Dynamic Traffic Zoning Methods and Systems for Urban Road Networks Based on Commuting Needs

Abstract

The present invention provides a method for dynamic traffic zoning of urban road networks for commuting needs, the method comprising the following steps: S1, preparing data and preprocessing it; S2, clustering commuting POIs based on a community detection algorithm; S3, dynamically zoning urban road networks based on a community detection algorithm. The present invention only performs traffic zoning on the road network of traffic management, effectively simplifying the content of traffic zoning. This simplification allows traffic zoning to be more focused on the core part of urban traffic management, namely the urban road network, thereby improving the efficiency and practicality of traffic zoning.

Description

Technical Field

[0001] The present invention relates to the field of traffic management, and in particular to a method and system for dynamic traffic zoning of an urban road network oriented to commuting needs. Background Art

[0002] The main purpose of traffic zone division is to optimize traffic flow and improve the overall efficiency of the traffic system. However, the current traffic zone division method mainly uses roads as boundaries to divide areas containing building complexes. This division based on building complexes has not shown clear practical value in actual traffic management. In terms of urban traffic management and control, traffic lights are mainly used to regulate the traffic flow of each section on the urban road network. Building complexes or road sections within urban communities do not fall within the scope of traffic management. Therefore, the traffic zoning method based on building complexes presents a lot of redundant information in this context, which is not conducive to refined traffic flow control.

[0003] In addition, the current traffic zone division method is universal, that is, it does not consider traffic changes in different time periods and traffic needs for different travel purposes. This comprehensive division ignores the unique characteristics of urban traffic flow in different time periods and under different demands, which to some extent affects the effective response and management of the traffic system to peak traffic pressure. Therefore, the dynamic zone division method for specific traffic demands (such as commuting) and traffic flow changes during special periods (such as peak hours) has become an urgent problem to be solved in current urban traffic planning and management. Summary of the invention

[0004] In order to solve the above existing technical problems, the present invention provides a method for dynamic traffic zoning of urban road networks for commuting needs. The method for dynamic traffic zoning of urban road networks for commuting needs of the present invention comprises the following steps:

[0005] S1. Prepare and preprocess data;

[0006] S2, clustering commuting POIs based on community detection algorithm;

[0007] S3. Dynamic traffic zoning of urban road networks based on community detection algorithm.

[0008] Furthermore, S1 specifically includes the following steps:

[0009] Step 1-1: Road network data preparation and preprocessing

[0010] Obtain road network data from the city map, that is, the road segment set R = (r0, r1, ..., r m );

[0011] Among them, the attributes contained in the mth road segment are Indicates the unique identifier of the mth road segment; Indicates the geographical location information of the mth road segment, specifically a POLYGON type ((x0,y0),(x1,y1),…,(x j ,y j ))((x j ,y j ) represents the coordinates of the jth vertex of the road segment boundary); Indicates the maximum speed limit value of the mth road section; Indicates the number of lanes contained in the mth road segment; represents the length of the mth road segment; Indicates the unique identifier of the starting intersection of the mth road segment; Indicates the unique identifier of the ending intersection of the mth road segment;

[0012] Step 1-2: Preprocessing of OD travel data during peak hours

[0013] Filter OD travel records: Get OD travel records. The record set is represented by T = {t0, t1, …, t l}, l is the record number; each record t l Filter out the following attribute values in, and Respectively represent the start and end date and time of the lth trip record, and Respectively represent the geographic coordinate information of the start and end of the lth trip record and represents the travel distance of the lth trip record; then, use the formula By cleaning out invalid records, and the peak hour information of the city to filter out the peak hour OD travel records;

[0014] Match the origin and destination segments of OD travel records; And the geographic data of the road segment obtained in step 1-1 Use the shortest Euclidean distance formula from point to polygon to get the identifier of the nearest road segment of the starting point of each OD trip record With the help of and the geographic data of the road segment Get the identifier of the nearest road segment to the end point of each OD travel record Update each record l The attribute value is

[0015] Step 1-3: Commute POI data preparation and preprocessing

[0016] First, commuting-related POIs are filtered out from all POIs in the city; then the data content required for subsequent steps is filtered out from all the data content of the POIs; finally, the data and matrix required for POI clustering are calculated.

[0017] Furthermore, steps 1-3 specifically include the following steps:

[0018] Step 1-3-1: Filter and extract attributes of commuting POIs

[0019] Filter commuting POI: Get data of all POIs in the current city area on the map;

[0020] Determine the basic attributes of commuting POI: Through the filtering in the above steps, we get the commuting POI set P = {p0, p1, ..., p i}, where the attributes of the i-th commuting POI are Represents the unique identifier of the i-th POI, Indicates the longitude and latitude information of the i-th POI (x i ,y i ), Indicates the type of the i-th POI;

[0021] Improve commuting POI attributes: with the help of geographical data of each POI And the geographic data of the road segment obtained in step 1-1 Use the shortest Euclidean distance formula from point to polygon to get the identifier of the nearest road segment for each POI Get further improved POI attributes

[0022] Furthermore, steps 1-3 specifically include the following steps:

[0023] Step 1-3-2: Calculation of the Euclidean distance matrix of commuting POIs

[0024] The geographic coordinates of each commuting POI obtained by step 1-3-1 And use the Euclidean distance formula to calculate the distance between the i-th and j-th commuting POIs

[0025]

[0026] Then, the Euclidean distance between two commuting POIs is converted into a two-dimensional Euclidean distance matrix D eucliden :

[0027]

[0028] Furthermore, steps 1-3 specifically include the following steps:

[0029] Step 1-3-3: Calculation of commuting POI path distance matrix and path segment number matrix

[0030] Calculate the shortest path between commuting POIs: Use the nearest road segment number of each commuting POI obtained in step 1-3-1 And use Dijkstra's shortest path search algorithm to obtain the shortest path between the i-th and j-th commuting POIs

[0031] Calculate the path distance matrix between commuting POIs: Using the length attribute of the road segment Use the following formula to calculate the path between the i-th and j-th commuting POIs: ij Length

[0032]

[0033] in, Indicates path path ij The length of the road segment closest to the k-th POI, represents the arc length of the intersection between the kth to k+1th sections in the path. Then, the path distance between two commuting POIs is converted into a two-dimensional path matrix D path :

[0034]

[0035] Calculate the number of road segments in the path between commuting POIs: Statistical path Dpath ij The number of segments in the middle, and the number of path segments n between the i-th and j-th commuting POIs ij :

[0036]

[0037] Convert the number of sections in the path between two commuting POIs into a two-dimensional path section number matrix N:

[0038]

[0039] Furthermore, S2 specifically includes the following steps:

[0040] Step 2-1: Graph structuring of commuting POI information

[0041] Before applying the community detection algorithm to cluster commuting POIs, the relevant information needs to be represented by a graph structure. The representation process first establishes the basic graph structure, and then adds the commuting POI information processed in steps 1-3-2 and 1-3-3 to the edge weight.

[0042] Step 2-3: Commute POI clustering based on community detection Louvain algorithm

[0043] For the graph structure G obtained in step 2-1 poi =(V poi ,E poi ), the community detection Louvain algorithm is used to cluster commuting POIs. The core of this algorithm is to optimize the community structure by continuously moving each POI node to different communities, achieving modular Q graph structure. poi The goal of maximizing. Among them, the modularity of the graph structure Q poi The calculation formula is as follows:

[0044]

[0045] in, Representation graph G poi The composite weight sum of all edges in , is the composite weight between the i-th and j-th POI nodes calculated in step 2-1-2, r poi is the clustering resolution parameter, k i and k j represents the sum of the weights of all edges associated with the i-th and j-th POI nodes, c i and c j denote the communities to which the i-th and j-th POI nodes are assigned, δ(c i ,c j ) is an indicator function, c i =c j The function value is 1 when , otherwise it is 0;

[0046] When Figure G poi Modular Q poi After maximization, the partition of commuting POIs is completed, and the cluster label of each POI is obtained. The partition result diagram is as follows Figure 3 Update the result to the POI attribute set of step 1-3-1 to obtain the attribute set of the i-th commuting POI

[0047] Furthermore, step 2-1 specifically includes the following steps:

[0048] Step 2-1-1: Construction of basic graph structure

[0049] Each commuting POI is a node in the graph structure, with the unique identifier of the POI As the node attribute, a connecting edge is established between different commuting POIs to construct the basic graph structure G of commuting POIs poi =(V poi ,E poi ). Among them, V poi represents a node, E poi represents an edge;

[0050] Step 2-1-2: Calculation of edge composite weight

[0051] Define the composite weight of the edge: edge E obtained in step 2-1-1 poi In the example, the edge between the i-th and j-th commuting POIs has a composite weight Specifically involved: The Euclidean distance sub-weight between two POIs The sub-weight of the same nearest road segment between two POIs Sub-weight for limiting the number of road segments in the path between two POIs The Jaccard similarity sub-weight between the closest node sets of the closest k-path distance between two POIs

[0052] Calculate the composite weight of the edge: the composite weight of the edge between the i-th and j-th commuting POI nodes The calculation formula is as follows:

[0053]

[0054] Among them, (α eucliden ,α same ,α n ,α J ) are the coefficients of the four sub-weights. In order to reflect the importance of different sub-weights, the order is Then the values ​​of the four sub-weight coefficients are set to (α eucliden ,α same ,α n ,α J )=(0.4,0.3,0.2,0.1);

[0055] Calculate the sub-weight of the edge: the Euclidean distance sub-weight of the edge between the i-th and j-th commuting POI nodes The calculation formula is as follows:

[0056]

[0057] Same nearest road segment sub-weight The calculation formula is as follows:

[0058]

[0059] Sub-weight for limiting the number of segments in a path The calculation formula is as follows:

[0060]

[0061] The Jaccard similarity sub-weight between the nearest k-path distance and the closest node set The calculation formula is as follows:

[0062]

[0063] Among them, N i and N j Respectively represent the POI sets with the k closest path distances to the i-th and j-th commuting POIs.

[0064] Furthermore, S3 specifically includes the following steps:

[0065] Step 3-1: Graph structuring of road network information

[0066] First, the historical OD travel data, road network data and commuting POI clustering results during peak hours are represented by a graph structure. The representation process first establishes a basic graph structure, and reflects the starting and ending frequencies of the historical OD travel data during peak hours in the weights of the nodes. Then, the attribute differences between road sections, as well as the commuting POI clustering results and the association between POI and road sections, are reflected in the composite weights of the edges of the graph structure.

[0067] Step 3-2: Road network partitioning based on community detection Leiden algorithm

[0068] For the graph structure G obtained in step 3-1 edge =(V edge ,E edge ), using the community detection Leiden algorithm that emphasizes connection relationships to partition the road network; by continuously moving each road node to different communities to optimize the community structure, the modular Q edge The goal of maximization; among them, the modularity of the graph structure Q edge The calculation formula is as follows:

[0069]

[0070] in, is the composite weight coefficient of the edge between the mth and nth road segment nodes calculated in step 3-1-2, r edge is the resolution parameter of the partition, k m and k nrepresents the sum of the weights of all edges associated with the mth and nth road segment nodes, respectively, c m and c n denote the communities to which the mth and nth road segment nodes are assigned, δ(c m ,c n ) is an indicator function;

[0071] When Figure G edge Modular Q edge After maximization, the partition of the road network is completed, and the partition label of each road section is obtained. Update the result to the segment attribute set in step 1-1 to obtain the attribute set of the mth segment Ultimately, dynamic traffic zoning of the road network can be achieved.

[0072] Furthermore, step 3-1 specifically includes the following steps:

[0073] Step 3-1-1: Determination of the basic diagram

[0074] Determine the node V edge Information: The road segment set R obtained by step 1-1 = (r0, r1, ..., r m ), each road segment is a node in the graph, and the unique identifier of the road segment is the node attribute, and the node weight is the normalized frequency f of each road segment as the origin and destination road segment in the historical OD travel records during peak hours. m , and its calculation formula is:

[0075]

[0076] Among them, c m is the total frequency of the mth road segment as the starting and ending point in the historical OD travel records during peak hours, max{c0,c1,…,c m} represents the maximum value of the total frequency of all road sections as the above starting and ending point sections;

[0077] Determine edge E edge Information: The starting intersection mark of the m-section obtained in step 1-1 and end intersection sign Determine the link relationship between each road segment; if there is a link relationship between road segment m and road segment n, an edge is established between the two road segments, and the basic sub-weight of the edge is assigned The calculation formula is as follows:

[0078]

[0079] Step 3-1-2: Calculate the composite weight of the edges in the graph

[0080] Define the composite weight of the edge: edge E obtained in step 3-1-1 edge In the example, the edge between the mth and nth segments has a composite weight Specifically involved: Basic sub-weight The weight of the nearest neighboring road segment of POI with the same clustering result Sub-weights of roads with the same speed limit Sub-weights of road segments with the same number of lanes Sub-weights of two road segments with similar lengths

[0081] Calculate the composite weight of the edge: the composite weight of the edge between the mth and nth segment nodes The calculation formula is as follows:

[0082]

[0083] Among them, (β b ,β p ,β s ,β lane ,β length ) are the coefficients of the five sub-weights. In order to reflect the importance of different sub-weights, the order is Then the values ​​of the five sub-weight coefficients are set to (β b ,β p ,β s ,β lane ,β length )=(0.01,0.699,0.1,0.1,0.1);

[0084] Calculate the sub-weight of the edge: the sub-weight of the nearest neighbor segment of the POI with the same clustering result The calculation method is: if there are the same clustering results in the POIs associated with two road segments, then Otherwise, it is 0; the sub-weight of the road section with the same speed limit The calculation formula is as follows:

[0085]

[0086] Sub-weights of road segments with the same number of lanes The calculation formula is as follows:

[0087]

[0088] Sub-weights of two road segments with similar lengths The calculation formula is as follows:

[0089]

[0090] The present invention also relates to a commuting demand-oriented urban road network dynamic traffic zoning system, including a computer module, which applies the above-mentioned commuting demand-oriented urban road network dynamic traffic zoning method.

[0091] Beneficial Effects

[0092] Compared with the prior art, the present invention has the following advantages:

[0093] 1. Simplify traffic zoning content

[0094] The present invention only divides traffic into zones on the road network of traffic management, effectively simplifying the content of traffic zones. This simplification makes traffic zones more focused on the core part of urban traffic management, namely the urban road network, thereby improving the efficiency and practicality of traffic zones.

[0095] By focusing on the road network, the present invention avoids the information redundancy caused by the partition based on building groups in the traditional method, thereby making traffic management more accurate and efficient.

[0096] 2. Accurate traffic zoning for commuting needs

[0097] The present invention combines the application of two community detection algorithms for the first time, affects the partitioning of the road network through the clustering results of commuting-related POIs, and realizes traffic partitioning for special commuting travel needs.

[0098] The present invention ensures that traffic zoning is more targeted in handling commuter traffic, and effectively copes with commuter pressure in the city, especially during peak hours.

[0099] 3. Effective use of dynamic travel data

[0100] The present invention realizes the regularity embodiment of dynamic travel data in traffic partitions by analyzing the geographical location information of the starting point and the end point and the associated road section information in the historical OD travel data.

[0101] The method of the present invention ensures the effectiveness and practicability of the zoning results in actual traffic management, especially in the application of dynamic traffic management and responding to changes in urban traffic demand.

[0102] In general, the innovation of the present invention in the traffic zoning method not only simplifies the zoning content, but also significantly improves the practicality and efficiency of traffic zoning in urban traffic management by accurately targeting commuting needs and effectively using historical OD data. These beneficial effects make the present invention significantly innovative and progressive on the basis of the existing technology, and provide a new perspective and method for urban traffic planning and management. BRIEF DESCRIPTION OF THE DRAWINGS

[0103] Figure 1This is a flow chart of the road network dynamic traffic zoning method of the present invention;

[0104] Figure 2 A schematic diagram of the commuting POI and road network scenario of the present invention;

[0105] Figure 3 This is a schematic diagram of the commuting POI partitioning result of the present invention;

[0106] Figure 4 This is a schematic diagram of the commuting POI and road network partition results of the present invention. DETAILED DESCRIPTION

[0107] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0108] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments may be combined with each other.

[0109] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments, but they are not intended to limit the present invention.

[0110] Step S1: Data preparation and preprocessing

[0111] First, we obtain the city road network data and POI data from the map, and then pre-process all the POI data in the city to filter out the commuting POIs. The processed city road network and commuting POI diagram is as follows: Figure 2 shown.

[0112] Step 1-1: Road network data preparation and preprocessing

[0113] Obtain road network data from the city map, that is, the road segment set R = (r0, r1, ..., r m ). Among them, the attributes contained in the mth road segment are Indicates the unique identifier of the mth road segment; Indicates the geographical location information of the mth road segment, specifically a POLYGON type ((x0,y0),(x1,y1),…,(x j ,y j ))((x j ,y j ) represents the coordinates of the jth vertex on the road segment boundary); Indicates the maximum speed limit value of the mth road section; Indicates the number of lanes contained in the mth road segment; represents the length of the mth road segment; Indicates the unique identifier of the starting intersection of the mth road segment; Indicates the unique identifier of the ending intersection of the mth road segment.

[0114] Step 1-2: Preprocessing of OD travel data during peak hours

[0115] Filter OD travel records. Obtain OD travel records from connected and autonomous vehicles, taxis, online ride-hailing vehicles, and navigation platforms. The record set is represented by T = {t0, t1, …, t l}, l is the record number. Each record t l Filter out the following attribute values in, and Respectively represent the start and end date and time of the lth trip record, and Respectively represent the geographic coordinate information of the start and end of the lth trip record and represents the travel distance of the lth trip record (unit: meter). Then, use the formula By cleaning out invalid records, And the peak hour information of the city is used to filter out the peak hour OD travel records.

[0116] Match the origin and destination segments of OD travel records. And the geographic data of the road segment obtained in step 1-1 Use the shortest Euclidean distance formula from point to polygon to get the identifier of the nearest road segment of the starting point of each OD trip record Similarly, with the help of and the geographic data of the road segment Get the identifier of the nearest road segment to the end point of each OD travel record Finally, update each record t l The attribute value is

[0117] Step 1-3: Commute POI data preparation and preprocessing

[0118] The data preprocessing part first filters out commuting-related POIs from all POIs in the city; then, filters out the data content required for subsequent steps from all the data content of the POI; finally, calculates the data and matrix required for POI clustering.

[0119] Step 1-3-1: Filter and extract attributes of commuting POIs

[0120] Filter commuting POIs. After obtaining the data of all POIs in the current city area on the map (including the unique identifier, geographic coordinates and type of the POI), use the POI type to filter out commuting-related POIs. Among them, the POI types used are amenity and building, which specifically involve the following secondary classifications: office (building.office, building.commercial, amenity.research_institute, amenity.language_school, amenity.courthouse, amenity.hospital, amenity.college, amenity.kindergrarten, amenity.university, amenity.post_office, amenity.school, amenity.bank, ...), residential (building.house, building.residential, building.apartments, ...) and parking (building.parking, ...).

[0121] Determine the basic attributes of commuting POI. Through the filtering in the above steps, we get the commuting POI set P = {p0, p1, ..., p i}, where the attributes of the i-th commuting POI are Represents the unique identifier of the i-th POI, Indicates the longitude and latitude information of the i-th POI (x i ,y i ), Indicates the type of the i-th POI (.

[0122] Improve the commuting POI attributes. With the help of the geographic data of each POI And the geographic data of the road segment obtained in step 1-1 Use the shortest Euclidean distance formula from point to polygon to get the identifier of the nearest road segment for each POI Get further improved POI attributes

[0123] Step 1-3-2: Calculation of the Euclidean distance matrix of commuting POIs

[0124] The geographic coordinates of each commuting POI obtained by step 1-3-1 And use the Euclidean distance formula to calculate the distance between the i-th and j-th commuting POIs

[0125]

[0126] Then, the Euclidean distance between two commuting POIs is converted into a two-dimensional Euclidean distance matrix D eucliden :

[0127]

[0128] Step 1-3-3: Calculation of commuting POI path distance matrix and path segment number matrix

[0129] Calculate the shortest path between commuting POIs. Use the nearest road segment number of each commuting POI obtained in step 1-3-1 And use Dijkstra's shortest path search algorithm to obtain the shortest path between the i-th and j-th commuting POIs

[0130] Calculate the path distance matrix between commuting POIs. Using the length attribute of the road segment Use the following formula to calculate the path between the i-th and j-th commuting POIs: ij Length

[0131]

[0132] in, Indicates path path ij The length of the road segment closest to the k-th POI, represents the arc length of the intersection between the kth to k+1th sections in the path. Then, the path distance between two commuting POIs is converted into a two-dimensional path matrix D path :

[0133]

[0134] Calculate the number of road segments in the path between commuting POIs. Statistics path path ij The number of segments in the middle, and the number of path segments n between the i-th and j-th commuting POIs ij :

[0135]

[0136] Then, the number of sections in the path between two commuting POIs is converted into a two-dimensional path section number matrix N:

[0137]

[0138] Step S2: Commute POI clustering based on community detection algorithm

[0139] Commuting POI clustering based on community detection algorithm first uses steps 1-3 to obtain commuting POI information to build a graph structure, and then uses the community detection Louvain algorithm to cluster commuting POIs on this graph structure data to obtain the classification results of each commuting POI.

[0140] Step 2-1: Graph structuring of commuting POI information

[0141] The input of the community detection algorithm is graph-structured data, so before applying the community detection algorithm to commuting POI clustering, the relevant information needs to be represented by a graph structure. The representation process first establishes the basic graph structure, and then adds the commuting POI information processed in steps 1-3-2 and 1-3-3 to the edge weight.

[0142] Step 2-1-1: Construction of basic graph structure

[0143] Each commuting POI is a node in the graph structure, with the unique identifier of the POI As the node attribute, a connecting edge is established between different commuting POIs to construct the basic graph structure G of commuting POIs poi =(V poi ,E poi ). Among them, V poi represents a node, E poi Represents an edge.

[0144] Step 2-1-2: Calculation of edge composite weight

[0145] Define the composite weight of the edge. The edge E obtained in step 2-1-1 poi In the example, the edge between the i-th and j-th commuting POIs has a composite weight Specifically involved: The Euclidean distance sub-weight between two POIs The sub-weight of the same nearest road segment between two POIs Sub-weight for limiting the number of road segments in the path between two POIs The Jaccard similarity sub-weight between the closest node sets of the closest k-path distance between two POIs

[0146] Calculate the composite weight of the edge. The composite weight of the edge between the i-th and j-th commuting POI nodes The calculation formula is as follows:

[0147]

[0148] Among them, (α eucliden ,α same ,α n ,α J) are the coefficients of the four sub-weights. In order to reflect the importance of different sub-weights, the order is Then the values ​​of the four sub-weight coefficients are set to (α eucliden ,α same ,α n ,α J )=(0.4,0.3,0.2,0.1).

[0149] Calculate the sub-weight of the edge. The Euclidean distance sub-weight of the edge between the i-th and j-th commuting POI nodes The calculation formula is as follows:

[0150]

[0151] Same nearest road segment sub-weight The calculation formula is as follows:

[0152]

[0153] Sub-weight for limiting the number of segments in a path The calculation formula is as follows:

[0154]

[0155] The Jaccard similarity sub-weight between the nearest k-path distance and the closest node set The calculation formula is as follows:

[0156]

[0157] Among them, N i and N j Respectively represent the POI sets with the k closest path distances to the i-th and j-th commuting POIs.

[0158] Step 2-3: Commute POI clustering based on community detection Louvain algorithm

[0159] For the graph structure G obtained in step 2-1 poi =(V poi ,E poi ), the community detection Louvain algorithm is used to cluster commuting POIs. The core of this algorithm is to optimize the community structure by continuously moving each POI node to different communities, achieving modular Q graph structure. poi The goal of maximization. Among them, the modularity Qp of the graph structure oi的 The calculation formula is as follows:

[0160]

[0161] in, Representation graph Gpoi The composite weight sum of all edges in , is the composite weight between the i-th and j-th POI nodes calculated in step 2-1-2, r poi is the clustering resolution parameter (the larger the value, the fewer clusters there are in the clustering result. This parameter should be adjusted according to the specific scenario and requirements). i and k j represents the sum of the weights of all edges associated with the i-th and j-th POI nodes, c i and c j denote the communities to which the i-th and j-th POI nodes are assigned, δ(c i ,c j ) is an indicator function (c i =c j The function value is 1 when , otherwise it is 0).

[0162] When Figure G poi Modular Q poi After maximization, the partition of commuting POI is completed, and the cluster label of each POI is obtained. The partition result diagram is as follows Figure 3 Update the result to the POI attribute set of step 1-3-1 to obtain the attribute set of the i-th commuting POI

[0163] Step 3: Road network partitioning based on community detection algorithm

[0164] Road network partitioning based on community detection algorithm, first use step 1-1 to obtain road network data and step 2-3 to obtain the attributes of each commuting POI Construct a graph structure of road network information. Then, use the community detection Leiden algorithm on the graph structure to partition the road network and obtain the partition results for each road section.

[0165] Step 3-1: Graph structuring of road network information

[0166] The input of the community detection algorithm is graph-structured data. Therefore, before applying the community detection algorithm to partition the urban road network, it is necessary to first represent the historical OD travel data during peak hours, the road network data, and the commuting POI clustering results using a graph structure. The representation process first establishes a basic graph structure, and reflects the starting and ending frequencies of the historical OD travel data during peak hours in the weights of the nodes. Then, the attribute differences between road sections, as well as the commuting POI clustering results and the association between POI and road sections, are reflected in the composite weights of the graph structure edges.

[0167] Step 3-1-1: Determination of the basic diagram

[0168] Determine the node Vedge With the help of the road segment set R = (r0, r1, ..., rm) obtained in step 1-1, each road segment is a node in the graph, and the unique identifier of the road segment is the node attribute, and the node weight is the normalized frequency f of each road segment as the origin and destination road segment in the historical OD travel records during peak hours. m , and its calculation formula is:

[0169]

[0170] Among them, c m is the total frequency of the mth road segment as the starting and ending point in the historical OD travel records during peak hours, max{c0,c1,…,c m} represents the maximum value of the total frequency of all road segments as the above starting and ending point sections.

[0171] Determine edge E edge The starting intersection mark of the m road segment obtained in step 1-1 and end intersection sign Determine the link relationship between each road segment. If there is a link relationship between road segment m and road segment n, an edge is established between the two road segments, and the basic sub-weight of the edge is assigned The calculation formula is as follows:

[0172]

[0173] Step 3-1-2: Calculate the composite weight of the edges in the graph

[0174] Define the composite weight of the edge. The edge E obtained in step 3-1-1 edge In the example, the edge between the mth and nth segments has a composite weight Specifically involved: Basic sub-weight The weight of the nearest neighboring road segment of POI with the same clustering result Sub-weights of roads with the same speed limit Sub-weights of road segments with the same number of lanes Sub-weights of two road segments with similar lengths

[0175] Calculate the composite weight of the edge. The composite weight of the edge between the mth and nth segment nodes The calculation formula is as follows:

[0176]

[0177] Among them, (β b ,β p ,β s ,β lane ,β length) are the coefficients of the five sub-weights. In order to reflect the importance of different sub-weights, the order is Then the values ​​of the five sub-weight coefficients are set to (β b ,β p ,β s ,β lane ,β length )=(0.01,0.699,0.1,0.1,0.1).

[0178] Calculate the sub-weight of the edge. The sub-weight of the nearest neighboring road segment of the POI with the same clustering result The calculation method is: if there are the same clustering results in the POIs associated with two road segments, then Otherwise, it is 0. Sub-weights of roads with the same speed limit The calculation formula is as follows:

[0179]

[0180] Sub-weights of road segments with the same number of lanes The calculation formula is as follows:

[0181]

[0182] Sub-weights of two road segments with similar lengths The calculation formula is as follows:

[0183]

[0184] Step 3-2: Road network partitioning based on community detection Leiden algorithm

[0185] For the graph structure G obtained in step 3-1 edge =(V edge ,E edge ), using the community detection Leiden algorithm that emphasizes connection relationships to partition the road network. The core of this algorithm is to optimize the community structure by continuously moving each road node to different communities, achieving modular Q graph structure. edge The goal of maximizing. Among them, the modularity of the graph structure Q edge The calculation formula is as follows:

[0186]

[0187] in, is the composite weight coefficient of the edge between the mth and nth road segment nodes calculated in step 3-1-2, r edge is the resolution parameter of the partition (the larger the value, the fewer the number of partitions, and this parameter is adjusted according to the specific scenario and requirements), k m and k nrepresents the sum of the weights of all edges associated with the mth and nth road segment nodes, respectively, c m and c n denote the communities to which the mth and nth road segment nodes are assigned, δ(c m ,c n ) is an indicator function (c m =c n The function value is 1 when , otherwise it is 0).

[0188] When Figure G edge Modular G edge After maximization, the partition of the road network is completed, and the partition label of each road section is obtained. Update the result to the segment attribute set in step 1-1 to obtain the attribute set of the mth segment Finally, the dynamic traffic partition of the road network is realized, and the result diagram is as follows Figure 4 shown.

[0189] Although the present invention is described herein with reference to specific embodiments, it should be understood that these embodiments are merely examples of the principles and applications of the present invention. It should therefore be understood that many modifications may be made to the exemplary embodiments and that other arrangements may be devised without departing from the spirit and scope of the present invention as defined by the appended claims. It should be understood that the various dependent claims and features described herein may be combined in a manner different from that described in the original claims. It should also be understood that features described in conjunction with individual embodiments may be used in other described embodiments.

Claims

1. A method for dynamic traffic zoning of urban road networks for commuting needs, characterized in that: The method comprises the following steps: S1. Prepare and preprocess data; S2, clustering commuting POIs based on community detection algorithm; S3, Dynamic traffic zoning of urban road network based on community detection algorithm; S1 specifically includes the following steps: Step 1-1: Road network data preparation and preprocessing Obtain road network data from the city map, that is, the road segment set R = (r0, r1, ..., r m ); Among them, the attributes contained in the mth road segment are Indicates the unique identifier of the mth road segment; Indicates the geographical location information of the mth road segment, specifically a POLYGON type ((x0,y0),(x1,y1),…,(x j ,y j )),(x j ,y j ) represents the coordinates of the jth vertex on the road segment boundary; Indicates the maximum speed limit value of the mth road section; Indicates the number of lanes contained in the mth road segment; represents the length of the mth road segment; Indicates the unique identifier of the starting intersection of the mth road segment; Indicates the unique identifier of the ending intersection of the mth road segment; Step 1-2: Preprocessing of OD travel data during peak hours Filter OD travel records: Get OD travel records. The record set is represented by T = {t0, t1, …, t l }, l is the record number; each record t l Filter out the following attribute values in, and Respectively represent the start and end date and time of the lth trip record, and Respectively represent the geographic coordinate information of the start and end of the lth trip record and represents the travel distance of the lth trip record; then, use the formula Clean out invalid records, using and the peak hour information of the city to filter out the OD travel records during peak hours; Match the origin and destination segments of OD travel records; And the geographic data of the road segment obtained in step 1-1 Use the shortest Euclidean distance formula from point to polygon to get the identifier of the nearest road segment of the starting point of each OD trip record With the help of and the geographic data of the road segment Get the identifier of the nearest road segment to the end point of each OD travel record Update each record l The attribute value is Step 1-3: Commute POI data preparation and preprocessing First, filter out commuting-related POIs from all POIs in the city; then filter out the data content required for subsequent steps from all the data content of the POIs; finally, calculate the data and matrix required for POI clustering; Steps 1-3 specifically include the following steps: Step 1-3-1: Filter and extract attributes of commuting POIs Filter commuting POI: Get data of all POIs in the current city area on the map; Determine the basic attributes of commuting POI: Through the filtering in the above steps, we get the commuting POI set P = {p0, p1, ..., p i }, where the attributes of the i-th commuting POI are Represents the unique identifier of the i-th POI, Indicates the longitude and latitude information of the i-th POI (x i ,y i ), Indicates the type of the i-th POI; Improve commuting POI attributes: with the help of geographical data of each POI And the geographic data of the road segment obtained in step 1-1 Use the shortest Euclidean distance formula from point to polygon to get the identifier of the nearest road segment for each POI Further improved POI attributes Steps 1-3 specifically include the following steps: Step 1-3-2: Calculation of the Euclidean distance matrix of commuting POIs The geographic coordinates of each commuting POI obtained by step 1-3-1 And use the Euclidean distance formula to calculate the distance between the i-th and j-th commuting POIs Then, the Euclidean distance between two commuting POIs is converted into a two-dimensional Euclidean distance matrix D eucliden : Steps 1-3 specifically include the following steps: Step 1-3-3: Calculation of commuting POI path distance matrix and path segment number matrix Calculate the shortest path between commuting POIs: Use the nearest road segment number of each commuting POI obtained in step 1-3-1 And use Dijkstra's shortest path search algorithm to obtain the shortest path between the i-th and j-th commuting POIs Calculate the path distance matrix between commuting POIs: Using the length attribute of the road segment Use the following formula to calculate the path between the i-th and j-th commuting POIs: ij Length in, Indicates path path ij The length of the road segment closest to the k-th POI, represents the arc length of the intersection between the kth to k+1th road segments in the path; then, the path distance between two commuting POIs is converted into a two-dimensional path matrix D path : Calculate the number of road segments in the path between commuting POIs: Statistical path path ij The number of segments in the middle, and the number of path segments n between the i-th and j-th commuting POIs ij : n ij =|path ij | Convert the number of sections in the path between two commuting POIs into a two-dimensional path section number matrix N: S2 specifically includes the following steps: Step 2-1: Graph structuring of commuting POI information Before applying the community detection algorithm to cluster commuting POIs, the relevant information needs to be represented by a graph structure. The representation process first establishes the basic graph structure, and then adds the commuting POI information processed in steps 1-3-2 and 1-3-3 to the edge weight. Step 2-1 specifically includes the following steps: Step 2-1-1: Construction of basic graph structure Each commuting POI is a node in the graph structure, with the unique identifier of the POI As the node attribute, a connecting edge is established between different commuting POIs to construct the basic graph structure G of commuting POIs poi =(V poi ,E poi ), where V poi represents a node, E poi represents an edge; Step 2-1-2: Calculation of edge composite weight Define the composite weight of the edge: edge E obtained in step 2-1-1 poi In the example, the edge between the i-th and j-th commuting POIs has a composite weight Specifically involved: The Euclidean distance sub-weight between two POIs The sub-weight of the same nearest road segment between two POIs Sub-weight for limiting the number of road segments in the path between two POIs The Jaccard similarity sub-weight between the closest node sets of the closest k-path distance between two POIs Calculate the composite weight of the edge: the composite weight of the edge between the i-th and j-th commuting POI nodes The calculation formula is as follows: Among them, (α eucliden ,α same ,α n ,α J ) are the coefficients of the four sub-weights respectively; in order to reflect the importance of different sub-weights, the order is Then set the values ​​of the four sub-weight coefficients to (α eucliden ,α same ,α n ,α J )＝(0.4,0.3,0.2,0.1)； Calculate the sub-weight of the edge: the Euclidean distance sub-weight of the edge between the i-th and j-th commuting POI nodes The calculation formula is as follows: Same nearest road segment sub-weight The calculation formula is as follows: Sub-weight for limiting the number of segments in a path The calculation formula is as follows: The Jaccard similarity sub-weight between the nearest k-path distance and the closest node set The calculation formula is as follows: Among them, N i and N j Respectively represent the POI sets with the closest path distances to the i-th and j-th commuting POIs; Step 2-2: Commute POI clustering based on community detection Louvain algorithm For the graph structure G obtained in step 2-1 poi =(V poi ,E poi ), the community detection Louvain algorithm is used to cluster commuting POIs; the core of this algorithm is to optimize the community structure by continuously moving each POI node to different communities, achieving the modular Q poi The goal of maximization is: poi The calculation formula is as follows: in, Represents the graph structure G poi The composite weight sum of all edges in , is the composite weight between the i-th and j-th POI nodes calculated in step 2-1-2, r poi is the clustering resolution parameter, k i and k j represents the sum of the weights of all edges associated with the i-th and j-th POI nodes, c i and c j denote the communities to which the i-th and j-th POI nodes are assigned, δ(c i ,c j ) is an indicator function, c i =c j The function value is 1 when , otherwise it is 0; When the graph structure G poi Modular Q poi After maximization, the partition of commuting POIs is completed, and the cluster label of each POI is obtained. Update the partition result to the POI attribute set in step 1-3-1 to obtain the attribute set of the i-th commuting POI Step S3 specifically includes the following steps: Step 3-1: Graph structuring of road network information First, the historical OD travel data, road network data and commuting POI clustering results during peak hours are represented by a graph structure. The representation process first establishes a basic graph structure, and reflects the starting and ending frequencies of the historical OD travel data during peak hours in the weights of the nodes. Then, the attribute differences between road sections, as well as the commuting POI clustering results and the association between POI and road sections, are reflected in the composite weights of the edges of the graph structure. Step 3-1 specifically includes: Step 3-1-1: Determine the basic graph, including: Determine the node V edge Information and determination of edge E edge information; Step 3-1-2: Calculate the composite weight of the edges in the graph Define the composite weight of the edge: edge E obtained in step 3-1-1 edge In the example, the edge between the mth and nth segments has a composite weight Specifically involved: Basic sub-weight The weight of the nearest neighboring road segment of POI with the same clustering result Sub-weights of roads with the same speed limit Sub-weights of road segments with the same number of lanes Sub-weights of two road segments with similar lengths Calculate the composite weight of the edge; Step 3-2: Road network partitioning based on community detection Leiden algorithm For the graph structure G obtained in step 3-1 edge =(V edge ,E edge ), using the community detection Leiden algorithm that emphasizes connection relationships to partition the road network; by continuously moving each road node to different communities to optimize the community structure, the modular Q edge The goal of maximization is: edge The calculation formula is as follows: in, is the composite weight coefficient of the edge between the mth and nth road segment nodes calculated in step 3-1-2, r edge is the resolution parameter of the partition, k m and k n represents the sum of the weights of all edges associated with the mth and nth road segment nodes, respectively, c m and c n denote the communities to which the mth and nth road segment nodes are assigned, δ(c m ,c n ) is an indicator function; When the graph structure G edge Modular Q edge After maximization, the partition of the road network is completed, and the partition label of each road section is obtained. Update the result to the road segment attribute set in step 1-1 to obtain the attribute set of the mth road segment. Ultimately, dynamic traffic zoning of the road network can be achieved.

2. The method for dynamic traffic zoning of urban road networks for commuting needs according to claim 1 is characterized in that: In step 3-1-1, determine the node V edge The information includes: the road segment set R = (r0, r1, ..., r m ), each road segment is a node in the graph, and the unique identifier of the road segment is the node attribute, and the node weight is the normalized frequency f of each road segment as the origin and destination road segment in the historical OD travel records during peak hours. m , and its calculation formula is: in, is the total frequency of the mth road segment as the starting and ending point in the historical OD travel records during peak hours, Indicates the maximum value of the total frequency of all road sections as the above starting and ending point sections; Determine edge E edge The information includes: the starting intersection mark of the m road segment obtained in step 1-1 and end intersection sign Determine the link relationship between each road segment; if there is a link relationship between road segment m and road segment n, an edge is established between the two road segments, and the basic sub-weight of the edge is assigned The calculation formula is as follows: In step 3-1-2, calculate the composite weight of the edge: the composite weight of the edge between the mth and nth segment nodes The calculation formula is as follows: Among them, (β b ,β p ,β s ,β lane ,β length ) are the coefficients of the five sub-weights respectively; in order to reflect the importance of different sub-weights, the order is Then the values ​​of the five sub-weight coefficients are set to (β b ,β p ,β s ,β lane ,β length )=(0.001,0.699,0.1,0.1,0.1); Calculate the sub-weight of the edge: the sub-weight of the nearest neighbor segment of the POI with the same clustering result The calculation method is: if there are the same clustering results in the POIs associated with two road segments, then Otherwise, it is 0; the sub-weight of the road section with the same speed limit The calculation formula is as follows: Sub-weights of road segments with the same number of lanes The calculation formula is as follows: Sub-weights of two road segments with similar lengths The calculation formula is as follows:

3. A dynamic traffic zoning system for urban road networks oriented to commuting needs, characterized in that: It comprises a computer module, which applies the urban road network dynamic traffic zoning method for commuting needs as described in any one of claims 1 to 2.

Images