A method for dividing urban road network into sub-areas based on congestion index distribution

Abstract

The present invention discloses a method for dividing urban area road traffic network into sub-areas based on congestion index distribution, which includes abstracting the urban area road traffic network into a directed graph. 、 Intersections are abstracted into node sets V , the interval roads are abstracted into a directed edge set E; the congestion index of the node is used to sort the V Rearrange the middle nodes in descending order to get a new node set V’ , and then perform subdivision, thereby grouping highly congested nodes and their adjacent roads and intersections of the same congestion level into the same subset. The resulting partitioning is characterized by fine granularity, high cohesion, and low coupling, which facilitates the development of different congestion management solutions for areas with different congestion levels, reducing resource waste and improving management efficiency. This invention also proposes a comprehensive congestion index based on three traffic flow indicators based on vehicle trajectory data, overcoming the one-sidedness of using a single traffic flow indicator to identify urban road traffic congestion conditions.

Description

Technical Field

[0001] The present invention relates to a method for dividing a road traffic network into sub-areas, and in particular to a method for dividing a road traffic network into sub-areas in an urban area based on congestion index distribution. Background Art

[0002] Urban road traffic is the foundation of urban development. With the rapid development of the national economy and the rapid advancement of urban construction, the huge number of motor vehicles has placed tremendous pressure on urban road traffic. Urban road traffic congestion has shown a trend of spreading from point to line to surface. In particular, as the coverage of urban road traffic congestion expands, it has seriously hindered the development of urban construction.

[0003] The most direct manifestation of the growing scope of urban road traffic congestion is saturated traffic, which has become a common phenomenon in all large and medium-sized cities. Oversaturated urban road traffic has seriously impacted the efficiency of the road network, directly causing traffic delays and congestion. This situation poses new challenges to traditional traffic control theory. Coordinated control of urban regional road traffic is an effective way to alleviate urban road congestion and improve the efficiency of urban road networks. Its foundation lies in the accurate and effective subdivision of urban regional road networks.

[0004] The sub-division of urban road traffic network is to divide the complex urban road traffic network into several sub-regions according to certain established rules. The roads and intersections within each sub-region are highly correlated, and there are obvious regular distinctions and boundaries between the sub-regions.

[0005] Essentially, the division of urban road traffic networks into subregions is a graph partitioning problem. Graph partitioning involves dividing a graph's vertex set into k disjoint subsets, satisfying certain constraints between the subsets. This k-partitioning problem is an NP-complete problem. Currently, the main graph partitioning algorithms include geometric partitioning, spectral partitioning, combinatorial partitioning, and hierarchical partitioning.

[0006] (1) Graph partitioning algorithms based on geometric techniques: They are mainly partitioned by the coordinate relationship between nodes, but do not consider the coordinate relationship between nodes, including coordinate nested bisection method, recursive inertia bisection method, space filling curve method, etc.

[0007] (2) Graph partitioning algorithm based on spectral partitioning technology: It mainly focuses on the binary partitioning of graphs. Its basic idea is to divide the graph into smaller subgraphs according to the eigenvectors of the graph matrix to improve the parallel efficiency of graph partitioning.

[0008] (3) Graph partitioning algorithms based on combinatorial techniques: These algorithms attempt to group highly connected nodes together without considering the coordinate relationships between the nodes. This algorithm partitions the graph based on the connectivity of the graph, resulting in slow partitioning and low parallelization.

[0009] (4) Graph partitioning algorithm based on hierarchical partitioning technology: First, the original graph is compressed multiple times to generate a graph with fewer nodes. Then, the compressed graph is partitioned and the partitioned graph is mapped back to the original graph to achieve the partitioning of the original graph. In the process of using graph partitioning algorithms to divide urban area road traffic network into sub-areas, the key is to formulate partitioning rules. Common partitioning plans include:

[0010] 1. Classification by road congestion level: This involves classifying roads into different congestion levels based on urban road congestion parameters, such as vehicle queue time, average vehicle speed, and vehicle queue length. The road network is then divided into sub-areas based on the congestion levels, with different roads being assigned to different road network sub-areas.

[0011] 2. Division by vehicle distribution: This method uses the geographic location of vehicles, vehicle distribution, and density as a benchmark to divide the road network into sub-areas. The density of vehicles can intuitively reflect congested areas, thereby accurately dividing the urban road network into different levels of sub-areas.

[0012] 3. Segmentation by congestion type: This involves selecting congested areas based on the congestion situation on inter-section roads or intersections. The congestion type, such as recurring or sporadic, is then determined based on the spatiotemporal variation of congestion trends. Further segmentation is then performed based on this congestion type. Analyzing congestion based on spatiotemporal trends can better reflect regional congestion and reduce errors in road network sub-area division caused by sporadic congestion.

[0013] The concept of traffic control sub-zone was first proposed by American scholar Walinchus. Shen G et al. proposed a method for identifying the oversaturation state of intersections and a method for classifying congestion levels. Qian Liu et al. proposed that urban main lines need to be divided into several segments through dynamic division, so as to achieve coordinated control of each segment within the segment. Cuijiao Cheng described the research status of static and dynamic traffic control sub-zone division, pointed out the existing deficiencies, gave the principles for dividing traffic control sub-zones according to node status, and designed a traffic control sub-zone division algorithm in detail. J Luo et al. studied the traffic area of ​​urban road network divisions and proposed a division method that combines static division with dynamic division. Qiang Liu et al. proposed an improved Newman fast clustering algorithm based on weighted traffic networks. The algorithm determines each sub-zone according to the contribution of each sub-network to the modularity of the entire network, so as to overcome the shortcomings of the local search of the traditional Newman fast algorithm. Shaohu Tang et al. proposed a traffic collaborative control model for urban oversaturated zones and their associated areas. Tomas Potuzak proposed an effective road traffic network partitioning method based on a partitioning genetic algorithm that utilizes graph coarsening to improve its results. He also proposed an improved road traffic network partitioning method for distributed road traffic simulation based on multi-level graph partitioning and an improved partitioning genetic algorithm. Luhua Zhao et al. proposed a spatial statistics method for traffic partitioning based on the K-means algorithm. Li E et al. proposed a control sub-region partitioning method based on an improved FN partitioning algorithm. This method comprehensively considers factors such as intersection spacing, link flow, traffic dispersion, signal cycle, and link traffic status, and calculates the correlation between link flow, intersection signal cycle, and link traffic status at adjacent intersections.

[0014] Although the above-mentioned literatures all propose methods for traffic network sub-division from different perspectives, few of them take into account the dynamic nature of urban road traffic flow. Alternatively, the granularity of the road network sub-division is too large to be applied to actual urban road coordination control, or the methods are limited to optimizing the traffic network sub-division algorithm.

[0015] Currently, there are numerous methods for identifying urban road traffic congestion. Congestion standards vary across countries, regions, and roads, and are difficult to reconcile. Furthermore, existing congestion identification methods often only consider a few independent traffic flow indicators, failing to comprehensively consider urban road traffic congestion from multiple perspectives. Assuming the congestion index depends on the length of vehicle queues, even if the vehicle queue length is large, the overall travel time may be low due to the long green light duration at the intersection, which is insufficient to constitute severe congestion. Therefore, using a single traffic flow indicator to identify urban road traffic congestion is one-sided.

[0016] Glossary:

[0017] Vehicle trajectory data: refers to data that describes the attributes of vehicles in a spatiotemporal segment, mainly including the speed, location, and other attribute information of different vehicles at different time points. Combined with geographic coordinate information, it is possible to obtain the road sections and intersections where different vehicles travel at different time points, and then analyze the congestion of the urban road traffic network. Vehicle trajectory data can accurately reflect the attributes of vehicles and their movement trajectories. Therefore, compared with instantaneous time traffic flow data and discrete time traffic flow data, vehicle trajectory data can better reflect the traffic flow conditions and road traffic congestion conditions over a period of time. In addition, vehicle trajectory data has better versatility and adaptability than traffic flow data such as videos and license plates.

[0018] Vehicle trajectory data consists of multiple single-vehicle trajectory data within a specified road section and a specified time segment. A single vehicle refers to a single vehicle. Summary of the Invention

[0019] The purpose of the present invention is to provide a method for dividing urban regional road traffic network sub-areas based on congestion index distribution, which can overcome the defects of using a single traffic flow indicator to identify urban road traffic congestion conditions, and has low time complexity and reduces time cost; the coupling between areas is low, and the regional division is clearer.

[0020] In order to achieve the above object, the technical solution adopted by the present invention is as follows: a method for dividing urban area road traffic network sub-areas based on congestion index distribution, comprising the following steps:

[0021] (1) The urban area road traffic network is abstracted into a directed graph G = (V, E), where the intersection is abstracted into a node set V = {v1, v2, ..., v i ,…,v n}, the interval road is abstracted into a set of directed edges E = {e ij},1<=i,j<=n, e ij represents a directed edge from node i to node j;

[0022] (2) Calculate the congestion index of each directed edge and construct the congestion index set EC of directed edges = {ec ij},1≤i,j≤n, ec ij For directed edge e ij Congestion index;

[0023] (3) Calculate the congestion index of each node based on the congestion index of the directed edge, and construct the node congestion index set VC = {vc1, vc2, ..., vc i ,…,vcn}, vc i is the congestion index of node i;

[0024] (4) Rearrange the nodes in the node set V and the node congestion index set VC in descending order of the node congestion index to obtain a new node set V' = {v'1, v'2, ..., v' i ,…,v' n}, and the node congestion index set VC'={vc'1,vc'2,…,vc' i ,…,vc' n};

[0025] (5) Initialize the number of road network sub-areas k = 0;

[0026] (6) Let k = k + 1, create two empty sets, which are partition sets V k , temporary set V′ k ;

[0027] (7) The kth node v' in V' k Divide into set V k In, as V k The source point and initialize V′ k ;

[0028] (8) Create a set S containing v' k For each node in S, determine whether it belongs to V′. If it belongs to V′, add it to V k and V′ k and subtract it from V′;

[0029] (9) Process V′ in sequence k Among the nodes, the processing method of one node p is:

[0030] (91) Create a set S' containing all one-hop neighbor nodes of p. For each node in S', determine whether it belongs to V'. If it belongs to V' and its congestion index is not lower than that of the source node, add it to V. k and V′ k and subtract the node from V′, from V′ k Subtract node p from

[0031] (92) Determine V′ k and V′ is an empty set, if V′ k If V′ is not an empty set, repeat step (91). If V′ is not an empty set, repeat step (6).

[0032] (10) The division is completed, and the set V of k road network sub-areas is obtained. p={V1,V2,...,V k}.

[0033] As a preference: in step (3), the congestion index vc of node i is i Calculated by the following formula;

[0034]

[0035] Where l is the number of directed edges connected to node i, the directed edges are marked from 1 to l, and the congestion index of the mth directed edge is ec m , 1≤m≤l.

[0036] Preferably, in step (2), the congestion index is a comprehensive congestion index, and the comprehensive congestion index of a directed edge is obtained by the following method;

[0037] (21) Determine the interval road corresponding to the directed edge in the urban area road traffic network, specify a time segment and divide it into t sub-time segments, sample once at the end of each sub-time segment, and obtain the accumulated bicycle trajectory data within the sub-time segment, on the interval road, and in the corresponding direction. The bicycle trajectory data includes the time, coordinates, and speed of the vehicle at the time of sampling;

[0038] (22) Calculate the average speed of the interval travel of the directed edge

[0039]

[0040] Among them, 1≤x≤t is the xth sub-time period, is the average speed of all vehicles on the road in the xth sub-time period and in the corresponding direction;

[0041] (23) Calculate the traffic index TCRI of the directed edge link ;

[0042]

[0043] Flow(x) is the total number of vehicles on the road in the xth sub-time period in the corresponding direction, C power The traffic capacity of the road in this section;

[0044] (24) Calculate the average queuing time T of directed edges link_wait ;

[0045]

[0046] N wait (x) is the number of vehicles waiting in the corresponding direction of the road in the x-th sub-time period;

[0047] (25) Calculate the comprehensive congestion index C of the directed edge according to the following formula: link ;

[0048]

[0049] (26) Comprehensive congestion index C link Normalize it and convert it into a value in the range [0,1].

[0050] As a preference, the traffic capacity of the interval road in step (23) is the number of vehicles that can travel on the interval road in unit time.

[0051] As a preference: Step (4) reorders the nodes in descending order of their congestion index, and its time complexity is O(nlogn). From step (5) to step (10), a breadth-first search based on the degree sequence is performed. The time complexity of the breadth-first search depends on the storage structure used. When a two-dimensional array is used as the storage structure of the graph, the time complexity of the breadth-first search is O(nlogn). 2 ); When the adjacency list is used as the storage structure of the graph, the time complexity of breadth-first search is O(n+e); where e represents the number of edges in the graph, therefore, the time complexity of the urban area road traffic network sub-area division method based on congestion index distribution is between O(nlogn) and O(n 2 )between.

[0052] The present invention proposes a new method for calculating the comprehensive congestion index, which is related to the average speed V of the interval travel. link , Road Traffic Index TCRI link , average queuing time T link_wait This method is based on traditional traffic flow indicators on the one hand, and reflects the congestion of designated interval roads in regional roads on the other hand, which is more conducive to the division of urban regional road traffic network sub-areas.

[0053] About the average speed V of the interval link : This refers to the average speed of all vehicles on a given road section during a specified time interval. The average speed of vehicles can provide a direct indicator of road congestion. A traffic congestion speed threshold is also set for a given road, and the congestion level of the section is calculated by analyzing the average speed over different time periods.

[0054] About the traffic flow index TCRI of directed edges link : The ratio of the traffic flow of a road section to the road section's traffic capacity, which is used to determine the traffic flow status.

[0055] On the average queuing time T of directed edges link_wait: Refers to the average queue time of all vehicles on a given road section within a specified time interval. This section's average queue time reflects the amount of time vehicles are blocked on that section of road, providing a more intuitive picture of congestion.

[0056] Compared with the prior art, the advantages of the present invention are:

[0057] (1) A method for dividing urban road network subregions based on congestion index distribution is proposed. This method is applicable not only to the traffic congestion index in the prior art but also to the comprehensive congestion index of the present invention. Based on the congestion index, highly congested nodes and their adjacent roads and intersections with the same congestion level are divided into the same subregion. The result of the division has the characteristics of fine granularity, high cohesion, and low coupling. The precise division of road network subregions helps us to provide different congestion control solutions for areas with different congestion levels, reduce resource waste, and improve control efficiency.

[0058] (2) While proposing a sub-area division method, the present invention also redefines and calculates the traffic congestion index, and proposes a new calculation method for the comprehensive congestion index. The comprehensive congestion index is based on the fluidity characteristics of urban road traffic flow, and analyzes and calculates the time changes of vehicle trajectory data and geographic coordinates, so that the comprehensive congestion index is related to the average speed V of the interval travel of the directed edge. link , the traffic index TCRI of the directed edge link , the average queuing time T of directed edges link_wait These three traffic flow indicators are all related, thus forming a traffic flow congestion index with a trend. When we conduct a comprehensive analysis of these three traffic flow indicators, the comprehensive congestion index formed can more objectively describe the degree of road traffic congestion.

[0059] (3) The comprehensive congestion index of the present invention is based on vehicle trajectory data and has better versatility and adaptability than traffic flow data such as video and license plate data. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 Flowchart of the present invention;

[0061] Figure 2 In embodiment 2 of the present invention, a congestion index diagram of each directed edge is obtained through step (2);

[0062] Figure 3 In the second embodiment of the present invention, a congestion index diagram of each node is obtained through step (3);

[0063] Figure 4 This is a schematic diagram of dividing the traffic status of sub-areas in Example 2 of the present invention;

[0064] Figure 5This is a schematic diagram of the simulated road network for the n×n urban area road traffic in Example 3;

[0065] Figure 6a This is a schematic diagram of the mean square error of the congestion index of internal nodes in the unblocked area of ​​simulation experiment 1;

[0066] Figure 6b This is a schematic diagram of the mean square error of the congestion index of internal nodes in a lightly congested area of ​​simulation experiment 1;

[0067] Figure 6c This is a schematic diagram of the mean square error of the congestion index of internal nodes in the moderately congested area of ​​simulation experiment 1;

[0068] Figure 6d This is a schematic diagram of the mean square error of the congestion index of internal nodes in a heavily congested area of ​​simulation experiment 1;

[0069] Figure 7a This is a schematic diagram of the mean square error of the congestion index of internal nodes in the unblocked area of ​​simulation experiment 2;

[0070] Figure 7b This is a schematic diagram of the mean square error of the congestion index of internal nodes in the lightly congested area of ​​simulation experiment 2;

[0071] Figure 7c This is a schematic diagram of the mean square error of the congestion index of internal nodes in the second-degree congestion area of ​​the simulation experiment;

[0072] Figure 7d This is a schematic diagram of the mean square error of the congestion index of internal nodes in the second most congested area of ​​the simulation experiment;

[0073] Figure 8 The time consumption for dividing the urban area road traffic network into sub-areas in simulation experiment three. DETAILED DESCRIPTION

[0074] The present invention will be further described below with reference to the accompanying drawings.

[0075] Example 1: See Figure 1 ,A method for dividing urban regional road traffic network sub-areas based on congestion index distribution, comprising the following steps;

[0076] (1) The urban area road traffic network is abstracted into a directed graph G = (V, E), where the intersection is abstracted into a node set V = {v1, v2, ..., v i ,…,v n}, the interval road is abstracted into a set of directed edges E = {e ij},1<=i,j<=n, e ij represents a directed edge from node i to node j;

[0077] (2) Calculate the congestion index of each directed edge and construct the congestion index set EC of directed edges = {ec ij},1≤i,j≤n, ec ij For directed edge e ij Congestion index;

[0078] (3) Calculate the congestion index of each node based on the congestion index of the directed edge, and construct the node congestion index set VC = {vc1, vc2, ..., vc i ,…,vc n}, vc i is the congestion index of node i, the congestion index vc of node i i Calculated by the following formula;

[0079]

[0080] Where l is the number of directed edges connected to node i, the directed edges are marked from 1 to l, and the congestion index of the mth directed edge is ec m , 1≤m≤l.

[0081] (4) Rearrange the nodes in the node set V and the node congestion index set VC in descending order of the node congestion index to obtain a new node set V' = {v'1, v'2, ..., v' i ,…,v' n}, and the node congestion index set VC'={vc'1,vc'2,…,vc' i ,…,vc' n};

[0082] (5) Initialize the number of road network sub-areas k = 0;

[0083] (6) Let k = k + 1, create two empty sets, which are partition sets V k , temporary set V′ k ;

[0084] (7) The kth node v' in V' k Divide into set V k In, as V k The source point and initialize V′ k ;

[0085] (8) Create a set S containing v' k For each node in S, determine whether it belongs to V′. If it belongs to V′, add it to V k and V′ k and subtract it from V′;

[0086] (9) Process V′ in sequence k Among the nodes, the processing method of one node p is:

[0087] (91) Create a set S' containing all one-hop neighbor nodes of p. For each node in S', determine whether it belongs to V'. If it belongs to V' and its congestion index is not lower than that of the source node, add it to V. k and V′ k and subtract the node from V', from V' k Subtract node p from

[0088] (92) Determine V′ k and V′ is an empty set, if V′ k If V′ is not an empty set, repeat step (91). If V′ is not an empty set, repeat step (6).

[0089] (10) The division is completed, and the set V of k road network sub-areas is obtained. p ={V1,V2,...,V k}.

[0090] In step (2), the congestion index is a comprehensive congestion index, and the comprehensive congestion index of a directed edge is obtained by the following method:

[0091] (21) Determine the interval road corresponding to the directed edge in the urban area road traffic network, specify a time segment and divide it into t sub-time segments, sample once at the end of each sub-time segment, and obtain the accumulated bicycle trajectory data within the sub-time segment, on the interval road, and in the corresponding direction. The bicycle trajectory data includes the time, coordinates, and speed of the vehicle at the time of sampling;

[0092] (22) Calculate the average speed of the interval travel of the directed edge

[0093]

[0094] Among them, 1≤x≤t is the xth sub-time period, is the average speed of all vehicles on the road in the xth sub-time period and in the corresponding direction;

[0095] (23) Calculate the traffic index TCRI of the directed edge link ;

[0096]

[0097] Flow(x) is the total number of vehicles on the road in the xth sub-time period in the corresponding direction, C power The traffic capacity of the road in this section;

[0098] (24) Calculate the average queuing time T of directed edges link_wait ;

[0099]

[0100] N wait (x) is the number of vehicles waiting in the corresponding direction of the road in the x-th sub-time period;

[0101] (25) Calculate the comprehensive congestion index C of the directed edge according to the following formula: link ;

[0102]

[0103] (26) Comprehensive congestion index C link Normalize it and convert it into a value in the range [0,1].

[0104] The traffic capacity of the section road in step (23) is the number of vehicles that can travel on the section road in a unit time.

[0105] Step (4) reorders the nodes in descending order of their congestion index, and its time complexity is O(nlogn). From step (5) to step (10), a breadth-first search based on the degree sequence is performed. The time complexity of the breadth-first search depends on the storage structure used. When a two-dimensional array is used as the storage structure of the graph, the time complexity of the breadth-first search is O(nlogn). 2 ); When the adjacency list is used as the storage structure of the graph, the time complexity of breadth-first search is O(n+e); where e represents the number of edges in the graph. Therefore, the time complexity of the urban area road traffic network sub-area division method based on the congestion index distribution is between O(nlogn) and O(n 2 )between.

[0106] Example 2: See Figure 1-Figure 4 , we take a specific urban area road traffic network as an example, and divide the road network subnet based on Example 1. Figure 2 The urban area road network has a 4×4 structure, with a total of 16 intersections, each of which connects to 8 sections of road. We sort the nodes from 1 to 16 and mark them in the figure, then process them according to the following steps:

[0107] (1) The urban area road traffic network is abstracted into a directed graph G = (V, E) with 16 nodes, and the 16 intersections are abstracted into a node set V = {v1, v2, v3, ..., v 16}, the interval road is abstracted into a set of directed edges E = {e ij},1<=i,j<=n,

[0108] (2) Calculate the congestion index of each directed edge. In this embodiment, the congestion index of the directed edge is specifically the comprehensive congestion index of the directed edge, which is calculated by formulas (2) to (5) of the present invention and marked in Figure 2 middle;

[0109] (3) Calculate the congestion index of each node and mark it in Figure 3 In this embodiment, the congestion index of a node is calculated by formula (1);

[0110] (4) Rearrange the nodes in descending order of congestion index to obtain a new node set V' = {v'1, v'2, ..., v'3, ..., v' 16}, v'1~v' 16 The corresponding node numbers are 6, 9, 5, 2, 10, 1, 4, 7, 3, 11, 15, 8, 12, 14, 16;

[0111] (5) Initialize the number of road network sub-areas k = 0;

[0112] Regarding steps (6)-(9) as a loop, we only take the first loop k=0+1 as an example, and the processing methods of the remaining loops are the same.

[0113] (6) Let k = 0 + 1 = 1, and create two empty sets V1 and V1′;

[0114] (7) Partition the first node v6 in V′ into V1 and make it the source of V1, obtaining V1 = {v6};

[0115] (8) Create a set S containing all one-hop neighbor nodes of v6 = {v2, v5, v7, v 10}, and judge whether they belong to V′ in turn. They all belong to V′, so we get V1={v6,v2,v5,v7,v 10}, and V1′={v2,v5,v7,v 10}, now the set V' has subtracted v2, v5, v7, and v 10 , contains only 12 nodes;

[0116] (9) Process V1′={v2,v5,v7,v 10 If the congestion index of a node in v6 is greater than or equal to the congestion index of v6, it will be divided into v1 and v1′, and the processed node will be deleted from v1′;

[0117] And so on, until V1′ is empty, we get V1={v6,v2,v5,v7,v 10,v9},V1′={v1,v4,v3,v 11 ,v 13 ,v 15 ,v8,v 12 ,v 14 ,v 16}, then repeat steps (6)-(9) until all nodes in V' are divided;

[0118] (10) Finally, we get the set V of 6 road network sub-areas p = {V1, V2, V3, V4, V5, V6}, the 6 sub-areas are: V1 = {v6, v2, v5, v7, v 10 ,v9}, V2={v1}, V3={v4,v3,v8}, V4={v 11 ,v 12 ,v 15}、V5={v 13 ,v 14}、V6={v 16}.

[0119] The traffic status of the sub-area can be defined according to the actual situation. For example, in this embodiment, we divide the traffic status into four types: smooth, lightly congested, moderately congested and heavily congested. Among them, the congestion index of the smooth state is [0, 0.4), the congestion index of the lightly congested state is [0.4, 0.6), the congestion index of the moderately congested state is [0.6, 0.8), and the congestion data of the heavily congested state is [0.8, 1]. And the congestion level of the sub-area is determined according to the congestion level of the source point of each road network sub-area, and finally it can be obtained that V1 = {v6, v2, v5, v7, v 10 ,v9} is a heavy congestion area, V2={v1} and V3={v4,v3,v8} are moderate congestion areas, V4={v 11 ,v 12 ,v 15} is a lightly congested area, V5={v 13 ,v 14} and V6={v 16} is a smooth area, such as Figure 4 shown.

[0120] Example 3: See Figure 5 -arrive Figure 8 According to the fine-grained, high-cohesion, and low-coupling goals of the urban area road traffic network sub-area division proposed by the present invention, this paper uses the SUMO traffic simulation tool to design three simulation experiments to verify the effectiveness of the method of the present invention.

[0121] The first simulation experiment used the mean squared deviation of the node congestion index within the partition set as an evaluation metric to verify the high cohesion of the proposed urban regional road traffic network sub-region partitioning algorithm. The second simulation experiment used the mean squared deviation of the node congestion index between partition sets as an evaluation metric to verify the low coupling of the proposed urban regional road traffic network sub-region partitioning algorithm. The third simulation experiment used the time required for sub-region partitioning as an evaluation metric to verify the feasibility of the proposed urban regional road traffic network sub-region partitioning algorithm.

[0122] In order to better calculate the experimental results of the urban area road traffic network sub-area division algorithm proposed in this paper, this paper uses the SUMO traffic simulation tool to construct an n×n urban area road traffic simulation network, such as Figure 5 As shown in Figure 2, the scale of traffic network nodes ranges from 100 to 10,000. The node congestion index adopts random distribution, uniform distribution, and normal distribution in [0,1], and the adjacency list is used as the storage structure of the graph.

[0123] Simulation experiment 1: Due to the different road network structures and congestion distributions in urban areas, the number of sub-areas k that are finally divided is uncertain. According to the division algorithm steps in Section 3.4.1, the highly congested nodes and their one-hop nodes, as well as the nodes with a congestion level not lower than the two-hop (inclusive) or more nodes of the division set source point are finally divided into the same set, forming a set of nodes with the same congestion level and a high mutual influence. Therefore, the cohesion of the urban area road traffic network sub-area division can be evaluated based on the mean square deviation of the congestion index of each node in the set. According to the division method of the present invention, because the source point of each road network sub-area division set has a strong correlation with its one-hop node, only the mean square deviation of the congestion index of each road network sub-area division set source point and its two-hop and above nodes is calculated.

[0124] The steps for calculating the mean square error of the congestion index of internal nodes in the road network sub-area division set are as follows.

[0125] (1) Assume that the urban area road traffic network has n intersections and is divided into k subsets. Each road network sub-area partition set is defined as V i , 1≤i≤k,v i1 V i The source point.

[0126] (2) Let V i_1hop v i1 One-hop node set, let V i_nhop v i1 The set of nodes with two or more hops, that is, V i ={v i1 ,V i_1hop ,Vi_nhop}.

[0127] (3) Let V i ′ is v i1 and the set of nodes with two hops or more, that is, V i ′={v i1 ,V i_nhop}, defined as V i ′={v′ i1 ,v′ i2 ,...,v′ im}, 1≤m≤|V i_nhop |+1.

[0128] (4) Define V i ′’s node congestion index set VC′ i ={vc′ i1 ,vc′ i2 ,...,vc′ im}, there exists C(v′ im )=vc′ im ,1≤m≤|V i_nhop |+1.

[0129] (5) Urban area road traffic network sub-area division set V i The mean square error of the congestion index of the midpoint is defined as follows:

[0130]

[0131] in, For the set V′ i The average value of the node congestion index; 1≤i≤k; 1≤m≤|V i_nhop |+1.

[0132] This paper conducts 100 road network subdivision experiments for each traffic network and each traffic congestion index distribution, and then calculates the urban area road traffic network subdivision set V k The average value of the mean square error of the node congestion index. The experimental results are shown in Tables 1-3.

[0133] Table 1 Results of the cohesion experiment on the urban road network sub-division (congestion index follows random distribution)

[0134]

[0135] Table 2 Results of the cohesion experiment on the urban road network sub-division (congestion index follows random distribution)

[0136]

[0137]

[0138] Table 3 Results of the cohesion experiment on the urban road network sub-division (congestion index follows Poisson distribution)

[0139]

[0140] From the cohesion test results of the urban area road traffic network sub-area division, it can be seen that under the conditions of different road network scales and different congestion index distributions, the mean square deviation of the congestion index of the internal nodes of the unobstructed area after division is between 0.15 and 0.25, and the mean square deviation of the congestion index of the internal nodes of the lightly congested, moderately congested and heavily congested areas is between 0.08 and 0.12. The mean square deviation of the congestion index of the internal nodes of the divided areas is consistent with the distribution of the congestion index of the divided areas, and the values ​​are small, indicating that the congestion indexes of the internal nodes of the urban area road traffic network sub-area division are similar. The urban area road traffic network sub-area division algorithm proposed in the present invention achieves the goal of high cohesion of the congestion index of the divided set.

[0141] The second simulation experiment used the mean square error (MSE) of the node congestion index across network subdivisions. This is because, according to the inventive method, the congestion levels of nodes within urban road network subdivisions are uniform and highly influenced by each other, while the congestion levels between subdivisions vary and are clearly demarcated.

[0142] The steps for calculating the mean square error of the node congestion index between the road network sub-area division sets are as follows.

[0143] (1) Assume that the urban area road traffic network has n intersections and is divided into k subsets. Each road network sub-area partition set is defined as V i , 1≤i≤k,v i1 V i The source point.

[0144] (2) Let V i_1hop v i1 One-hop node set, let V i_nhop v i1 The set of nodes with two or more hops, that is, V i ={v i1 ,V i_1hop ,V i_nhop}.

[0145] (3) Let V i ′ is v i1 and the set of nodes with two hops or more, that is, V i ′={v i1 ,V i_nhop}, defined as Vi ′={v′ i1 ,v′ i2 ,...,v′ im}, 1≤m≤|V i_nhop |+1.

[0146] (4) Define V i ′’s node congestion index set VC′ i ={vc′ i1 ,vc′ i2 ,...,vc′ im}, there exists C(v′ im )=vc′ im ,1≤m≤|V i_nhop |+1.

[0147] (5) Calculate the set V i Congestion Index 1≤i≤k.

[0148] (6) Define V i_neighbor ={V j |V j V i The adjacent congested area partition set}, set V i The mean square error of the congestion index between the adjacent road network sub-area division sets Among them, |V i_neighbor | is the set V i The number of adjacent road network sub-area partition sets, 1≤i≤k.

[0149] The present invention performs 100 road network sub-division experiments for each traffic network and each traffic congestion index distribution, and then calculates the urban area road traffic network sub-division set V k The average value of the mean square error of the congestion index of the adjacent road network sub-area division is shown in Table 4 to Table 6. Figures 7a to 7d shown.

[0150] Table 4 Results of coupling experiments on urban road network sub-division (congestion index follows random distribution)

[0151]

[0152] Table 5 Results of coupling experiments on urban road network sub-division (congestion index follows uniform distribution)

[0153]

[0154]

[0155] Table 6 Results of coupling experiments on urban road network sub-division (congestion index follows Poisson distribution)

[0156]

[0157] According to the cohesion test results of the urban regional road traffic network sub-area division, under the conditions of different road network scales and different congestion index distributions, the mean square deviation of the congestion index of nodes between unobstructed areas after division is between 0.45 and 0.55, the mean square deviation of the congestion index of nodes between lightly congested and moderately congested areas is between 0.25 and 0.35, and the mean square deviation of the congestion index of nodes between heavily congested areas is between 0.38 and 0.48. The mean square deviations of the congestion index between the divided areas are consistent with the distribution of the congestion index of the divided areas, and the values ​​are large, indicating that the congestion index of the divided sub-areas of the urban regional road traffic network varies greatly. The urban regional road traffic network sub-area division algorithm proposed in this paper achieves the goal of low coupling of the congestion index of the divided sets.

[0158] Simulation experiment 3: For each traffic network and each traffic congestion index distribution, 100 road network sub-area division experiments were performed, and then the average value was taken. The experimental results are shown in Table 7 and Figure 8 shown.

[0159] Table 7 Experimental results on the time complexity of urban road network sub-division

[0160]

[0161] The test results for subdividing urban road networks show that the time required for subdividing increases with the number of nodes in the road network. Across the node scale, the subdivision time exhibits an O(nlogn) distribution, consistent with the time complexity analysis results previously described. Furthermore, the test results also demonstrate that the proposed subdivision algorithm is independent of the congestion index distribution. The subdivision time is nearly identical when the congestion index follows a random, uniform, and Poisson distribution, indicating that the subdivision time is only related to the number of nodes.

[0162] In summary, the urban area road traffic network sub-area division algorithm based on congestion index distribution proposed in the present invention has the characteristics of fine granularity, high cohesion and low coupling. Moreover, under the condition of using adjacency list as the graph storage structure, its time complexity is O(nlogn), which has good practicality.

[0163] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for dividing urban area road traffic network into sub-areas based on congestion index distribution, characterized by: The following steps are included: (1) The urban area road traffic network is abstracted into a directed graph G = (V, E), where the intersection is abstracted into a node set V = {v1, v2, ..., v i ,…,v n }, the interval road is abstracted into a set of directed edges e ij represents a directed edge from node i to node j; (2) Calculate the congestion index of each directed edge and construct the congestion index set of directed edges ec ij For directed edge e ij Congestion index; (3) Calculate the congestion index of each node based on the congestion index of the directed edge, and construct the node congestion index set VC = {vc1, vc2, ..., vc i ,…,vc n }, vc i is the congestion index of node i; (4) Rearrange the nodes in the node set V and the node congestion index set VC in descending order of the node congestion index to obtain a new node set V' = {v'1, v'2, ..., v' i ,…,v' n }, and the node congestion index set VC'={vc'1,vc'2,…,vc' i ,…,vc' n }; (5) Initialize the number of road network sub-areas k = 0; (6) Let k = k + 1, create two empty sets, which are partition sets V k , temporary set V k '; (7) The kth node v' in V' k Divide into set V k In, as V k The source point and initialize V k '; (8) Create a set S containing v' k For each node in S, determine whether it belongs to V′. If it belongs to V′, add it to V k and V k ′ and subtract it from V′; (9) Process V in sequence k ′, the processing method of one node p is: (91) Create a set S' containing all one-hop neighbor nodes of p. For each node in S', determine whether it belongs to V'. If it belongs to V' and its congestion index is not lower than that of the source node, add it to V. k and V′ k and subtract the node from V′, from V′ k Subtract node p from (92) Determine V k ' and V' are empty sets, if V k If V′ is not an empty set, repeat step (91). If V′ is not an empty set, repeat step (6). (10) The division is completed, and the set V of k road network sub-areas is obtained. p ={V1,V2,...,V k }.

2. The method for dividing urban area road traffic network into sub-areas based on congestion index distribution according to claim 1, characterized in that: In step (3), the congestion index vc of node i i Calculated by the following formula; Where l is the number of directed edges connected to node i, the directed edges are marked from 1 to l, and the congestion index of the mth directed edge is ec m , 1≤m≤l.

3. The method for dividing urban area road traffic network into sub-areas based on congestion index distribution according to claim 2, characterized in that: In step (2), the congestion index is a comprehensive congestion index, and the comprehensive congestion index of a directed edge is obtained by the following method: (21) Determine the interval road corresponding to the directed edge in the urban area road traffic network, specify a time segment and divide it into t sub-time segments, sample once at the end of each sub-time segment, and obtain the accumulated bicycle trajectory data within the sub-time segment, on the interval road, and in the corresponding direction. The bicycle trajectory data includes the time, coordinates, and speed of the vehicle at the time of sampling; (22) Calculate the average speed of the interval travel of the directed edge Among them, 1≤x≤t is the xth sub-time period, is the average speed of all vehicles on the road in the xth sub-time period and in the corresponding direction; (23) Calculate the traffic index TCRI of the directed edge link ; Flow(x) is the total number of vehicles on the road in the xth sub-time period in the corresponding direction, C power The traffic capacity of the road in this section; (24) Calculate the average queuing time T of directed edges link_wait ; N wait (x) is the number of vehicles waiting in the corresponding direction of the road in the x-th sub-time period; (25) Calculate the comprehensive congestion index C of the directed edge according to the following formula: link ; (26) Comprehensive congestion index C link Normalize it and convert it into a value in the range [0,1].

4. The method for dividing urban area road traffic network into sub-areas based on congestion index distribution according to claim 3, characterized in that: The traffic capacity of the section road in step (23) is the number of vehicles that can travel on the section road in a unit time.

Images