A disaster point network community discovery method and device fusing geographical space constraints

Abstract

The application discloses a disaster point network community discovery method and device fusing geographical space constraints, and the method steps comprise the following steps: step S01, acquiring geographical position data and a topological connection relationship matrix of each disaster point in a disaster point network; step S02, calculating a spatial similarity matrix between each disaster point, and fusing and calculating a spatial connection strength with the topological connection relationship matrix; step S03, calculating a geographical modularity according to the spatial connection strength, and constructing a hybrid modularity according to the geographical modularity and a topological modularity; step S04, initializing each disaster point as an independent community, and dividing each community node based on the hybrid modularity by adopting a greedy strategy, and judging whether node decomposition is needed in the division process by a persistence degree, and calculating a spatial cohesion degree to judge whether the community needs to be decomposed. The application can reduce resource scheduling paths, and realizes collaborative improvement of geographical and topological joint optimization.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of emergency management resource scheduling, and in particular to a disaster point network community discovery method and device fusing geographical space constraints. BACKGROUND

[0002] In the field of emergency management, the rapid response to public emergencies depends on the accurate clustering of disaster points and the efficient scheduling of resources. Community discovery algorithm, as the core technology of complex network analysis, aims to divide network nodes into node groups with close internal connections. Its application effect directly affects the allocation efficiency of rescue resources and the timeliness of emergency response.

[0003] Traditional community discovery algorithms mainly perform clustering analysis based on network topology structure or density distribution. For example, modularity optimization method (such as Louvain algorithm) realizes community division by maximizing modularity, which needs to rely on the internal edge density of network topology structure. Label propagation method (LPA) iteratively updates community affiliation based on the majority voting principle of neighbor node labels. Density clustering method (such as SCAN algorithm) defines communities in combination with node density and reachability, and is suitable for non-uniformly distributed networks. However, when the above traditional community discovery algorithms are applied to emergency management resource scheduling, the following problems may exist:

[0004] 1. Traditional community discovery algorithms only consider topology connection (such as functional dependence relationship between disaster points) for clustering, without considering the geographical space relationship between disaster points in actual emergency scenarios. Geographical coordinates are usually only used as auxiliary attributes. However, in fact, resource scheduling will be limited by geographical location. Clustering based only on topology connection may lead to deviation of clustering results from actual physical distribution, and it is difficult to support nearby rescue.

[0005] 2. In traditional community discovery algorithms, only node connection strength is considered for modularity calculation, without considering geographical distance. Therefore, geographical space proximity and functional association cannot be quantified at the same time, which may lead to community division results that cannot meet the emergency needs of "functional cooperation closeness" and "geographical distribution concentration" at the same time, and cannot adapt to dynamic scene needs.

[0006] 3. Traditional community discovery algorithms usually adopt a one-way merging strategy, i.e. only allow community merging, but do not allow merged communities to be further decomposed. Once a wrong merge occurs (such as merging nodes that are too far apart into the same community), it cannot be traced back and corrected. The wrong merged community will also occupy computing resources, affecting the subsequent optimization direction, so that the final division deviates from the optimal solution and reduces the functional association within the community. Therefore, the merging result is prone to local optimum, and the functional association in the merging decision cannot be quantified, which may misdivide key functional association nodes (such as upstream and downstream disaster points being divided into different communities), so that important functional nodes are forced to be divided into different communities, leading to resource scheduling path redundancy (cross-community deployment) and reduced response speed. SUMMARY

[0007] The technical problem solved by the present application is that, in view of the technical problems existing in the prior art, the present application provides a disaster point network community discovery method and device fusing geographical space constraints, which can dynamically balance geographical position and topological connection relationship, so as to meet the emergency demand of functional cooperation and close geographical distribution at the same time, and can also avoid falling into local optimum in the community merging process and avoid missegmentation of key nodes.

[0008] To solve the above technical problems, the technical solution provided by the present application is:

[0009] A disaster point network community discovery method fusing geographical space constraints, comprising the following steps:

[0010] Step S01. Obtain geographical position data and topological connection relationship matrix of each disaster point in the disaster point network;

[0011] Step S02. Calculate the spatial similarity matrix between each disaster point according to the geographical position data, and fuse the spatial similarity matrix and the topological connection relationship matrix to calculate the spatial connection strength to quantify the geographical proximity and functional correlation between disaster points at the same time;

[0012] Step S03. Calculate the geographical modularity according to the spatial connection strength, and construct the hybrid modularity according to the geographical modularity and the topological modularity, wherein the topological modularity is used to measure the edge density of the network;

[0013] Step S04. Initialize each disaster point as an independent community, and divide each community node based on the hybrid modularity by using a greedy strategy, wherein in the division process, the maintenance degree of each community node is calculated to determine whether the community node needs to be decomposed to other communities, and the spatial cohesion degree of each community is calculated to determine whether the community needs to be decomposed, and finally the optimized community division result is output, wherein the maintenance degree is calculated according to the number of edges of the community node in the corresponding community to measure the connection tightness of the node and the community intelligence, and the spatial cohesion degree is calculated according to the spatial similarity matrix between each community node in the community to measure the geographical dispersion degree of the nodes in the community.

[0014] Further, step S01 comprises:

[0015] Step S101. Obtain a disaster point data set and geographical position data of the disaster point in a two-dimensional geographical coordinate system wherein, ~ each disaster point is represented by, n the number of disaster points is represented by, respectively represent the longitude and latitude of the geographical position, i the serial number of the disaster point;

[0016] Step S102. Using the disaster point dataset initialize the community as and according to the geographical position data of the disaster point initialize the topological connection relationship matrix of the disaster point wherein, if the disaster point has a topological association with the disaster point , the topological association weight is set to 1, otherwise it is set to 0, and if there is a direct physical connection path or functional association between the two disaster points, the edge between the two disaster points is marked as an adjacent edge.

[0017] Further, step S02 comprises:

[0018] Step S201. According to the geographical position data, use the Gaussian kernel function to calculate the spatial similarity matrix between each disaster point:

[0019]

[0020] wherein, is the Gaussian kernel bandwidth, is the spatial similarity matrix between the disaster point and , respectively, , are the geographical position coordinates of the disaster point and ;

[0021] Step S202. Fuse the spatial similarity matrix and the topological connection relationship matrix to calculate the spatial connection strength :

[0022]

[0023] wherein, is the spatial weight factor, is the topological connection relationship matrix between the disaster point and the disaster point .

[0024] Further, in step S03, according to the geographical modularity and the topological modularity, the calculation expression of the hybrid modularity is constructed as:

[0025]

[0026]

[0027]

[0028]

[0029] wherein, is the hybrid modularity, is the weight coefficient for controlling the topological modularity and the geospatial modularity, , is the topological modularity, and respectively represent the degrees of nodes , , represents the number of edges in the network, is the geospatial modularity, is the sum of the spatial connection strengths of all edges, is the spatial connection strength between node and node , is an indicator function for screening node pairs within the same community.

[0030] Further, in step S04, in the process of dividing the nodes of each community based on the hybrid modularity using the greedy strategy, the weight of the topological modularity and the geospatial modularity is adjusted according to the current disaster stage, wherein in the initial stage, the weight of the topological modularity is reduced and the weight of the geospatial modularity is increased, and in the later stage, the weight of the topological modularity is increased and the weight of the geospatial modularity is reduced.

[0031] Further, in step S04, the step of dividing the nodes of each community based on the hybrid modularity using the greedy strategy comprises:

[0032] Step S401. Calculate all possible community pairs of the merged , and select the pair of communities with the largest to merge, update the community division and the hybrid modularity , that is, select the community pair corresponding to the largest gain to merge;

[0033] Step S402. Record the global maximum modularity , perform local optimal detection, and determine whether the local optimal is reached according to the size of the persistence degree of each node community, if yes, proceed to step S403, otherwise return to step S401;

[0034] Step S403. Node-level disintegration: determine whether the persistence degree of each node of the current community is less than a preset persistence degree threshold , if yes, the corresponding community node is disintegrated into other communities;

[0035] ​Step S404. Community-level disintegration: Determine whether the spatial cohesion of each community is less than the preset spatial cohesion value. If so, then decompose the corresponding community into sub-communities;

[0036] Step S405. Determine the current community size Is it less than the preset size threshold S or the degree of mixed modules? If the change before and after the greedy merging is less than the preset value, and if so, the community discovery and clustering are completed, and the optimized community division results are output.

[0037] Furthermore, the formula for calculating the degree of continuity is:

[0038]

[0039] in, Represents community nodes The degree of maintenance, Community Node Total number of sides Represents community nodes In the community Total number of sides inside, Indicates community Except All other nodes except those mentioned above;

[0040] The formula for calculating the spatial cohesion is:

[0041]

[0042] in, For the community Spatial cohesion, For the community Inner Community Node With community nodes Spatial similarity matrix between them For the community Size.

[0043] Furthermore, in step S04, after outputting the optimized community division results, the method further includes calculating the spatial density and topological connectivity index of each community, and calculating the scheduling priority score of each community based on the calculated spatial density and topological connectivity index. The scheduling priority of each community is determined according to the aforementioned scheduling priority score, wherein, Indicates community Scheduling priority score, To measure the weight of community connectivity, Indicates community Spatial compactness, representing a community a topological connectivity index.

[0044] Further, the spatial density is a spatial cohesion degree to measure the geographical dispersion of disaster points in the community, and the topological connectivity index is a network density to reflect the proportion of actual edges to the theoretical maximum number of edges in the community, and the calculation expression of the network density is:

[0045]

[0046] wherein, is an element in an adjacency matrix , when and node have an edge, , otherwise , is the size of the community .

[0047] A disaster point network community discovery device fusing geographical space constraints comprises a processor and a memory, the memory is used for storing a computer program, and the processor is used for executing the computer program to perform the method as described above.

[0048] Compared with the prior art, the present application has the beneficial effects that:

[0049] 1. The present application uses a spatial similarity quantification mechanism to map geographical position coordinates into a dynamic weight gain item of a topological adjacent edge, simultaneously fuses geographical proximity and topological connection strength in modularity calculation to form a hybrid modularity, can simultaneously quantify geographical proximity and functional relevance, establishes a dual-target joint optimization framework of geographical proximity and network functional relevance, solves the problem of geographical dispersion or functional fragmentation of community division caused by single-dimensional analysis of traditional algorithms, makes the division results meet the requirements of "functional synergy" and "geographical compactness" at the same time, can allow the merging of long-distance but functionally critical nodes, improve the cross-geographical merging ratio of functionally strongly correlated nodes, and reduce redundant scheduling paths.

[0050] 2. The present application adopts a dynamic disintegration optimization strategy of a double-trigger disintegration mechanism when performing community division on disaster points, performs node-level decomposition according to the maintenance degree of community nodes and community decomposition according to the spatial cohesion degree of the community, can split and reconstruct low-embedded nodes and weakly connected communities to avoid local optimal problems, adaptively splits and reconstructs the topological global structure of weakly connected communities, can ensure that the community has the advantages of geographical compactness and functional synergy, and thus can optimize the global division, improve the community division modularity, and improve the stability / fault tolerance of the community division result through iteration.

[0051] 3、The present application can realize disaster point clustering from the dual dimensions of function and space by combining iterative disassembly reconstruction and mixed modularity optimization community division process, improve the clustering effect of disaster points under the background of emergency response, ensure the geographical distribution of nodes in the community is concentrated while ensuring the close functional association, so as to meet the emergency demand of functional cooperation and close geographical distribution at the same time, avoid the problem of low resource scheduling efficiency caused by geographical space constraints, improve the geographical and actual distribution consistency of community division, and realize the synergistic improvement of geographical and topological joint optimization.

[0052] 4、The present application can also avoid the error segmentation caused by local optimization of traditional one-way merging algorithm by dynamically correcting the merging strategy in the community merging process, thereby improving the global division rationality and solving the suboptimal community division problem caused by local optimization. BRIEF DESCRIPTION OF DRAWINGS

[0053] Figure 1 is the implementation flowchart of the disaster point network community discovery method of the present embodiment.

[0054] Figure 2 is the result schematic diagram of community division in the specific application embodiment by using the method of the present application. DETAILED DESCRIPTION

[0055] The present application will be further described below in combination with the drawings and specific preferred embodiments of the present application, but the protection scope of the present application is not limited thereby.

[0056] As shown in Figure 1 , the steps of the disaster point network community discovery method of the present embodiment include:

[0057] Step S01. Obtain the geographical position data and topological connection relationship matrix of each disaster point in the disaster point network.

[0058] In the present embodiment, after inputting the disaster point network G, the geographical position data and topological connection relationship matrix of each disaster point in the network G are obtained, and then the geographical position data of each disaster point is used to initialize the community C and the topological connection relationship matrix, and a standardized disaster point input community is constructed.

[0059] As an optional embodiment, the specific steps of step S01 include:

[0060] Step S101. Obtain the disaster point data set and the geographical position data of the disaster point in the two-dimensional geographical coordinate system , wherein, ~ respectively represent each disaster point, n represents the number of disaster points, respectively represent the longitude and latitude of the geographical position, i the serial number of the disaster point;

[0061] Step S102. Using the disaster point dataset Initialize the community as , and according to the geographical position data of the disaster point Initialize the topological connection relationship matrix of the disaster point , wherein if the disaster point has a topological association with the disaster point , the topological association weight is set to 1, otherwise it is set to 0, if there is a direct physical connection path or functional association between the two disaster points, mark the edge between the two disaster points as an adjacent edge, and construct a standardized disaster point input community.

[0062] Step S02. According to the geographical position data, calculate the spatial similarity matrix between each disaster point, and fuse the spatial similarity matrix and the topological connection relationship matrix to calculate the spatial connection strength to quantify the geographical proximity and functional association between disaster points.

[0063] This embodiment considers the functional heterogeneity of disaster point data, uses a spatial similarity quantification mechanism to map geographical position coordinates to the spatial connection strength of topological adjacent edges, and in the subsequent modularity calculation process, integrates geographical modularity and topological modularity to construct a hybrid modularity, establishes a dual-target joint optimization framework of geographical proximity and network functional association, and can quantify geographical proximity and functional association. From the dual dimensions of geographical space and function, community division is realized, and the problem of geographical dispersion or functional fragmentation caused by single-dimensional analysis of traditional algorithms is solved.

[0064] As an optional implementation, step S02 includes:

[0065] Step S201. According to the geographical position data, use the Gaussian kernel function to calculate the spatial similarity matrix between each disaster point:

[0066] (1)

[0067] wherein, is the Gaussian kernel bandwidth, is the spatial similarity matrix between the disaster point and , and , respectively are the geographical position coordinates of the disaster point and ;

[0068] Step S202. Fuse the Gaussian kernel space similarity matrix with the topological connection relationship matrix to calculate the spatial connection strength :

[0069] (2)

[0070] wherein, is a spatial weight factor, is a topological connection relationship matrix between the disaster point and the disaster point . That is, only when there is a connection on the topology , the geographical similarity takes effect; if there is no connection on the topology , no matter how close the geographical position is, the spatial connection strength is still 0.

[0071] Step S03. Calculate the geographical modularity according to the spatial connection strength, and construct a hybrid modularity according to the geographical modularity and the topological modularity, wherein the topological modularity is used to measure the edge density of the network.

[0072] In this embodiment, the hybrid modularity is constructed by combining the geographical modularity and the topological modularity, so that the geographical space and the topological connection relationship can be considered comprehensively in the subsequent community division process. By setting the weight between the geographical modularity and the topological modularity, the balance between the topology and the space can also be controlled, so that the division result can meet the requirements of “functional cooperation” and “geographical compactness” at the same time, the merging of long-distance but functionally key nodes can be allowed, the cross-geographical merging ratio of functionally strongly correlated nodes can be improved, and the redundant scheduling path can be reduced.

[0073] As an optional implementation, the calculation expression for constructing the hybrid modularity according to the geographical modularity and the topological modularity is as follows:

[0074] (3)

[0075] wherein, is the hybrid modularity, is a weight coefficient for controlling the topological modularity and the geographical spatial modularity, , is the topological modularity, is the geographical spatial modularity.

[0076] As an optional implementation, the calculation of the topological modularity adopts the traditional modularity, for example, the calculation expression is as follows:

[0077] (4)

[0078] wherein, and respectively represent the node , (i.e., corresponding disaster point) and disaster sites (degree) This represents the number of edges in the network.

[0079] As shown in equation (4), the first node is selected from all nodes in the entire network as... For each selected Then select the sorting position from the entire network. Subsequent nodes as until all "" in the entire network have been traversed in front, The accumulation stops after each pair of unique nodes, and the topological modularity is finally calculated. .

[0080] In the formula This is an indicator function used to filter node pairs within the same community:

[0081] (5)

[0082] As an optional implementation, the spatial modularity calculation adopts an optimized design based on the traditional modularity, and the calculation expression is:

[0083] (6)

[0084] in, It is the sum of the spatial connectivity strengths of all edges.

[0085] Step S04. Initialize each disaster point as an independent community, and use a greedy strategy to divide the community nodes based on the mixed modularity. During the division process, the continuity of each community node is calculated to determine whether it is necessary to decompose the community node into other communities, and the spatial cohesion of each community is calculated to determine whether it is necessary to decompose the community. Finally, the optimized community division result is output. The continuity is calculated based on the number of edges of the community node in the corresponding community to measure the tightness of the connection between the node and the community. The spatial cohesion is calculated based on the spatial similarity matrix between the community nodes in the community to measure the geographical dispersion of the nodes in the community.

[0086] The dynamic disintegration optimization strategy using the dual trigger disintegration mechanism is used in the community division in this embodiment. The node-level decomposition is performed according to the maintenance degree of the community nodes, and the community decomposition is performed according to the spatial cohesion degree of the community. The low-embedded nodes and weakly connected communities can be split and reconstructed to avoid the local optimal problem. The adaptive splitting and reconstruction of the weakly connected community can optimize the global topology structure, ensure that the community has the advantages of geographical compactness and functional synergy, and improve the community division modularity and the stability / fault tolerance of the community division result.

[0087] As an optional implementation, in the process of dividing the community nodes based on the hybrid modularity by using the greedy strategy, the specific steps of the dynamic disintegration strategy include:

[0088] Step S401. Calculate all possible community pairs after merging , and select the pair of communities with the maximum gain to merge, update the community division , and the hybrid modularity , that is, select the community pair corresponding to the maximum gain to merge, , and the hybrid modularity difference

[0089] Step S402. Record the global maximum modularity , perform local optimal detection, and determine whether the local optimal is reached according to the size of the maintenance degree of each node community. If yes, go to step S403, otherwise return to step S401.

[0090] Step S403. Node-level disintegration: determine whether the maintenance degree of each community node is less than a preset maintenance degree threshold . If yes, the corresponding community node is decomposed into other communities.

[0091] Step S404. Community-level disintegration: determine whether the spatial cohesion degree of each community is less than a preset spatial cohesion degree value . If yes, perform community decomposition, that is, divide all the communities into multiple community regions.

[0092] Step S405. Determine whether the current community size is less than a preset size threshold S or the hybrid modularity changes by less than a preset value before and after the greedy merging. If yes, complete the community discovery clustering, and output the optimized community division result.

[0093] Traditional community detection algorithms typically only select the merging operation with the maximum current gain during the community merging process, such as simulated annealing. This is prone to getting stuck in suboptimal solutions, making it difficult to achieve global optimization. Furthermore, they often employ a one-way merging strategy, which is prone to incorrectly segmenting functionally related nodes (e.g., classifying upstream and downstream disaster points as different communities), and cannot correct suboptimal solutions through disassembly. The merging process usually uses a rigid geographical distance threshold; if the node distance exceeds this threshold, merging is completely prohibited. This rigid threshold cannot distinguish special scenarios such as "geographically distant but functionally highly related" nodes (e.g., power hub A and refuge B that relies on its power supply), leading to the blocking of merging of strongly functionally related nodes. This embodiment adopts a dynamic disassembly strategy, prioritizing node-level disassembly before addressing community-level disassembly, ensuring iterative stability. After reconstruction and decomposition, the algorithm returns to the merging phase to re-merge until the iteration stops, completing community detection clustering. This solves the local optimum problem during the merging process, allowing for the decomposition of low-embedded nodes or weakly connected communities, generating clustering results that balance geographical concentration and functional synergy. Meanwhile, by dynamically adjusting the merging strategy during the community merging process, we can avoid erroneous partitioning caused by local optima in traditional one-way merging algorithms, thereby improving the rationality of global partitioning and solving the problem of suboptimal community partitioning caused by local optima.

[0094] Local optima refer to a community structure with no room for optimization. Node continuity can quantify the rationality of a single node's affiliation to the current community. The existence of nodes with low continuity means that node affiliation can be adjusted, indicating that the structure has not reached local optima. Conversely, a high continuity indicates that local optima may have been reached. In step S402 above, the continuity of each node is used to determine whether local optima have been reached. Specifically, the proportion of nodes with low continuity globally can be statistically analyzed. If the proportion is lower than a preset threshold (i.e., there are not many nodes that need adjustment), it indicates that there is no room for optimization and is determined to be local optima, proceeding to step S403. If the proportion is higher than the preset threshold (i.e., there are many misallocated nodes), it is determined that local optima have not been reached, and dynamic dismantling and adjustment are required, returning to step S401.

[0095] As an optional implementation method, the formula for calculating the degree of maintenance is:

[0096] (7)

[0097] in, Represents community nodes The degree of maintenance, Community Node Total number of sides Represents community nodes In the community Total number of sides inside, Indicates community Except all other nodes, starting from any node in the community, until all nodes in the community are traversed all non nodes in the community are traversed.

[0098] As an optional implementation, the calculation expression of the spatial cohesion degree is:

[0099] (8)

[0100] wherein, is the spatial cohesion degree of the community, is the size of the community, is the spatial similarity matrix between the community nodes in the community, is the spatial similarity matrix between the community nodes in the community, is the spatial similarity matrix between the community nodes in the community, is the spatial similarity matrix between the community nodes in the community, is the spatial similarity matrix between the community nodes in the community, is the spatial similarity matrix between the community nodes in the community, is the spatial similarity matrix between the community nodes in the community, is the spatial similarity matrix between the community nodes in the community, is the spatial similarity matrix between the community nodes in the community, is the spatial similarity matrix between the community nodes in the community, is the spatial similarity matrix between the community nodes in the community, is the spatial similarity matrix between the community nodes in the community, is the spatial similarity matrix between the community nodes in the community, is the spatial similarity matrix between the community nodes in the community, is the spatial similarity matrix between the community nodes in the community,

[0101] In this embodiment, by combining the iterative disintegration reconstruction and the mixed modularity optimization community division process, the dual trigger disintegration strategy of the node-level decomposition based on the node maintenance degree and the community-level decomposition based on the community cohesion is adopted, and the dynamic disintegration reconstruction is combined in the iteration, which can eliminate the local optimal trap and improve the division efficiency, is suitable for the emergency scene with high real-time requirement, and realizes the suboptimal solution correction and global convergence.

[0102] In this embodiment, in the process of dividing each community node based on the mixed modularity by using the greedy strategy, the weights of the topological modularity and the geographical spatial modularity are adjusted according to the current disaster stage, wherein in the initial stage, the weight of the topological modularity is reduced and the weight of the geographical spatial modularity is increased, and in the later stage, the weight of the topological modularity is increased and the weight of the geographical spatial modularity is reduced, so as to dynamically adjust the geographical and topological weights in combination with the disaster stage, to strengthen the geographical proximity merging tendency in the initial stage, and focus on the community division of the function correlation optimization in the later stage, so as to improve the emergency response efficiency in the whole cycle and realize the dynamic balance adaptive to the disaster stage.

[0103] Specifically, the division time parameters of the disaster stage are configured as and , the initial stage is The intermediate stage is The later stage is And set weight coefficients according to different stages. The importance of topology and space is dynamically adjusted at different stages. For example, a smaller topology and space are configured in the initial stage. The initial value (e.g., 0.3) emphasizes the importance of its topological structure, while it can be appropriately increased in later stages. Values ​​(such as 0.8) are used to emphasize the priority of geographical structure before disaster point aggregation, so that disaster points are aggregated based on geographical proximity during the emergency phase, while the focus of functional network reconstruction should be placed on the recovery phase.

[0104] Traditional spatial clustering methods (such as DBSCAN) use fixed geographical thresholds (such as neighborhood radius ε), which cannot adapt to complex emergency scenarios. This leads to a loose geographical distribution of communities during the emergency phase, insufficient functional connectivity during the recovery phase, and increased global scheduling costs. This embodiment adopts a phase-aware dynamic weight adjustment mechanism, emphasizing geographical proximity in the early stages of a disaster and tilting towards functional connectivity in the later stages. This improves geographical density during the emergency phase, which is beneficial for functional connectivity during the recovery phase and enhances overall response efficiency.

[0105] In step S04 of this embodiment, after outputting the optimized community division results, the method further includes calculating the spatial density and topological connectivity index of each community, and calculating the scheduling priority score of each community based on the calculated spatial density and topological connectivity index. The dispatch priority of each community is determined by scoring according to dispatch priority, resulting in an emergency priority sequence for each community. This facilitates the generation of emergency resource allocation decisions, among which, Indicates community Scheduling priority score, To measure the weight of community connectivity, Indicates community Spatial compactness, Indicates community The topological connectivity index.

[0106] In this embodiment, spatial density is used to measure the geographical dispersion of disaster points within the community, and topological connectivity is used to measure network density to reflect the ratio of the actual number of edges to the theoretical maximum number of edges within the community. The expression for calculating network density is as follows:

[0107]

[0108] in, Adjacency matrix The elements in the node, when the node With nodes When there are connecting edges, Otherwise , For the community Size.

[0109] Specifically, the spatial cohesion of each community in , that is , is calculated, wherein is the community in , to measure the geographical dispersion of disaster points within the community; the network density of each community in is calculated to reflect the ratio of the actual edge number to the theoretical maximum edge number, and the value is closer to 1, indicating that the connection is more intensive.

[0110] In summary, the present application linearly fuses spatial similarity and topological connection, so that the modularity calculation quantifies both types of weights, realizes disaster point clustering from the functional and spatial dimensions, and improves the clustering effect of disaster points under the background of emergency response. While ensuring the closeness of functional association, it ensures the geographical distribution of nodes within the community, so that it can meet the emergency needs of functional synergy and geographical distribution concentration, avoid the problem of low resource scheduling efficiency caused by geographical space constraints, improve the geographical and actual distribution consistency of community division, and enable functionally strongly associated nodes to be merged across geography under high weight compensation. Ultimately, the dual objectives of "not fragmented in function and not loose in geography" are optimized, the resource scheduling path is reduced, the geographical and topological joint optimization is realized, and the synergy is improved.

[0111] To verify the effect of the present application, in a specific application example, the above-mentioned method of the present application is used to perform disaster point community clustering on a 10-node, 15-edge undirected network, wherein the geographical positions of the 10 nodes are set as follows: node 1 [10, 20], node 2 [30, 40], node 3 [50, 60], node 4 [70, 30], node 5 [20, 70], node 6 [40, 20], node 7 [60, 40], node 8 [30, 30], node 9 [50, 20], and node 10 [20, 50], and the connection relationship adjacency matrix of each node is as follows:

[0112]

[0113] When calculating the spatial connection strength , take , and the 10-node network is merged through a greedy strategy as follows:

[0114] (1) Merge operation: merge community 10 (node

[10] ) into community 5 (node [5]), and the remaining 9 communities

[0115] Community 5 after merge: nodes [5, 10], modularity gain: 0.0582, current modularity: -0.0471

[0116] (2) Merge operation: merge community 9 (node [9]) into community 4 (node [4]), remaining 8 communities

[0117] Community 4 after merge: nodes [4, 9], modularity gain: 0.0565, current modularity: 0.0094

[0118] (3) Merge operation: merge community 4 (nodes [4, 9]) into community 3 (node [3]), remaining 7 communities

[0119] Community 3 after merge: nodes [3, 4, 9], modularity gain: 0.0894, current modularity: 0.0988

[0120] (4) Merge operation: merge community 8 (node [8]) into community 6 (node [6]), remaining 6 communities

[0121] Community 6 after merge: nodes [6, 8], modularity gain: 0.0513, current modularity: 0.1501

[0122] (5) Merge operation: merge community 6 (nodes [6, 8]) into community 1 (node [1]), remaining 5 communities

[0123] Community 1 after merge: nodes [1, 6, 8], modularity gain: 0.0786, current modularity: 0.2288

[0124] (6) Merge operation: merge community 2 (node [2]) into community 1 (nodes [1, 6, 8]), remaining 4 communities, community 1 after merge: nodes [1, 2, 6, 8], modularity gain: 0.0478, current modularity: 0.2765

[0125] (7) Merge operation: merge community 7 (node [7]) into community 3 (nodes [3, 4, 9]), remaining 3 communities, community 3 after merge: nodes [3, 4, 7, 9], modularity gain: 0.0232, current modularity: 0.2998

[0126] Optimize low-embedded nodes and weakly connected communities during the partitioning process through dynamic disintegration strategy, and the final community distribution after optimization:

[0127] Community 1: 4 nodes → [1, 2, 6, 8]

[0128] Community 2: 4 nodes → [3, 4, 7, 9]

[0129] Community 3: 2 nodes -> [5, 10]

[0130] The division result is specifically as shown in Figure 2

[0131] According to the final community distribution, the spatial cohesion of each community is calculated as follows:

[0132]

[0133] Community 2: 0.1115; Community 3: 0.1401.

[0134] The result of the network density is as follows:

[0135] Community 1: 0.8333; Community 2: 0.6667; Community 3: 1.0.

[0136]

[0137] Further, the priority score of each community is calculated as follows: The emergency priority sequence of each community is obtained by sorting the community dispatch priority, that is, to support emergency resource scheduling decisions.

[0138] The embodiment further provides a disaster point network community discovery device integrating geographic spatial constraints, including a processor and a memory, the memory is used to store a computer program, and the processor is used to execute the computer program to perform the above method.

[0139] It can be understood that the above method of the embodiment can be executed by a single device, such as a computer or a server, etc., and can also be applied to a distributed scenario to be completed by multiple devices cooperating with each other. In the case of a distributed scenario, one of the multiple devices can only execute one or more steps in the above method of the embodiment, and the multiple devices interact to complete the above method. The processor can be implemented in the form of a general-purpose CPU, a microprocessor, an application-specific integrated circuit, or one or more integrated circuits, etc., for executing related programs to implement the above method of the embodiment. The memory can be implemented in the form of a read-only memory ROM, a random access memory RAM, a static storage device, and a dynamic storage device, etc. The memory can store an operating system and other application programs, and when the above method of the embodiment is implemented by software or firmware, the related program codes are saved in the memory and executed by the processor.

[0140] ​​​​​​​The above are only the preferred embodiments of the present application, and do not limit the present application in any form. Although the present application has been disclosed with the preferred embodiments as above, it is not intended to limit the present application. Therefore, any simple modification, equivalent change and modification made to the above embodiments without departing from the technical solutions of the present application, and according to the technical essence of the present application, should fall within the scope of protection of the technical solutions of the present application.

Claims

1. A method for discovering disaster point network communities that incorporates geospatial constraints, characterized by the following steps: include: Step S01. Obtain the geographical location data and topology connection matrix of each disaster point in the disaster point network. The geographical location data includes the longitude and latitude of the geographical location. In the topology connection matrix, if there is a topological association between two disaster points, the corresponding topological association weight is set to 1, otherwise it is set to 0. Step S02. Calculate the spatial similarity matrix between each disaster point based on the geographical location data, and fuse the spatial similarity matrix with the topological connection matrix to calculate the spatial connection strength so as to simultaneously quantify the geographical proximity and functional correlation between disaster points; Step S03. Calculate the geographic modularity based on the spatial connectivity strength, and construct a hybrid modularity based on the geographic modularity and the topological modularity. The topological modularity is used to measure the edge density of the network. Step S04. Initialize each disaster point as an independent community, and use a greedy strategy to divide the community nodes based on the hybrid modularity. During the division process, the continuity of each community node is calculated to determine whether it is necessary to decompose the community node into other communities, and the spatial cohesion of each community is calculated to determine whether it is necessary to decompose the community. Finally, the optimized community division result is output. The continuity is calculated based on the number of edges of the community node in the corresponding community to measure the tightness of the connection between the node and the community. The spatial cohesion is calculated based on the spatial similarity matrix between the community nodes in the community to measure the geographical dispersion of the nodes in the community. Step S04, which involves using a greedy strategy to partition each community node based on the hybrid modularity, includes: Step S401. Calculate all possible community pairs after the merger. , and select The largest pair of communities were merged, and the community boundaries were updated. With hybrid modularity That is, selecting the community pair corresponding to the maximum gain. To merge, Indicates the difference in hybrid modularity; Step S402. Record the global maximum modularity. Perform local optimum detection and determine whether a local optimum has been reached based on the degree of community maintenance of each node. If so, proceed to step S403; otherwise, return to step S401. Step S403. Node-level disintegration: Determine whether the current continuity of each community node is less than the preset continuity threshold. If so, the corresponding community node will be distributed to other communities; Step S404. Community-level disintegration: Determine whether the spatial cohesion of each community is less than the preset spatial cohesion value. If so, then community decomposition will be performed, that is, all communities will be divided into multiple community areas; Step S405. Determine the current community size Is it less than the preset size threshold S or the degree of mixed modules? If the change before and after the greedy merging is less than the preset value, and if so, the community discovery and clustering are completed, and the optimized community division results are output.

2. The disaster point network community discovery method integrating geospatial constraints according to claim 1, characterized in that, Step S01 includes: Step S101. Obtain disaster point dataset and geographical location data of disaster points in a two-dimensional geographic coordinate system ,in, ~ Each represents a different disaster point. n Indicates the number of disaster sites. These represent the longitude and latitude of the geographical location, respectively. i Indicates the sequence number of the disaster point; Step S102. Use the disaster point dataset Initialize the community as And based on the geographical location data of the disaster site Initialize the topological connectivity matrix of disaster points Among them, if the disaster point and disaster sites If there is a topological association between two disaster points, the topological association weight is set to 1; otherwise, it is set to 0. If there is a direct physical connection path or functional association between two disaster points, the edge between the two disaster points is marked as an adjacent edge.

3. The disaster point network community discovery method integrating geospatial constraints according to claim 1, characterized in that, Step S02 includes: Step S201. Calculate the spatial similarity matrix between each disaster point using the Gaussian kernel function based on the geographical location data: in, For Gaussian kernel bandwidth, Disaster point and Spatial similarity matrix between them , Disaster points and Geographical coordinates; Step S202. Merge the spatial similarity matrix and the topological connectivity matrix to calculate the spatial connectivity strength. : in, Spatial weighting factor, Disaster point and disaster sites The topological connection matrix between them.

4. The disaster point network community discovery method integrating geospatial constraints according to claim 1, characterized in that, In step S03, the calculation expression for the hybrid modularity is constructed based on the geographic modularity and the topological modularity as follows: in, For mixed modularity, These are the weighting coefficients used to control the topological modularity and the geospatial modularity. , For topological modularity, and Representing nodes respectively and The degree, Represents the number of edges in the network. For geospatial modularity, It is the sum of the spatial connectivity strengths of all edges. For nodes With nodes Spatial connection strength between them This is an indicator function used to filter node pairs within the same community.

5. The disaster point network community discovery method integrating geospatial constraints according to claim 4, characterized in that, In step S04, during the process of dividing each community node based on the hybrid modularity using a greedy strategy, the weights of topological modularity and geospatial modularity are adjusted according to the current disaster stage. In the initial stage, the weight of topological modularity is reduced and the weight of geospatial modularity is increased, while in the later stage, the weight of topological modularity is increased and the weight of geospatial modularity is decreased.

6. The disaster point network community discovery method integrating geospatial constraints according to any one of claims 1 to 5, characterized in that, The formula for calculating the degree of continuity is: in, Represents community nodes The degree of maintenance, Community Node Total number of sides Represents community nodes In the community Total number of sides inside, Indicates community Except All other nodes except those mentioned above; The formula for calculating the spatial cohesion is: in, For the community Spatial cohesion, For the community Inner Community Node With community nodes Spatial similarity matrix between them For the community Size.

7. The disaster point network community discovery method integrating geospatial constraints according to any one of claims 1 to 5, characterized in that, In step S04, after outputting the optimized community partitioning results, the method further includes calculating the spatial density and topological connectivity index of each community, and calculating the scheduling priority score of each community based on the calculated spatial density and topological connectivity index. The scheduling priority of each community is determined according to the aforementioned scheduling priority score, wherein, Indicates community Scheduling priority score, To measure the weight of community connectivity, Indicates community Spatial compactness, Indicates community The topological connectivity index is defined as follows: spatial density is spatial cohesion to measure the geographical dispersion of disaster points within the community, and network density is the topological connectivity index to reflect the ratio of the actual number of edges to the theoretical maximum number of edges within the community.

8. The disaster point network community discovery method integrating geospatial constraints according to claim 7, characterized in that, The expression for calculating the network density is: in, Adjacency matrix The elements in the node, when the node With nodes When there are connecting edges, ,otherwise , For the community Size.

9. A disaster point network community discovery device integrating geospatial constraints, comprising a processor and a memory, wherein the memory is used to store a computer program, characterized in that, The processor is used to execute the computer program to perform the method as described in any one of claims 1 to 8.

Images