization of routing paths; The dynamic adjacency matrix update described in step S3 is as follows: Step S3: Update the dynamic adjacency matrix Based on the updated node representation, the network connectivity needs to be dynamically adjusted to reflect the reduction in traffic capacity and traffic redistribution caused by the event; this specifically includes the following sub-steps: Sub-step S31: Weight update of affected road segments For the set of nodes v affected by the event affected The connection strength between them needs to be reduced to reflect the weakening of traffic capacity caused by abnormal events; the updated dynamic adjacency matrix Defined as: in, This is the original adjacency matrix. α is the deviation rate between the node's dynamic representation and the original features, and α is the connection strength adjustment parameter that controls the extent of weight weakening; Sub-step S32: Potential detour path diversion update For nodes other than those affected, their connection weights with unaffected nodes need to be increased to encourage traffic to be redistributed to alternative paths; the updated dynamic adjacency matrix Defined as: in, The original adjacency matrix is ​​given, β is the detour path weight adjustment parameter, and d is the detour path weight adjustment parameter. max To determine the maximum permissible detour distance, d(v) is obtained based on the statistical characteristics of actual traffic patterns in the network. i ,v k () represents the geographical distance between nodes; Sub-step S33: Overall road network connectivity adjustment Based on the adjacency matrix of the affected region obtained in step S31 above. The adjacency matrix of potential bypass paths obtained from S32 The weighted fusion yields the final dynamic adjacency matrix: Where γ is a learnable weight fusion coefficient. When γ→1, the weakening of the connection in the affected area is given priority, emphasizing the direct impact of the event on the network. When γ→0, the weight enhancement of the detour path is given priority, promoting the diversion of traffic to the unaffected area. The generation and optimization of the traffic splitting path described in step S4 are as follows: After the dynamic adjacency matrix is ​​updated, the path planning algorithm based on the dynamic adjacency matrix generates a set of candidate routing paths. Taking into account network capacity, detour time and traffic efficiency, the optimal path is selected and traffic is allocated reasonably. The dynamic adjacency matrix obtained using S3 is As input, combined with the shortest path algorithm, a candidate branching path set P = {p1, p2, ..., p...} is generated. k }, where p i ={v1,v2,...,v m }, where k is the number of candidate paths; and to ensure the actual feasibility of the paths, the length and detour cost of the candidate paths are limited: L(p i )≤ψ1·L direct ,C(p i )≤ψ2·C direct, Wherein, L(p i ) and L direct For path p i The length of the direct path and the length of the direct path, C(p) i ) and C direct For path p i Travel time and direct route travel time, where ψ1 and ψ2 are detour cost coefficients; For each candidate path p i ∈P, calculate capacity utilization rate Q ij , C ij These represent the flow and capacity of each edge in the path; and the travel time. Where T ij Let p be the travel time for each edge in the path; considering the trade-off between capacity utilization and travel time, calculate the path cost C(p). i Then, sort the candidate paths according to their path costs, and select the path with the lowest cost as the optimal path p. opt : C(p i )=λ·U(p i )+(1-λ)·T(p i ) Where λ∈[0,1] is the weight balance coefficient between capacity and time; To avoid overload on a single path, the optimal path p opt Other candidate paths p i ∈P distributes the flow proportionally: Q total The total traffic that needs to be diverted; The traffic splitting path and traffic allocation ratio are output as the final traffic splitting strategy, and the final output is the optimal path p. opt Path traffic allocation ratio It provides diversion nodes and guidance information for the implementation of traffic guidance systems; it can combine real-time monitoring of path execution effects, record traffic changes, travel time and delay data, and dynamically adjust path weights based on feedback results to optimize the next round of strategies.

2. The road network traffic diversion method based on dynamic topology reconstruction according to claim 1, characterized in that: The identification of abnormal events and determination of their impact scope in step S1 are as follows: Step S1: Identification of Abnormal Events and Determination of Impact Scope When an anomaly occurs in the traffic network, timely and accurate identification of the event's location and impact range is fundamental to developing diversion strategies. By analyzing traffic data, the specific location of the event is dynamically determined, the affected road segments and their node sets are identified, and the degree of impact on the traffic network is quantified. This provides data support for subsequent topology reconstruction and diversion path optimization. Specifically, this includes the following sub-steps: Sub-step S11: Traffic data input and representation The input data consists of three parts: a weighted traffic network map, historical traffic data, and abnormal event data. The weighted graph structure of a traffic network is defined as G = (V, E, A); where V is the set of nodes, representing key nodes in the road network, E is the set of edges, representing road segments between nodes, and A is the adjacency matrix, describing the static connection relationships between nodes. Among them, W ij For edge e ij The weight represents the passage capacity; Historical traffic data Including each edge e ij Historical traffic Historical speed Road segment occupancy rate Abnormal event data is defined as a quadruple E = {L, S, T, D}, where L represents the location of the event, S represents the severity of the event, reflecting the percentage decrease in traffic capacity, T represents the time of the event, and D represents the duration of the event. Sub-step S12: Determining the location of the abnormal event Abnormal events are divided into explicit input events and potential abnormal events affected by input events. Among them, potential abnormal events need to be implicitly identified. For events that display input, simply use event e. k Position l k Mapped to network node v i ∈V: Among them, v c For core nodes; d(l k ,v i ) indicates the event location l k to node v i The standardized geographic distance, cos(θ(l) k ,v i The cosine of the angle between the event propagation direction and the road segment direction is used to measure directional consistency; w d and w θ These are the learnable weight coefficients for geographical distance and orientation factors, respectively; For potential anomalies that require implicit identification, the location of potential anomalies is dynamically identified by analyzing the differences between historical traffic data and input flow, specifically including the flow change rate ΔQ. ij Traffic condition degradation rate R i Calculation of (t): Among them, Q ij (t), V ij (t) represents the current flow rate and speed. For historical flow and speed; if |ΔQ ij |Exceeds the threshold and R i If (t) is less than the threshold and the state duration exceeds a certain value, then the road segment is determined to be an abnormal road segment, and the nodes included in the road segment are recorded in the affected node set v. affected middle; Sub-step S13: Identification of the scope of impact of abnormal events The attention mechanism model is used to quantify the weights of affected nodes and determine the scope of the event's influence; for each pair of nodes (v i ,v j Attention weight α ij Represents node v j For node v i Importance: Among them, W a F is a trainable weight matrix. i F j represents the feature representation of a node, and || represents the vector concatenation operation; Based on attention weights, the set of affected nodes v is further filtered. affected : v affected ={v i |α ic >T,d(v i ,v c )≤φ(s k )} Wherein, φ(s) k ) is based on the severity of the event. k The mapping function, α ic T represents the attention weights between the events and the nodes that influence them, and is the weight selection threshold.

3. The road network traffic diversion method based on dynamic topology reconstruction according to claim 2, characterized in that: The node dynamic representation update mentioned in step S2 is as follows: Step S2: Dynamically update the node representation After an abnormal event occurs, in order to accurately reflect the direct and indirect impacts of the event on traffic network nodes, it is necessary to dynamically update the feature representation of the nodes; by propagating the event features from the core nodes to the surrounding nodes, the spatial impact range and time decay effect of the abnormal event can be described. Based on step S12, the core node v is obtained. c By directly injecting event characteristics into its representation, the node's sensitivity to abnormal events is enhanced. The node's representation is updated as follows: in, For node v c The original feature representation, F e As a characteristic of the event, W e Weights are injected into learnable event features; For nodes affected by the core node, the node representation is decayed based on its distance to the core node and the severity of the event. The node representation is updated as follows: Where, η i The attenuation coefficient is defined as follows: Wherein d(v i ,v c ) represents node v i To the core node v c Geographical distance, hop(v i ,v c ) indicates that in the original road network topology, from v i To the core node v c The number of hops on the shortest path, i.e., the topological distance; λ1 and λ2 are learnable decay rates controlling the two distances; t k and d k It indicates the time and duration of the event, reflecting the process of the event's impact decaying over time.

Images