A method for identifying key transfer stations in AGV material delivery networks based on heterogeneous graphs

By constructing a heterogeneous graph and employing a graph compression strategy to calculate betweenness centrality, the problem that the Brandes algorithm cannot be applied to heterogeneous graphs is solved, achieving the effect of efficiently identifying important transfer stations in AGV material delivery networks.

CN116777333BActive Publication Date: 2026-11-17SHANDONG UNIV +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202310589170.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-05-24
Publication Date
2026-11-17
Estimated Expiration
2043-05-24

AI Technical Summary

Technical Problem

The existing Brandes algorithm cannot be directly applied to the calculation of BC values ​​of vertices in heterogeneous graphs and the identification of important transfer stations, resulting in the inability to accurately identify important transfer stations in the AGV material delivery network.

Method used

By constructing a heterogeneous graph and defining meta-paths, we employ a basic algorithm, a graph compression strategy that removes edge vertices, and a graph compression strategy that merges consistent vertex sets to calculate the betweenness centrality of vertices in the heterogeneous graph and identify important transit points.

Benefits of technology

It enables accurate identification of key transfer stations in large-scale AGV material delivery networks, improving computational efficiency and reducing computation time from 20 hours to 0.5 hours.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN116777333B_ABST
    Figure CN116777333B_ABST
Patent Text Reader

Abstract

This application belongs to the field of data mining technology, and specifically relates to a method for identifying important transfer stations in AGV material delivery networks based on heterogeneous graphs. The method includes the following steps: S1. Collecting data from a real AGV material delivery network and constructing a heterogeneous graph containing vertices of different categories; S2. Selecting meta-paths in the heterogeneous graph constructed in step S1, defining the connectivity between vertices of the same type in the heterogeneous graph; S3. Calculating the betweenness centrality of vertices in the heterogeneous graph, whereby the betweenness centrality is used to measure the importance of vertices; S4. Selecting the k vertices with the highest betweenness centrality as important transfer stations in the AGV material delivery network. The method provided by this invention defines the BC value of vertices in a heterogeneous graph based on meta-paths, effectively solving the problem of measuring the importance of vertices in heterogeneous graphs, and can accurately identify transfer stations that have a significant impact on material delivery in heterogeneous AGV material delivery networks.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the field of data mining technology, and specifically relates to a method for identifying important transfer stations in AGV material distribution networks based on heterogeneous graphs. Background Technology

[0002] Graphs, as a form of data organization, are frequently used in various fields to model complex relationships between objects. A graph contains two elements: vertices and edges. Vertices represent real-world entities, and edges represent relationships between entities. Based on the number of types of vertices and edges, graphs can be divided into two categories: homogeneous graphs and heterogeneous graphs. Homogeneous graphs contain only one type of vertex and one type of edge, while heterogeneous graphs contain multiple types of vertices and edges. Real-world AGV material delivery networks typically include different types of entities such as shippers, receivers, transit stations, and warehouses, as well as different types of relationships between them. Heterogeneous graphs can be used for modeling and analysis, containing a wealth of valuable information. Therefore, the analysis and research of heterogeneous graphs have significant practical implications.

[0003] The importance of vertices is a key research topic in heterogeneous graphs. There are many measures of vertex importance, such as degree centrality, betweenness centrality, and compact centrality. Among these, betweenness centrality (BC value) is an important indicator of the global influence of vertices in a graph. Vertices with high BC values ​​are important transit points for connecting different vertices in the graph. Using betweenness centrality, important transit stations can be identified in AGV material delivery networks. The larger the BC value of a transit station, the greater the impact of its failure on the smooth delivery of materials throughout the network, indicating that the transit station is important. Therefore, using betweenness centrality to measure the importance of vertices in heterogeneous graphs is a worthwhile research topic.

[0004] A classic algorithm for calculating the BC value of vertices in a graph is the Brandes algorithm. The Brandes algorithm consists of two phases: the first phase uses breadth-first search to find the shortest paths from the source vertex *s* to other vertices; the second phase accumulates the source dependencies of *s* on other vertices *v*. Finally, by summing the source dependencies of all vertices on vertex *v*, the BC value of vertex *v* is obtained. However, because heterogeneous graphs contain vertices of different types, the Brandes algorithm cannot be directly applied to heterogeneous graphs, nor can it be used to calculate the BC values ​​of vertices in heterogeneous graphs or to identify important transfer stations in AGV material delivery networks. Summary of the Invention

[0005] To address the aforementioned problems, this invention provides a method for identifying important transfer stations in AGV material delivery networks based on heterogeneous graphs. This method defines the BC value of vertices in the heterogeneous graph based on metapaths, effectively solving the problem of measuring vertex importance in heterogeneous graphs and accurately identifying transfer stations that have a significant impact on material transport in heterogeneous AGV material delivery networks.

[0006] This invention is achieved through the following technical solution:

[0007] A method for identifying important transfer stations in an AGV material delivery network based on heterogeneous graphs includes the following steps:

[0008] S1. Collect data from real AGV material delivery networks and construct a heterogeneous graph containing vertices of different categories;

[0009] S2. Select meta-paths in the heterogeneous graph constructed in step S1 to specify the connectivity between vertices of the same type in the heterogeneous graph;

[0010] S3. Calculate the betweenness centrality of the vertices in the heterogeneous graph, whereby the betweenness centrality is used to measure the importance of the vertices;

[0011] S4. Select several vertices with the highest intermediary centrality as important transfer stations in the AGV material delivery network.

[0012] Furthermore, step S1 specifically includes:

[0013] Data is collected from real AGV material delivery networks, and the AGV material delivery network is modeled as a heterogeneous graph.

[0014]

[0015] Where V is the set of vertices, E is the set of edges, A is the set of vertex types, and R is the set of edge types. A mapping from each vertex to its corresponding vertex type is defined. The mapping from each edge to its corresponding edge type is defined;

[0016] Using the heterogeneous graph pattern T G = (A, R) represents all vertex types and the types of edges between vertex types in a heterogeneous graph.

[0017] Furthermore, in step S2, the heterogeneous graph pattern T is used. G Metapath Metapath represents the connectivity between similar vertices in a heterogeneous graph. The specific path corresponding to a heterogeneous graph is called a meta-path. Path instance.

[0018] Among them, metapath Among them, A1…A l+1 Indicates the type of the vertices, R1…R l Indicates the type of edge between vertices, where l is the metapath. The number of edge types.

[0019] Furthermore, in step S3, the calculation of the betweenness centrality of vertices in the heterogeneous graph is performed using any one of the following three methods:

[0020] (1) Calculate the betweenness centrality of all vertices using the basic algorithm;

[0021] (2) Use a graph compression strategy that removes edge vertices to accelerate the calculation of vertex betweenness centrality;

[0022] (3) Use a graph compression strategy that merges consistent vertex sets to accelerate the computation of vertex betweenness centrality.

[0023] Furthermore, the calculation of the betweenness centrality of all vertices using the basic algorithm specifically includes the following steps:

[0024] S3.1. Initialize the betweenness centrality metric of each vertex in the heterogeneous graph, so that the initial BC value of each vertex is 0, and store it in the BC array;

[0025] S3.2. Traverse each vertex in the heterogeneous graph and calculate the betweenness centrality metric for each vertex. The calculation steps are as follows:

[0026] S3.2.1. Using any vertex in the heterogeneous graph as the source vertex s, perform a breadth-first search to calculate the number of shortest paths σ[v] from the source vertex s to other vertices v in the heterogeneous graph: where other vertices v represent any vertex in the heterogeneous graph that the source vertex s can reach through breadth-first search.

[0027]

[0028] Where Pred(v) represents the set of preceding vertices of vertex v in the breadth-first search process, vertex u is the preceding vertex of vertex v, and σ[u] represents the number of shortest paths from source vertex s to vertex u. It is the number of path instances between vertices u and v; Obtained synchronously during the breadth-first search process;

[0029] During the breadth-first search process, the predecessor vertex of vertex v is recorded in the Pred(v) set; and the access order of vertices during the breadth-first search is stored in a temporary stack structure S.

[0030] S3.2.2. Traverse the vertices in reverse order of the vertices visited during breadth-first search, pop the vertex from the temporary stack structure S and take it as vertex w. Update the source dependency δ(v) of the source vertex s on any vertex v in the set of preceding vertices Pred(w) of vertex w using the following formula:

[0031]

[0032] In the formula, w:v∈Pred(w) means traversing all vertices w that contain vertex v in the previous vertex set; σ[w] represents the number of path instances between vertices v and w; σ[w] represents the number of shortest paths from source vertex s to vertex w; δ(w) represents the source dependency of source vertex s on vertex w.

[0033] When vertex w is not vertex s, add δ(w) to BC[w].

[0034] S3.2.3. Confirm whether all vertices have been traversed. If not, take the next vertex in the heterogeneous graph as the source vertex s and proceed to step S3.2.1 until all vertices in the heterogeneous graph have been traversed.

[0035] Furthermore, the graph compression strategy of removing edge vertices to accelerate the calculation of vertex betweenness centrality specifically involves the following steps:

[0036] (1) Traverse the heterogeneous graph, find all edge vertices in the heterogeneous graph, and perform the following operations for each edge vertex i:

[0037] Calculate the source dependency δ of edge vertex i on other vertices v. i (v):

[0038] Using vertex i as the source point, perform a breadth-first search according to the formula. Calculate the number σ of shortest paths from vertex i to other vertices v. i [v];

[0039] Where Pred(v) represents the set of preceding vertices of vertex v in the breadth-first search process, vertex u is the preceding vertex of vertex v, and σ i [u] represents the number of shortest paths from vertex i to vertex u. It is the number of path instances between vertices u and v; Obtained synchronously during the breadth-first search process;

[0040] During the breadth-first search process, the predecessor vertex of vertex v is recorded in the Pred(v) set; and the access order of vertices during the breadth-first search is stored in a temporary stack structure S.

[0041] ii) Traverse the vertices in reverse order of the breadth-first search, pop each vertex from the stack S as vertex w, and then use the formula... Update the source dependency δ of any vertex v in the set of predecessor vertices Pred(w) of vertex i to vertex w. i (v):

[0042] w:v∈Pred(w) means traversing all vertices w in the previous vertex set that contain vertex v; σ represents the number of path instances between vertices v and w; i [w] represents the number of shortest paths from vertex i to vertex w; δ i (w) represents the source dependency of edge vertex i on vertex w;

[0043] ②Add 2×δ to the BC value BC[v] of any vertex v. i (v);

[0044] ③ Remove edge vertex i from the heterogeneous graph, including vertex i and the edges adjacent to vertex i.

[0045] (2) After step (1) is completed, if a new edge vertex appears in the heterogeneous graph, return to step (1); otherwise, proceed to step (3).

[0046] (3) On the heterogeneous graph where edge vertices have been removed, perform step S3.2:

[0047] Iterate through each vertex, calculate the source dependency δ(v) of each vertex on every other vertex v, and accumulate it into BC[v].

[0048] Furthermore, a graph compression strategy of merging consistent vertex sets is used to accelerate the computation of vertex betweenness centrality. The specific steps are as follows:

[0049] (1) Identify all consistent vertex sets in the heterogeneous graph;

[0050] (2) Merge each consistent vertex set Z into a representative vertex x, that is, remove all vertices except x from the heterogeneous graph in the consistent vertex set Z, and record the number of vertices in the consistent vertex set Z in icount[x], which represents how many vertices a vertex x represents;

[0051] (3) After the processing in step (2) is completed, if a new set of consistent vertices appears in the heterogeneous graph, then step (1) is executed; otherwise, step (4) is executed.

[0052] (4) On the heterogeneous graph that has merged consistent vertices, traverse each vertex and calculate the betweenness centrality metric of the vertex. The calculation steps are as follows:

[0053] ① On a heterogeneous graph that has merged consistent vertices, traverse each vertex and calculate the betweenness centrality metric of the vertex. The calculation steps are as follows:

[0054] ① Using any vertex s in a heterogeneous graph that has merged consistent vertices as the source vertex, perform a breadth-first search according to the formula. Calculate the number of shortest paths σ[v] from the source vertex s to other vertices v;

[0055] Pred(v) represents the set of preceding vertices of vertex v in the breadth-first search process, vertex u is the preceding vertex of vertex v, and σ[u] represents the number of shortest paths from source vertex s to vertex u. It is the number of path instances between vertices u and v, and icount[u] represents the number of consistent vertices represented by vertex u; Obtained synchronously during the breadth-first search process;

[0056] During the breadth-first search process, the predecessor vertex of vertex v is recorded in the Pred(v) set; and the access order of vertices during the breadth-first search is stored in a temporary stack structure S.

[0057] ② Traverse the vertices in reverse order of the vertices visited during breadth-first search, pop the vertex from the temporary stack structure S and take it as vertex w. Update the source dependency δ(v) of the source vertex s on any vertex v in the set of predecessor vertices Pred(w) of vertex w using the following formula:

[0058]

[0059] icount[w] represents the number of consistent vertices represented by vertex w; δ(w)

[0060] When vertex w is not the source vertex s, for w and each vertex wi represented by w, add δ(w)×icount[w] to BC[w] and BC[wi].

[0061] Further, in step S4, the vertices in the heterogeneous graph are sorted from largest to smallest according to their betweenness centrality metric (BC value). The top k vertices with the largest betweenness centrality metric values ​​are selected, and the transfer stations corresponding to these selected top k vertices in the AGV material delivery network are designated as important transfer stations. Specifically, the user can choose the value of k according to their needs, i.e., select k important transfer stations.

[0062] Beneficial technical effects of the present invention:

[0063] (1) The method provided by the present invention can calculate the accurate BC value of all vertices in the heterogeneous graph. By measuring the intermediate centrality in the heterogeneous graph, the importance of the vertices in the heterogeneous network can be fully measured, and the transfer stations that have a significant impact on the material delivery in the heterogeneous network of AGV material distribution can be accurately identified.

[0064] (2) The method provided by this invention offers two graph compression strategies, which greatly improves the running efficiency of the algorithm, enabling the calculation of the BC value of vertices and the identification of important transfer stations even in large-scale heterogeneous AGV material delivery networks.

[0065] (3) The method provided by the present invention adopts an efficient optimization calculation strategy to accelerate the calculation of vertex intermediate centrality on heterogeneous graphs, so that the important transfer station identification method proposed in the present invention can be applied to large-scale AGV material distribution networks. Through experimental verification on large datasets, compared with the basic method, the optimization strategy can achieve an acceleration effect of 0.5 hours from 20 hours on large datasets. Attached Figure Description

[0066] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below:

[0067] Figure 1 This is a schematic diagram of the overall process of an AGV material distribution network important transfer station identification method based on heterogeneous graphs in an embodiment of the present invention.

[0068] Figure 2 This is a schematic diagram illustrating the process of using a basic algorithm to calculate the BC value of a vertex in a heterogeneous graph and identify important transfer stations in an embodiment of the present invention.

[0069] Figure 3 This is a schematic diagram illustrating the process of using an optimization strategy to accelerate the calculation of vertex BC values ​​in a heterogeneous graph and identify important transfer stations in an embodiment of the present invention.

[0070] Figure 4 Example diagram for identifying important transfer stations in an AGV material delivery network using the method of an embodiment of the present invention. Detailed Implementation

[0071] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0072] Conversely, this invention encompasses any substitutions, modifications, equivalent methods, and solutions made within the spirit and scope of the invention as defined in the claims. Furthermore, to provide a better understanding of the invention, certain specific details are described in detail below. However, those skilled in the art will fully understand the invention even without these detailed descriptions.

[0073] This invention proposes a method for identifying important transfer stations in AGV material delivery networks based on heterogeneous graphs, such as... Figure 1 As shown, the specific implementation is as follows:

[0074] S1. Collect data from a real AGV material delivery network and construct a heterogeneous graph containing vertices of different categories:

[0075] Data is collected from real AGV material delivery networks, and the AGV material delivery network is modeled as a heterogeneous graph.

[0076]

[0077] Where V is the set of vertices, E is the set of edges, A is the set of vertex types, and R is the set of edge types. A mapping from each vertex to its corresponding vertex type is defined. The mapping from each edge to its corresponding edge type is defined;

[0078] Using the heterogeneous graph pattern T G = (A, R) represents all vertex types in the heterogeneous graph and the types of edges between vertex types. In this embodiment, vertex types include recipient, sender, transit station, and warehouse.

[0079] like Figure 4 As shown, real-world AGV material delivery networks typically include different types of entities such as shippers, receivers, transit stations, and warehouses, as well as different types of relationships between these entities. Therefore, it is necessary to construct a heterogeneous graph containing different types of vertices. The heterogeneous graph pattern can store all vertex types in the heterogeneous graph and the types of edges between these vertex types. The heterogeneous graph stores all entities in the heterogeneous network and the edges that exist between them; each type of edge is stored in the form of an adjacency matrix.

[0080] S2. In the heterogeneous graph constructed in step S1, meta-paths are selected to specify the connectivity between vertices of the same type in the heterogeneous graph:

[0081] Use the definition in the heterogeneous graph pattern T G Metapath Metapath represents the connectivity between similar vertices in a heterogeneous graph. The specific path corresponding to a heterogeneous graph is called a meta-path. Path instance.

[0082] Among them, metapath Among them, A1…A l+1 Indicates the type of the vertices, R1…R l Indicates the type of edge between vertices, where l is the metapath. The number of edge types.

[0083] Since vertices of the same type in heterogeneous graphs often lack edges and cannot be directly connected, connections between them typically require vertices of other types as intermediaries. This is achieved using meta-paths. The connectivity between vertices can be specified. Metapaths are defined on heterogeneous graph patterns and are a sequence of vertex and edge types. The specific path corresponding to a metapath on a heterogeneous graph is called a metapath. A path instance, if vertex u is traversed via a metapath If a path instance can reach vertex v, then vertex u and vertex v are said to be... Neighbor.

[0084] S3. Calculate the BC value of the vertices in the heterogeneous graph:

[0085] The betweenness centrality metric of a vertex identifies the sum of the proportions of shortest paths passing through that vertex in the shortest paths between each pair of vertices in the graph. It can be used to measure the importance of that vertex pair in achieving connectivity with other vertices in the network. The betweenness centrality of vertices in the heterogeneous graph can be calculated using any of the following three methods:

[0086] (1) Calculate the betweenness centrality of all vertices using the basic algorithm;

[0087] (2) Use a graph compression strategy that removes edge vertices to accelerate the calculation of vertex betweenness centrality;

[0088] (3) Use a graph compression strategy that merges consistent vertex sets to accelerate the computation of vertex betweenness centrality.

[0089] like Figure 2 As shown, the calculation of the betweenness centrality of all vertices using the basic algorithm specifically includes the following steps:

[0090] S3.1. Initialize the betweenness centrality metric of each vertex in the heterogeneous graph, so that the initial BC value of each vertex is 0, and store it in the BC array; wait for the update in the subsequent calculation process, where BC[v] represents the betweenness centrality metric of vertex v.

[0091] S3.2. Traverse each vertex in the heterogeneous graph and calculate the betweenness centrality metric for each vertex. The calculation steps are as follows:

[0092] S3.2.1. Using any vertex in the heterogeneous graph as the source vertex s, perform a breadth-first search to calculate the number of shortest paths σ[v] from the source vertex s to other vertices v in the heterogeneous graph:

[0093]

[0094] Where Pred(v) represents the set of preceding vertices of vertex v in the breadth-first search process, vertex u is the preceding vertex of vertex v, and σ[u] represents the number of shortest paths from source vertex s to vertex u. It is the number of path instances between vertices u and v; Obtained synchronously during the breadth-first search process;

[0095] During the breadth-first search process, the predecessor vertex of vertex v is recorded in the Pred(v) set; and the access order of vertices during the breadth-first search is stored in a temporary stack structure S.

[0096] S3.2.2. Traverse the vertices in reverse order of the vertices visited during breadth-first search, pop the vertex from the temporary stack structure S and take it as vertex w. Update the source dependency δ(v) of the source vertex s on any vertex v in the set of preceding vertices Pred(w) of vertex w using the following formula:

[0097]

[0098] In the formula, w:v∈Pred(w) means traversing all vertices w that contain vertex v in the previous vertex set; σ[w] represents the number of path instances between vertices v and w; σ[w] represents the number of shortest paths from source vertex s to vertex w; δ(w) represents the source dependency of source vertex s on vertex w; δ(w) and δ(v) are calculated using the same logic. In fact, w and v in the formula represent vertices with predecessor and successor relationships in the calculation process, rather than referring to a specific vertex.

[0099] In the formula above In the middle, replace v with w, and replace w with the vertex with w as the preceding vertex, and you can calculate δ(w);

[0100] When vertex w is not vertex s, δ(w) is added to the BC value BC[w] of vertex w;

[0101] S3.2.3. Confirm whether all vertices have been traversed. If not, take the next vertex in the heterogeneous graph as the source vertex s and proceed to step S3.2.1 until all vertices in the heterogeneous graph have been traversed.

[0102] like Figure 3As shown, based on S3, two graph compression strategies are used to accelerate the calculation of the betweenness centrality of vertices in heterogeneous graphs: one is the strategy of removing edge vertices, and the other is the strategy of merging consistent vertex sets.

[0103] The graph compression strategy of removing edge vertices accelerates the computation of vertex betweenness centrality. This optimization strategy removes edge vertices in the heterogeneous graph that are not traversed by a shortest path, thus achieving graph compression. An edge vertex is a vertex that is not traversed by a shortest path, meaning its BC value is 0. Therefore, edge vertices can be removed from the heterogeneous graph to achieve graph compression. The specific steps are as follows:

[0104] (1) Traverse the heterogeneous graph, find all edge vertices in the heterogeneous graph, and perform the following operations for each edge vertex i:

[0105] ① Calculate the source dependency δ of edge vertex i on other vertices v. i (v):

[0106] i) Perform a breadth-first search using the edge vertex i as the source point, according to the formula Calculate the number σ of shortest paths from vertex i to other vertices v. i [v];

[0107] Where Pred(v) represents the set of preceding vertices of vertex v in the breadth-first search process, vertex u is the preceding vertex of vertex v, and σ i [u] represents the number of shortest paths from vertex i to vertex u. It is the number of path instances between vertices u and v; Obtained synchronously during the breadth-first search process;

[0108] During the breadth-first search process, the predecessor vertex of vertex v is recorded in the Pred(v) set; and the access order of vertices during the breadth-first search is stored in a temporary stack structure S.

[0109] ii) Traverse the vertices in reverse order of the breadth-first search, pop each vertex from the stack S as vertex w, and then use the formula... Update the source dependency δ of any vertex v in the set of predecessor vertices Pred(w) of vertex i to vertex w. i (v):

[0110] w:v∈Pred(w) means traversing all vertices w in the previous vertex set that contain vertex v; σ represents the number of path instances between vertices v and w; i [w] represents the number of shortest paths from vertex i to vertex w; δ i (w) represents the source dependency of edge vertex i on vertex w;

[0111] δi(w) and δi(v) follow the same calculation approach. In fact, w and v in the formulas represent vertices with predecessor and successor relationships in the calculation process, rather than referring to any specific vertex. (See the formula above.) In the equation, replace v with w, and replace w with the vertex that has w as its preceding vertex, and you can calculate the result.

[0112] ②Add 2×δ to the BC value BC[v] of any vertex v. i (v);

[0113] ③ Remove edge vertex i from the heterogeneous graph, including vertex i and the edges adjacent to vertex i;

[0114] (2) Because after removing the edge vertex in (1), it is the same as the removed edge vertex. A neighboring vertex may become a new edge vertex. Therefore, traverse these vertices. If a new edge vertex appears in the heterogeneous graph at this time, execute (1); otherwise, execute (3).

[0115] (3) On the heterogeneous graph with the edge vertices removed, perform S3.2 (i.e., step S3.2 in the basic algorithm described above):

[0116] Traverse each vertex in the heterogeneous graph and calculate the betweenness centrality metric for each vertex. The calculation steps are as follows:

[0117] S3.2.1. Using any vertex in the heterogeneous graph as the source vertex s, perform a breadth-first search to calculate the number of shortest paths σ[v] from the source vertex s to other vertices v in the heterogeneous graph: where other vertices v represent any vertex in the heterogeneous graph that the source vertex s can reach through breadth-first search.

[0118]

[0119] Where Pred(v) represents the set of preceding vertices of vertex v in the breadth-first search process, vertex u is the preceding vertex of vertex v, and σ[u] represents the number of shortest paths from source vertex s to vertex u. It is the number of path instances between vertices u and v; Obtained synchronously during the breadth-first search process;

[0120] During the breadth-first search process, the predecessor vertex of vertex v is recorded in the Pred(v) set; and the access order of vertices during the breadth-first search is stored in a temporary stack structure S.

[0121] S3.2.2. Traverse the vertices in reverse order of the vertices visited during breadth-first search, pop the vertex from the temporary stack structure S and take it as vertex w. Update the source dependency δ(v) of the source vertex s on any vertex v in the set of preceding vertices Pred(w) of vertex w using the following formula:

[0122]

[0123] In the formula, w:v∈Pred(w) means traversing all vertices w that contain vertex v in the previous vertex set; σ[w] represents the number of path instances between vertices v and w; σ[w] represents the number of shortest paths from source vertex s to vertex w; δ(w) represents the source dependency of source vertex s on vertex w.

[0124] When vertex w is not vertex s, add δ(w) to BC[w].

[0125] S3.2.3. Confirm whether all vertices have been traversed. If not, take the next vertex in the heterogeneous graph as the source vertex s and proceed to step S3.2.1. Continue traversing all vertices in the heterogeneous graph. For each vertex, calculate the source dependency δ(v) of each vertex on each other vertex v and accumulate it in the array BC[v].

[0126] Two vertices being consistent means that their neighborhood information is the same, that is, the two vertices have the same neighbors. When two vertices are consistent, their BC values ​​are equal, so they can be merged into one vertex for computation, achieving graph compression. Vertices in the same consistent vertex set are consistent with each other, so all vertices in a consistent vertex set Z can be merged into a representative vertex x, and x represents all vertices in Z for subsequent computation.

[0127] The graph compression strategy of merging consistent vertex sets is used to accelerate the computation of vertex betweenness centrality. The specific steps are as follows:

[0128] (1) Identify all consistent vertex sets in the heterogeneous graph.

[0129] (2) Merge each consistent vertex set Z into a representative vertex x, that is, remove all vertices except x from the heterogeneous graph in the consistent vertex set Z, and record the number of vertices in the consistent vertex set Z in icount[x], which records how many vertices a representative vertex x represents.

[0130] (3) Because after merging the set of consistent vertices, the common neighbors of the consistent vertices may become new consistent vertices. Traverse these vertices, and if there is a new set of consistent vertices in the heterogeneous graph at this time, execute (1); otherwise execute (4).

[0131] (4) On the heterogeneous graph with merged consistent vertices, traverse each vertex s and calculate the betweenness centrality metric of the vertex. The calculation steps are as follows:

[0132] ① Using any vertex s in a heterogeneous graph that has merged consistent vertices as the source vertex, perform a breadth-first search according to the formula. Calculate the number of shortest paths σ[v] from the source vertex s to other vertices v;

[0133] Pred(v) represents the set of preceding vertices of vertex v in the breadth-first search process, vertex u is the preceding vertex of vertex v, and σ[u] represents the number of shortest paths from source vertex s to vertex u. It is the number of path instances between vertices u and v, and icount[u] represents the number of consistent vertices represented by vertex u; Obtained synchronously during the breadth-first search process;

[0134] During the breadth-first search process, the predecessor vertex of vertex v is recorded in the Pred(v) set; and the access order of vertices during the breadth-first search is stored in a temporary stack structure S.

[0135] ② Traverse the vertices in reverse order of the vertices visited during breadth-first search, pop the vertex from the temporary stack structure S and take it as vertex w. Update the source dependency δ(v) of the source vertex s on any vertex v in the set of predecessor vertices Pred(w) of vertex w using the following formula:

[0136]

[0137] icount[w] represents the number of consistent vertices represented by vertex w, as above.

[0138] δ(w) and δ(v) follow the same calculation approach. In fact, w and v in the formula represent vertices that have a predecessor and successor relationship in the calculation process, rather than referring to a specific vertex.

[0139] In the formula above In the equation, replace v with w, and replace w with the vertex that has w as its preceding vertex, and you can calculate the result.

[0140] When w is not s, for w and each vertex wi represented by w, add δ(w)×icount[w] to BC[w] and BC[wi].

[0141] S4. Acquire an important transit point:

[0142] (1) Sort the vertices in the heterogeneous graph from largest to smallest according to their BC values.

[0143] (2) Select the k vertices with the largest BC values. The transfer stations corresponding to these vertices in the AGV material delivery network are the important transfer stations. Specifically, users can choose the value of k according to their needs, that is, select k important transfer stations.

[0144] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for identifying important transfer stations in an AGV material distribution network based on a heterogeneous graph, characterized in that, Includes the following steps: S1. Collect data from real AGV material delivery networks and construct a heterogeneous graph containing vertices of different categories; S2. Select meta-paths in the heterogeneous graph constructed in step S1 to specify the connectivity between vertices of the same type in the heterogeneous graph; S3. Calculate the betweenness centrality of the vertices in the heterogeneous graph, whereby the betweenness centrality is used to measure the importance of the vertices; The calculation of the betweenness centrality of vertices in the heterogeneous graph is performed using any one of the following three methods: (1) Calculate the betweenness centrality of all vertices using the basic algorithm; (2) Use a graph compression strategy that removes edge vertices to accelerate the computation of vertex betweenness centrality; (3) Use a graph compression strategy that merges consistent vertex sets to accelerate the computation of vertex betweenness centrality; The graph compression strategy of merging consistent vertex sets is used to accelerate the computation of vertex betweenness centrality. The specific steps are as follows: (1) Identify all consistent vertex sets in the heterogeneous graph; (2) merge each consistent vertex set Z into one representative vertex x, i.e. delete all vertices in the isomorphic graph except x in the consistent vertex set Z, and record the number of vertices in the consistent vertex set Z in wherein indicates how many vertices are represented by one vertex x; (3) After the processing in step (2) is completed, if a new set of consistent vertices appears in the heterogeneous graph, then step (1) is executed; otherwise, step (4) is executed. (4) On the heterogeneous graph that has merged consistent vertices, traverse each vertex and calculate the betweenness centrality metric of the vertex. The calculation steps are as follows: ① Using any vertex s in a heterogeneous graph that has merged consistent vertices as the source vertex, perform a breadth-first search according to the formula. Calculate the number of shortest paths from source vertex s to other vertices v. ; This represents the set of preceding vertices of vertex v in the breadth-first search process, where vertex u is a preceding vertex of vertex v. This represents the number of shortest paths from source vertex s to vertex u. It is the metapath between vertices u and v. The number of path instances, This indicates the number of consistent vertices represented by vertex u; Obtained synchronously during the breadth-first search process; During the breadth-first search process, the predecessor vertex of vertex v is recorded in... The set is stored in a temporary stack structure S, which stores the order in which vertices are visited during the breadth-first search. ② Traverse the vertices in reverse order of the vertices visited during breadth-first search, pop the vertex from the temporary stack structure S, and use it as vertex w. Update the set of the previous vertices of vertex w from the source vertex s using the following formula. Source dependencies of any vertex v : ; This indicates the number of consistent vertices represented by vertex w; This indicates the source dependency of source vertex s on vertex w; When vertex w is not the source vertex s, for w and each vertex wi represented by w, accumulate in BC[w] and BC[wi]. ,in This represents the number of consistent vertices represented by the source vertex s; S4. Select several vertices with the highest intermediary centrality as important transfer stations in the AGV material delivery network.

2. The method for identifying important transfer stations in an AGV material distribution network based on heterogeneous graphs according to claim 1, characterized in that, Step S1 is as follows: Data is collected from real AGV material delivery networks, and the AGV material delivery network is modeled as a heterogeneous graph. ; in, It is a set of vertices. It is a set of edges. It is a collection of vertex types. It is a collection of edge types. A mapping from each vertex to its corresponding vertex type is defined. The mapping from each edge to its corresponding edge type is defined; Use heterogeneous graph pattern This represents the type of all vertices in the heterogeneous graph and the type of edges between vertices.

3. The method for identifying important transfer stations in an AGV material distribution network based on heterogeneous graphs according to claim 2, characterized in that, In step S2, the heterogeneous graph pattern is defined. Metapath Metapath represents the connectivity between similar vertices in a heterogeneous graph. The specific path corresponding to a heterogeneous graph is called a meta-path. Path instance.

4. The method for identifying important transfer stations in an AGV material distribution network based on heterogeneous graphs according to claim 1, characterized in that, The calculation of the intermediate centrality of all vertices using the basic algorithm specifically includes the following steps: S3.

1. Initialize the betweenness centrality metric of each vertex in the heterogeneous graph, so that the initial BC value of each vertex is 0, and store it in the BC array; S3.

2. Traverse each vertex in the heterogeneous graph and calculate the betweenness centrality metric for each vertex. The calculation steps are as follows: S3.2.

1. Using any vertex in the heterogeneous graph as the source vertex s, perform a breadth-first search to calculate the number of shortest paths from the source vertex s to the other vertices v in the heterogeneous graph. Vertex v is the source vertex s via the metapath The vertices connected by the path instance; ; in, This represents the set of preceding vertices of vertex v in the breadth-first search process, where vertex u is a preceding vertex of vertex v. This represents the number of shortest paths from source vertex s to vertex u. It is the metapath between vertices u and v. The number of path instances; Obtained synchronously during the breadth-first search process; During the breadth-first search process, the predecessor vertex of vertex v is recorded in... The set is stored in a temporary stack structure S, which stores the order in which vertices are visited during the breadth-first search. S3.2.

2. Traverse the vertices in reverse order of visiting vertices during breadth-first search, pop each vertex from the temporary stack structure S as vertex w, and update the set of preceding vertices of vertex w from source vertex s using the following formula. Source dependencies of any vertex v : ; In the formula, This indicates traversing all vertices w that include vertex v in the preceding vertex set; Represents the metapath between vertices v and w. The number of path instances; This represents the number of shortest paths from source vertex s to vertex w; This indicates the source dependency of source vertex s on vertex w; When vertex w is not the source vertex s, the BC value BC[w] of vertex w is accumulated. ; S3.2.

3. Confirm whether all vertices have been traversed. If not, take the next vertex in the heterogeneous graph as the source vertex s and proceed to step S3.2.1 until all vertices in the heterogeneous graph have been traversed.

5. The method for identifying important transfer stations in an AGV material distribution network based on heterogeneous graphs according to claim 4, characterized in that, The graph compression strategy of removing edge vertices to accelerate the calculation of vertex betweenness centrality involves the following steps: (1) Traverse the heterogeneous graph, find all edge vertices in the heterogeneous graph, and perform the following operations for each edge vertex i: ① Calculate the source dependency of edge vertex i on other vertices v. : i) Perform a breadth-first search using the edge vertex i as the source point, according to the formula Calculate the number of shortest paths from vertex i to other vertices v. ; in, This represents the set of preceding vertices of vertex v in the breadth-first search process, where vertex u is a preceding vertex of vertex v. This represents the number of shortest paths from vertex i to vertex u. It is the metapath between vertices u and v. The number of path instances; Obtained synchronously during the breadth-first search process; During the breadth-first search process, the predecessor vertex of vertex v is recorded in... The set is stored in a temporary stack structure S, which stores the order in which vertices are visited during the breadth-first search. ii) Traverse the vertices in reverse order of the breadth-first search, pop each vertex from the stack S as vertex w, and then use the formula... Update the set of preceding vertices of vertex i to vertex w. Source dependencies of any vertex v : This indicates traversing all vertices w that include vertex v in the preceding vertex set; Represents the metapath between vertices v and w. The number of path instances; This represents the number of shortest paths from vertex i to vertex w. This represents the source dependency of vertex i on vertex w; ② Accumulate the BC value BC[v] of any vertex v. ; ③ Remove edge vertex i from the heterogeneous graph, including vertex i and the edges adjacent to vertex i; (2) After step (1) is completed, if a new edge vertex appears in the heterogeneous graph, return to step (1); otherwise, proceed to step (3). (3) On the heterogeneous graph where the edge vertices have been removed, perform step S3.2; Iterate through each vertex and calculate the source dependency of each vertex on every other vertex v. And add it to BC[v].

6. The method for identifying important transfer stations in an AGV material distribution network based on heterogeneous graphs according to claim 1, characterized in that, In step S4, the vertices in the heterogeneous graph are sorted from largest to smallest according to their betweenness centrality metric (i.e., BC value). The top k vertices with the largest betweenness centrality metric are selected, and the transfer stations corresponding to the selected top k vertices with the largest betweenness centrality metric in the AGV material distribution network are designated as important transfer stations.