A signal control unit division method under multi-element data fusion

Abstract

The application discloses a signal control unit division method under multi-element data fusion, which comprises the following steps: step one, setting a road network topology structure to convert a controllable area static road network into the road network topology structure; step two, statically layering the road network and taking the static road network as the first, second, third and fifth layers of road network; step three, dividing a dynamic road network and taking the dynamic road network as the fourth layer of road network; and step four, dividing a signal control unit, determining a control area node set node according to a selected control area, dividing the node set, and obtaining a final signal control unit set. The division method fuses the static road network and the dynamic road network, disperses signal control function loads to different units for processing, and thus improves the concurrent processing capacity and response speed of the system.

Description

Technical Field

[0001] This invention relates to the fields of secure digital information transmission and machine learning. Specifically, it relates to a method for dividing information control units (ICUs) under multi-source data fusion. Background Technology

[0002] Existing methods for dividing information control units (ICUs) mainly fall into two categories: the first is based on the division of static and dynamic data, and the second is based on business scenarios and requirements. The method based on static and dynamic data primarily divides ICUs according to different data attributes, generally processing static and dynamic data separately, with less attention paid to division approaches under multi-source data fusion.

[0003] As the amount and types of data in the field of intelligent transportation increase, and the real-time performance of data also improves significantly, conditions are provided for the division of multi-source data information control units under the fusion of static and dynamic data. Summary of the Invention

[0004] The present invention is proposed based on the above-mentioned situation of the prior art. The technical problem to be solved by the present invention is to provide a method for dividing information control units under multi-source data fusion, which can realize the accurate division and dynamic adjustment of information control units under source data fusion.

[0005] To solve the above problems, the present invention is implemented using the following technical solution:

[0006] A method for dividing traffic control units under multi-source data fusion is provided, including step one: setting a road network topology to convert the static road network of the controllable area into a road network topology. The conversion rules for setting the road network topology include: node design rules, edge design rules, surface design rules, attribute design rules, and database design rules. Step two: statically layering the road network and using the static road network as the first, second, third, and fifth layers of the road network. Here, the signal light parameter bool_TL∈{0,1} is defined, where 0 represents a non-signal-controlled intersection and 1 represents a signal-controlled intersection; the road level set parameter road_level∈{expressway, arterial road, secondary arterial road, local road} is defined. When bool_TL=0&road_level=expressway, this part of the road is divided into the first layer of the road network; the extended roads associated with the first layer of the road network and the auxiliary roads corresponding to expressways are divided into the second layer of the road network; according to the congestion delay index formula... Jam index This indicates that the index is the congestion delay index corresponding to the road in the index, t. jam t represents the travel time during congestion periods. freeThe travel time during the free flow period is represented by the number of roads with a duration of 2 or more, which are selected as the third layer of the road network. Roads that are spatially associated with the first, second, and third layers of the road network and thus serve to receive traffic flow within the first, second, and third layers of the road network are classified as the fifth layer of the road network. Step 3: Divide the dynamic road network and use it as the fourth layer of the road network. This includes: S301 Collecting traffic flow data, adding vehicle flow, flow distribution, and traffic direction as edge weights to the road network topology to form a traffic flow map; S302 Developing a dynamic road network division strategy based on the traffic flow map and changes in traffic conditions, calculating a similarity matrix using traffic flow information and performing cluster analysis to minimize the dispersion of traffic points within each cluster; S303 Dividing the road network in real time according to the division strategy determined in S302, and associating each division unit with the corresponding traffic flow data; S304 Updating the status information of the division units based on the real-time collected traffic flow data, and repeating steps S302 and S303 to determine the latest road network division strategy and division results, thus forming the fourth layer of the road network. Step 4: Divide the signal control units. Based on the selected control area, determine the set of control area nodes (node) and divide the node set to obtain the final set of signal control units. The different types of area nodes are denoted as node_type, including {Level 1 signal control nodes, Level 2 signal control nodes, Level 3 signal control nodes, merging and diverging nodes for expressways and highways, and merging and diverging nodes for urban roads}. Level 1 signal control nodes have signal control equipment and are located at key intersections of major commuter roads within the area during peak hours. Level 2 signal control nodes have signal control equipment and are located at key intersections of secondary commuter roads within the area during peak hours, providing moderate traffic flow management during peak hours. Level 3 signal control nodes have signal control equipment and are located at key intersections of non-major / secondary commuter roads within the area during peak hours. Merging and diverging nodes for expressways and highways, and merging and diverging nodes for urban roads, are located at the entrances and exits of urban roads and expressway / highway ramps within the area during peak hours.

[0007] Preferably, step four includes: S401 defining node labels as flag∈{0,1}, where 0 represents a node not yet partitioned and 1 represents a node already partitioned; initializing all nodes n with flag==0, and selecting the starting point of the control unit set; the starting point selection satisfies the first condition: first-level nodes > second-level nodes > third-level nodes, and the second condition: nodes with high traffic saturation > nodes with smooth traffic; S402 traversing each data point n, if the label flag of n is ==1, then skipping the node; otherwise, finding the neighbor node neighbor(n) of node n, with the formula:

[0008] neighbor(n) = {m|m∈node, distance(m,n)≤r}, where r is the neighborhood radius of data point n, node m is included within the radius of the neighborhood (neighbor(n), and distance(m,n) represents the distance between points m and n; in S403, the parameter Max is set. If the number of data points in the neighborhood (neighbor(n)) is greater than or equal to Max, then the data point n corresponding to the neighborhood (neighbor(n)) is used as the starting point and added to the starting point selected in step S401; a new cluster sub is created for this starting point. n And add n and its density-reachable points to the cluster; S404 recursively processes all points in the new cluster, that is, for each point k∈sub n If the label flag of k is 0, then add it to the new cluster and continue to find the neighbor(k) of k to expand the control unit set; repeat steps S402 to S404 in S404, traverse the data points in the control area, and put the traversed node into the control unit according to the condition. When the traversal end condition is met or the label flag of all nodes in the area is 1, the control unit set is expanded and the final control unit is generated.

[0009] Preferably, the conditions for placing the traversed node into the control unit in step S404 include being adjacent to an intersection in the control unit, having a correlation coefficient between intersections greater than a threshold, and having a number of nodes in the control unit less than a limit.

[0010] Preferably, step S302 includes: first, randomly initializing the road network regions, denoting one region as A and another region as B, where A∈net, B∈net, and net is the entire set of road network regions; then, using the formula mat(B)=P -1 mat(A)P calculates the similarity matrix P between the flow matrices of regions A and B, where mat(A) is the flow matrix of region A, mat(B) is the flow matrix of region B, and P is the similarity matrix; then, the formula P = QΛQ is used. T Perform spectral decomposition to decompose the similarity matrix P obtained above into eigenvectors and eigenvalues, where Q is an orthogonal matrix composed of eigenvectors of similarity matrix P, and Λ is a diagonal matrix composed of eigenvalues ​​of matrix P; then perform eigenvector clustering on the obtained Q matrix; and repeat the above steps until the dispersion of the partitioning results tends to be stable and the sum is minimized, and then end the iteration to obtain the partitioning strategy.

[0011] Preferably, the eigenvector clustering of the obtained Q matrix includes: preliminary clustering through agglomerative hierarchical clustering, followed by further subdivision clustering using k-means; the agglomerative hierarchical clustering divides region A∩B into several regions according to road level, while k-means further subdivides the regions, with the clustering process aiming to minimize intra-cluster dispersion; where dispersion is represented by variance, as shown in the following formula: Where, x i Let represent the i-th data point, μ represent the mean of the data, and N represent the total number of data points. Repeat the above steps until the dispersion of the partitioning results stabilizes and the sum is minimized. The iteration ends, yielding the final partitioning strategy.

[0012] Preferably, the node design rules include: the node set expression is node∈{intersection, expressway / urban road (diverging / merging), pedestrian crossing, boundary node}, where node represents the node set; the edge design rules include: the edge set expression is edge∈{urban road, expressway, connecting road segment}, where edge represents the edge set; the surface design rules include: the connection between node and edge is used as the design of zone, the surface set expression is zone∈{shopping mall, hospital, office area, school...}, where zone represents the surface set.

[0013] Preferably, the attribute design rules include designing attribute information for nodes, edges, and faces, adding some attribute information to nodes, edges, or faces, including information such as {road length, number of lanes, speed limit}.

[0014] Preferably, the attribute design rules further include treating the divergence / merging points as child nodes (sub_node), and extending the unique attribute (sub_attribute) on the basis of inheriting the original node attributes, where sub_attribute = {divergence direction, downstream merging point number, merging segment}.

[0015] Preferably, the database design rules include designing a corresponding database model, including a node table, an edge table, and a zone table.

[0016] Preferably, layers one, two, three, and five are static road networks, and layer four is a dynamic road network obtained by loading dynamic data and combining the static three-layer and five-layer road networks.

[0017] This partitioning method can integrate static and dynamic road networks, and distribute the signal control function load to different units for processing, thereby improving the system's concurrent processing capability and response speed. In addition, by partitioning the signal control units, different signal control strategies and parameter configurations can be adopted for different service types, realizing differentiated service management. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of this specification or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the embodiments of this specification. For those skilled in the art, other drawings can be obtained based on these drawings.

[0019] Figure 1 This is a flowchart of a method for dividing information control units under multi-source data fusion provided in a specific embodiment of the present invention;

[0020] Figure 2 This is a first-level topology diagram of the road network topology in a signal control unit partitioning method under multi-data fusion provided by a specific embodiment of the present invention;

[0021] Figure 3 This is a schematic diagram of the surface design rules for the road network topology in a signal control unit partitioning method under multi-data fusion provided by a specific embodiment of the present invention;

[0022] Figure 4 This is a flowchart of road network division in a signal control unit division method under multi-data fusion provided by a specific embodiment of the present invention;

[0023] Figure 5 This is a schematic diagram of the fourth-layer dynamic road network in a signal control unit partitioning method under multi-data fusion provided by a specific embodiment of the present invention;

[0024] Figure 6 This is a schematic diagram illustrating the relationship between the five layers of the road network in a signal control unit division method under multi-source data fusion provided by a specific embodiment of the present invention;

[0025] Figure 7 This is a flowchart of the information control unit division method under multi-source data fusion provided in a specific embodiment of the present invention. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0027] To facilitate understanding of the embodiments of the present invention, further explanations and descriptions will be provided below with reference to the accompanying drawings and specific embodiments. These embodiments do not constitute a limitation on the scope of protection of the present invention.

[0028] This specific implementation includes a signal control unit (SCU) partitioning method under multi-source data fusion. This method starts from the road network topology and partitions SCUs under multi-source data fusion to better leverage the advantages of traffic big data in the field of intelligent transportation and achieve accurate partitioning and dynamic adjustment of SCUs.

[0029] The steps of the information control unit division method under multi-data fusion are as follows: Figure 1 As shown, specifically, it includes:

[0030] Step 1: Set up the road network topology

[0031] Road network topology refers to the structure composed of relationships between various elements (such as nodes, edges, and faces) in a road network, which are usually represented graphically. It describes the connection methods, directions, and topological attributes between different elements in the road network. In this specific embodiment, to make the road network structure clearer and facilitate data partitioning and observation, this embodiment provides a road network topology suitable for the signal control unit partitioning method under multi-data fusion involved in this embodiment, so as to convert the static road network of controllable areas into a topological structure, providing support for subsequent static and dynamic area partitioning of the road network.

[0032] In this specific embodiment, the conversion rules for the road network topology setting include:

[0033] Node design rules: Nodes refer to intersections, turning points, etc., in a road network, and there are interconnected relationships between nodes. Specifically, in this specific implementation, the node type expression is node∈{intersection, expressway / urban road (diverging / merging), pedestrian crossing, boundary node}, where node represents a set of nodes.

[0034] Edge design rules: An edge refers to a road segment in a road network. Edges should have a clear start and end point and should be directional. In this specific implementation, the edge type expression is edge∈{urban roads, expressways, connecting road segments}, where edge represents the set of edges.

[0035] Surface design rules: A surface represents a road grid within a region, such as a city block. Since the relationship between zones, nodes, and edges needs to be considered, the connection between nodes and edges is added to the zone design, defined as a virtual parking lot, with the expression:

[0036] zone∈{shopping mall, hospital, office area, school...}, where zone represents a set of faces, mainly areas with high traffic volume.

[0037] Attribute Design Rules: In this specific implementation, for different application scenarios, some attribute information can be added to nodes, edges, or faces, including...

[0038] Information such as road length, number of lanes, and speed limit is used for analysis and calculation. Attribute information is designed for custom nodes, edges, and surfaces. Divestment / merge points are treated as child nodes (sub_node), inheriting the original node attributes and extending them with unique attributes (sub_attribute), where sub_attribute = {divergence direction, downstream merge point number, merge segment}.

[0039] Database design rules: In order to store and manage road network topology data, it is necessary to design a corresponding database model, including node table, edge table, zone table, etc. At the same time, it is necessary to consider indexing, data compression and other technologies to improve data query and processing efficiency.

[0040] The road network topology is designed based on the above method. The first-level topology includes edges and nodes, as illustrated below. Figure 2 From a macro perspective, the road network structure is divided into road segments and nodes. From a micro perspective, due to the large volume of traffic entering and exiting densely populated areas such as residential communities, shopping malls, schools, and hospitals around the road network, these areas can be uniformly divided into virtual parking lots. Only the entrance and exit roads are retained as connecting segments, connecting to their respective corresponding road segments, thus relating to the macro road network structure, such as... Figure 3 As shown. Based on this, the nodes, road segments, and enclosed surfaces involved are combined to form a two-level topology.

[0041] Based on the transformation rules designed above, database storage fields are designed for edges, nodes, and faces. The database field design for edges is shown in the table below.

[0042] Based on the transformation rules of the design, the database storage fields for edges, nodes and faces are designed. The edge_table database field for edges is designed as shown in the table below.

[0043]

[0044] The node_table database field for nodes is designed differently from that for edges, adding fields such as node location (latitude and longitude) and whether there are traffic lights, as shown in the table below.

[0045]

[0046]

[0047] The area enclosed by nodes and edges is defined as a virtual parking lot, also known as a virtual community. The corresponding database field zone_table is designed as shown in the table below.

[0048]

[0049] Step 2: Statically divide the road network into layers, and use the static road network as the first, second, third, and fifth layers of the road network.

[0050] Static layering of the road network is the process of spatially dividing the road network according to predetermined rules and methods to form spatial units, and associating or allocating various information in the road network, such as traffic flow, speed, and traffic restrictions, with these spatial units.

[0051] In this specific implementation, a graph-based segmentation method is used to perform hierarchical screening of the overall road network from the perspective of road hierarchy, establishing a five-layer road network. The critical paths in each layer are selected as static road networks, with layers one, two, three, and five being static road networks. The fourth layer is a dynamic road network obtained by loading dynamic data and combining the static third and fifth layers. The first four layers are superimposed as the macroscopic road network segmentation result. The segmentation method is as follows:

[0052] S201 is divided into a single-level road network.

[0053] Define traffic light parameters

[0054] bool_TL∈{0,1}, where 0 represents a non-signal-controlled intersection and 1 represents a signal-controlled intersection; define the road level set parameter road_level∈{expressway, arterial road, secondary arterial road, local road}, and when bool_TL=0&road_level=expressway, this part of the road is divided into a level of road network.

[0055] S202 is divided into a two-level road network.

[0056] Among them, the extension roads and auxiliary roads corresponding to expressways that are related to the first-level road network will be classified into the second-level road network.

[0057] S203 is divided into a three-level road network.

[0058] The third layer of the road network is defined by the frequency of congestion on main roads during morning and evening rush hours, specifically using a congestion delay index, as shown in the following formula:

[0059]

[0060] Jam index This indicates that the index is the congestion delay index corresponding to the road in the index, t. jamt represents the travel time during congestion periods. free This indicates the travel time during the free-flow period. When jam index ≥4 defines severe traffic congestion; when 2≤jam index <4, define road congestion; when 1.5 ≤ jam index <2, define slow-moving roads. Based on the congestion delay index, select roads with a value greater than or equal to 2 as the third layer of the road network.

[0061] S204 is divided into the fifth level of road network.

[0062] Roads that are spatially connected to the first, second, and third road networks, and thus serve to receive traffic flow within the first three road networks, are classified as the fifth road network.

[0063] Based on the above steps, taking Beijing as an example, each layer of the road network includes the following roads: Layer 1: Expressway and Expressway Network. This includes expressways and expressways within the Fifth Ring Road and intersecting with them. These are the main commuting routes within and outside Beijing during weekday morning and evening rush hours and are unsignaled. Traffic volume is high, and average speed is fast. Layer 2: Extended and Auxiliary Road Network. This includes extended sections of expressways and expressways converted into urban roads, and auxiliary roads for expressways. Building upon the first layer, this includes extended sections of expressways and expressways, as well as auxiliary roads for expressways and ring roads, handling the main commuting traffic within and outside Beijing during weekday morning and evening rush hours. These sections have signal control equipment, and traffic volume is high, with relatively fast average speed. Layer 3: Main Inbound and Outbound Road Network. This adds main inbound and outbound roads, which are arterial roads and the main commuting routes within the Fifth Ring Road during weekday morning and evening rush hours. These roads have signal control equipment. Traffic volume is moderate to high, average speed is moderate, and congestion is common. The fifth level comprises all connecting roads in the road network. It includes roads that are spatially connected to the first three levels and have good connectivity. The fifth level road network is formed by merging roads from bottom to top. Any road that crosses or connects the areas defined by the first three levels, effectively linking areas and transporting traffic, is included in the fifth level road network. The fifth level road network represents the finest division. It includes secondary arterial roads, local roads, and other secondary roads.

[0064] Step 3: Divide the road network into dynamic layers and use it as the fourth layer of the road network.

[0065] Dynamic road network partitioning is a time- and space-based method that divides the road network in real time according to changes in traffic flow. It can be used in applications such as real-time traffic monitoring and congestion management. Unlike static partitioning, dynamic partitioning can reflect the changing process of traffic conditions more precisely.

[0066] The division of the fourth-layer road network is based on the processing of the third-layer and fifth-layer road network data and dynamic data to obtain critical paths, and critical paths not displayed on the third-layer road network are completed.

[0067] The specific steps for obtaining the critical path include:

[0068] S301 collects traffic flow data

[0069] Traffic flow data at different locations on the road is acquired through traffic monitoring equipment and vehicle identification systems. Traffic flow maps are then created based on urban route flow data and intersection flow data. Specifically, after the road network topology design is completed, vehicle flow, flow distribution, and traffic direction on a specific road, intersection, or road segment are added as edge weights to form a traffic flow map for analysis and evaluation of traffic conditions.

[0070] S302 Determine the partitioning strategy

[0071] like Figure 4 As shown, a strategy for dynamic road network partitioning is formulated based on traffic flow maps and changes in traffic conditions. A similarity matrix is ​​calculated using traffic flow information and cluster analysis is performed to minimize the dispersion of traffic points within each cluster.

[0072] First, randomly initialize the road network regions, denoting one region as A and another as B. Then A ∈ net, B ∈ net, where net is the entire set of road network regions. Calculate the similarity matrix P between the traffic matrices corresponding to regions A and B, using the following formula:

[0073] mat(B)=P -1 mat(A)P

[0074] Where mat(A) is the flow matrix of region A, mat(B) is the flow matrix of region B, and P is the similarity matrix.

[0075] Next, spectral decomposition is performed to decompose the similarity matrix P obtained above into eigenvectors and eigenvalues, as shown in the following formula:

[0076] P = QΛQ T

[0077] Here, Q is an orthogonal matrix composed of the eigenvectors of the similarity matrix P, and Λ is a diagonal matrix composed of the eigenvalues ​​of matrix P. The next step is to perform eigenvector clustering on the obtained Q matrix.

[0078] A two-level clustering method is preferred, which involves initial clustering using agglomerative hierarchical clustering, followed by further subdivision using k-means clustering to achieve better clustering results. Agglomerative hierarchical clustering divides region A∩B into several regions according to road level, while k-means further subdivides these regions. The clustering process aims to minimize intra-cluster dispersion. Dispersion is represented by variance, as shown in the following formula:

[0079]

[0080] Where, x i Let represent the i-th data point, μ represent the mean of the data, and N represent the total number of data points. Repeat the above steps until the dispersion of the partitioning results stabilizes and the sum is minimized. The iteration ends, yielding the final partitioning strategy.

[0081] S303 road network division

[0082] The road network is divided in real time according to the division strategy determined in S302, and each division unit is associated with the corresponding traffic flow data.

[0083] S304 Updated Partition Information

[0084] Based on the real-time traffic flow data, update the status information of the segmented units, and repeat the above steps S302 and S303 to determine the latest road network segmentation strategy and segmentation results.

[0085] Taking Beijing as an example, such as Figure 6 As shown in the previous step, the static road network was divided into five layers. The fourth layer is a dynamic road network, derived from the data of the third and fifth layers, along with dynamic data. Based on available traffic data and the division strategy, traffic flow data for the morning and evening peak hours were selected for dynamic road network division. The selected data identified the top 50 roads with the highest traffic flow during peak hours citywide as critical commuting routes. Combining the strategies of the third and fifth layers, based on the morning and evening peak hour data, if a critical path is not in the third layer but is included in the fifth layer, it is added to the third layer to form the fourth layer road network. For details on how the fourth layer road network was derived, please refer to [link to relevant documentation]. Figure 5 illustrate.

[0086] Step 4: Divide the signal control units

[0087] In this step, such as Figure 7As shown, based on the selected control area, a set of control area nodes (node) is determined. Following a specific process, this node set is divided to obtain the final set of signal control and control units, which can be categorized into single-point, trunk, regional, and critical nodes. The different types of regional nodes are denoted as `node_type`, including {Level 1 signal control node, Level 2 signal control node, Level 3 signal control node, merging and diverging nodes for expressways and highways, and merging and diverging nodes for urban roads}.

[0088] Level 1 signal control nodes have signal control equipment and are mainly located at key intersections of major commuter roads within the area during peak hours, managing heavy traffic flow during peak hours. Level 2 signal control nodes have signal control equipment and are mainly located at key intersections of secondary commuter roads within the area during peak hours, managing moderate traffic flow during peak hours. Level 3 signal control nodes have signal control equipment and are mainly located at key intersections of non-major / secondary commuter roads within the area during peak hours, managing less traffic flow during peak hours. Finally, there are signal control devices at expressway / urban road merging / diversion points, mainly located at the entrances and exits of urban roads and expressway / highway ramps within the area during peak hours, primarily controlling ramp entrance and exit traffic flow.

[0089] Let the total number of nodes in the control region be l, and define the node set as node, where n ∈ node, and n is a node in the control region. Based on the set partitioning condition, the final control unit can be obtained as {n1, n2, ... n}. i}、{n i ,n i+1 ,…n b}、…、{n b ,n b+1 ,…n l}. The subscript of the letter 'n' indicates the node number.

[0090] The specific division process is as follows:

[0091] S401 defines node labels as flag∈{0,1}, where 0 represents a node that has not been partitioned, and 1 represents a node that has been partitioned. Initialize flag == 0 for all nodes n, and select the starting point for the control unit set. The selection of the starting point has two priority conditions that must be met: Level 1 nodes > Level 2 nodes > Level 3 nodes; high traffic saturation > nodes with smooth traffic.

[0092] S402 iterates through each data point n. If the label flag of n is 1, then skip that node; otherwise, find the neighbor node neighbor(n) of node n, as shown in the formula:

[0093] neighbor(n)={m|m∈node,distance(m,n)≤r}

[0094] Where r is the neighborhood radius of data point n, node m is included within the radius of the neighborhood (neighbor(n), and distance(m,n) represents the distance between points m and n;

[0095] S403 sets the parameter Max. If the number of data points in the neighborhood (n) is greater than or equal to Max, then the data point n corresponding to the neighborhood (n) is used as the starting point and added to the starting point selected in step S401. A new cluster sub is created for this starting point. n And add n and the points whose density is reachable to the cluster;

[0096] S404 recursively processes all points in the new cluster, that is, for each point k∈sub n If the label flag of k is 0, then add it to the new cluster and continue to find the neighbor(k) of k to expand the set of control units.

[0097] S404 repeats steps S402 to S404, traversing the data points in the control area. Based on certain conditions, each traversed node is placed into its corresponding control unit. When the traversal ends or the flag of all nodes in the area is equal to 1, the control unit set is expanded, generating the final control unit. The conditions for placing a traversed node into a control unit include: being adjacent to an intersection within that control unit; having a correlation coefficient between intersections greater than a threshold; having a control unit node count less than a limit; and having a strong correlation with traffic flow within the same category.

[0098] Taking Xueyuan Road in Haidian District, Beijing as an example, the morning rush hour nodes are divided as follows: the red line represents the main traffic flow, and the red nodes are level 1 traffic control nodes; the yellow line represents the secondary traffic flow, and the yellow nodes are level 2 traffic control nodes; all other nodes, except for the merging / diverting nodes of expressways / urban roads, are classified as level 3 traffic control nodes.

[0099] The signal control unit division results can be obtained according to the above division process, as shown in the figure. Among them, high-level control nodes (first and second-level signal control nodes) implement corresponding control strategies for trunk lines in different areas according to the road network division. Key road continuous control nodes adopt coordinated control strategies. If the third-level signal control nodes are divided into the same area, fixed timing control or regional coordinated control can be adopted.

[0100] This partitioning method distributes the signal control workload across different units, thereby improving the system's concurrent processing capabilities and response speed. Furthermore, different business scenarios may require different signal control strategies and rules. By partitioning into signal control units, different signal control strategies and parameter configurations can be applied to different business types, achieving differentiated business management. In summary, the purpose of signal control unit partitioning is to improve system performance, achieve scalability, enhance system stability, and support differentiated business management through effective partitioning and management, thereby meeting diverse business needs and optimizing system operation.

[0101] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for dividing information control units under multi-source data fusion, characterized in that, The method includes: Step 1: Set up a road network topology to convert the static road network in the controllable area into a road network topology. The conversion rules for setting up the road network topology include: node design rules, edge design rules, surface design rules, attribute design rules, and database design rules. Step 2: Statically layer the road network, designating the static road network as layers 1, 2, 3, and 5. Define the traffic light parameters. 0 indicates a non-signal-controlled intersection, and 1 indicates a signal-controlled intersection; define the road level set parameters. When satisfied The road network is divided into two layers: the first layer consists of roads connected to the first layer and auxiliary roads corresponding to expressways; these are then divided into the second layer. The congestion delay index formula is used to further classify these roads. ,in This indicates that the index represents the congestion delay index corresponding to the road specified by index. Indicates travel time during congestion periods. The travel time during the free flow period is represented by the number of roads with a duration of 2 or more, which are selected as the third layer of the road network. Roads that are spatially associated with the first, second, and third layers of the road network and thus serve to receive traffic flow within the first, second, and third layers of the road network are classified as the fifth layer of the road network. Step 3: Divide the dynamic road network and use it as the fourth layer of the road network. This includes: S301 Collecting traffic flow data, adding vehicle flow, flow distribution, and traffic direction as edge weights to the road network topology to form a traffic flow map; S302 Formulating a dynamic road network division strategy based on the traffic flow map and changes in traffic conditions, calculating a similarity matrix using traffic flow information and performing cluster analysis to minimize the dispersion of traffic points within each cluster; S303 Dividing the road network in real time according to the division strategy determined in S302, and associating each division unit with the corresponding traffic flow data; S304 Updating the status information of the division units based on the real-time collected traffic flow data, and repeating steps S302 and S3023 to determine the latest road network division strategy and division results, thus forming the fourth layer of the road network. Among these, layers one, two, three, and five are static road networks, and the fourth layer is a dynamic road network obtained by loading dynamic data and combining the static third and fifth layer road networks. Step S302 includes, First, the road network areas are randomly initialized, and one area is denoted as... The other area is ,but , It is the entire road network area; through the formula Computational area and Similarity matrix of corresponding flow matrices ,in It is a region Traffic matrix It is a region Traffic matrix It is a similarity matrix; then, use the formula Perform spectral decomposition calculations and convert the obtained similarity matrix into a spectral matrix. Decomposed into eigenvectors and eigenvalues, where, It is a similar matrix An orthogonal matrix formed by the eigenvectors of . It is a matrix The diagonal matrix is ​​constructed from the eigenvalues; then the obtained... The matrix is ​​used for eigenvector clustering; and the above steps are repeated until the dispersion of the partitioning results tends to be stable and the sum is minimized, and then the iteration ends to obtain the partitioning strategy; The obtained Matrix-based eigenvector clustering includes initial clustering using agglomerative hierarchical clustering, followed by further subdivision using k-means clustering; agglomerative hierarchical clustering divides regions... According to road classification The clustering process divides the data into several regions, and k-means further subdivides these regions, aiming to minimize the intra-cluster dispersion. This dispersion is represented by variance, as shown in the following formula: in, Indicates the first Data points, This represents the mean of the data. The total number of data points is represented by the above steps. The process is repeated until the dispersion of the partitioning results tends to be stable and the sum is minimized. The iteration ends and the final partitioning strategy is obtained. Step 4: Divide the signal control units Based on the selected control region, determine the set of control region nodes (node) and divide the node set to obtain the final set of signal control units. The different types of region nodes are denoted as follows: ,Include Level 1 signal control nodes have signal control equipment and are located at key intersections of major commuter roads within the area during peak hours; Level 2 signal control nodes have signal control equipment and are located at key intersections of secondary commuter roads within the area during peak hours, providing moderate traffic flow during peak hours; Level 3 signal control nodes have signal control equipment and are located at key intersections of non-major / secondary commuter roads within the area during peak hours; merging and diverging nodes for expressways and urban roads are located at the entrances and exits of urban roads and expressway / highway ramps within the area during peak hours; Step four includes, S401 defines node labels as follows 0 represents nodes that have not been partitioned, and 1 represents nodes that have been partitioned; initialize all nodes. of And select the starting point of the control unit set; the starting point is selected to meet the first condition: first-level node > second-level node > third-level node, and the second condition: high traffic saturation > smooth traffic. S402 iterates through each data point ,if tags If the node is found, skip it; otherwise, find the node. Neighboring nodes The formula is: in Data points neighborhood radius, node Included in the neighborhood Within the radius, This represents the distance between two points, m and n. S403 setting parameters If the neighboring region The data points contained are greater than or equal to Then the neighborhood Corresponding data points As a starting point, it is added to the starting point selected in step S401; a new cluster is created for this starting point. and will Points whose density is reachable are added to the cluster; S404 recursively processes all points in the new cluster, that is, for each point... ,if tags If so, add it to the new cluster and continue searching. neighborhood This achieves the goal of expanding the set of control units; S404 Repeat steps S402 to S404, traversing the data points in the control area, and placing the traversed nodes into their respective control units according to the conditions. The traversal ends when the end condition is met or the labels of all nodes in the area are satisfied. At that time, the control unit set is expanded and the final control unit is generated.

2. The method for dividing information control units under multi-source data fusion according to claim 1, characterized in that, The conditions for placing a traversed node into the control unit in step S404 include being adjacent to an intersection in the control unit, having a correlation coefficient between intersections greater than a threshold, and having a number of nodes in the control unit less than a limit.

3. The method for dividing information control units under multi-source data fusion according to claim 1, characterized in that, The node design rules include the following: the node set expression is... ,in Represents a set of nodes; The edge design rules include the following: the edge set expression is... Represents the set of edges; Surface design rules include, and The connection as The design, the face set expression is , Represents a set of faces.

4. The method for dividing information control units under multi-source data fusion according to claim 3, characterized in that, The attribute design rules include designing attribute information for nodes, edges, and faces, and adding attribute information to nodes, edges, or faces. ,Include information.

5. The method for dividing information control units under multi-source data fusion according to claim 4, characterized in that, The attribute design rules also include treating the branching / merging points as child nodes. In addition to inheriting the original node attributes, it also expands the unique attributes. ,in .

6. The method for dividing information control units under multi-source data fusion according to claim 5, characterized in that, Database design rules include designing the corresponding database model, including node tables. Side table Surface .

Images