Emergency evacuation information processing method, electronic equipment and storage medium

Abstract

The invention provides an emergency evacuation information processing method, electronic equipment and a storage medium, and the method comprises the steps: obtaining real-time data to construct a space-time diagram structure, predicting the space-time diagram structure through a demand prediction model, and obtaining a demand prediction result of each emergency place, and the demand prediction result of each emergency place is predicted through a resource allocation model to obtain a resource allocation scheme at the current time, and then emergency evacuation resource scheduling is performed based on the resource allocation scheme. According to the method and the device, dynamic adjustment based on real-time data can be realized, the utilization rate of resources is improved, and complex dynamic demand changes can be flexibly coped with.

Description

Technical Field

[0001] The present application relates to the technical field of emergency evacuation, and in particular to an emergency evacuation information processing method, electronic equipment and storage medium. Background Art

[0002] When encountering an emergency, it is necessary to quickly and reasonably arrange the evacuation of people and the distribution of materials in shelters. An effective evacuation system becomes the key to reducing casualties. The evacuation system can analyze demand based on data sources and make decisions such as resource allocation and evacuation route planning.

[0003] However, most existing evacuation systems make decisions based on a single data source, and the prediction accuracy is insufficient. In addition, the existing evacuation systems are difficult to make real-time dynamic adjustments to cope with emergencies in terms of resource allocation and route planning. They have problems with poor flexibility and timeliness, which greatly affects the efficiency of emergency evacuation management. Summary of the invention

[0004] The purpose of this application is to provide an emergency evacuation information processing method, electronic device and storage medium to address the deficiencies in the above-mentioned prior art, so as to solve the problem that the evacuation system in the prior art is also difficult to make real-time dynamic adjustments to cope with emergencies in terms of resource allocation, path planning, etc., and has poor flexibility and timeliness.

[0005] To achieve the above objectives, the technical solutions adopted in this application are as follows:

[0006] In a first aspect, the present application provides an emergency evacuation information processing method, the method comprising:

[0007] Acquire actual data of each emergency site at the current time, the actual data including: personnel density of the site, remaining capacity of the site, and resource inventory of the site;

[0008] Constructing a spatiotemporal graph structure according to the actual data, wherein the spatiotemporal graph structure includes a plurality of nodes and edges connecting the nodes, wherein the nodes are used to represent emergency sites, and the attribute information of the nodes includes: the personnel density of the sites, the remaining capacity of the sites, and the resource stock of the sites, and the edges are used to represent the traffic information between the emergency sites corresponding to the two nodes connected by the edges;

[0009] Input the spatiotemporal graph structure into a pre-trained demand forecasting model, and use the spatiotemporal graph convolutional network and variational autoencoder in the demand forecasting model to predict the demand forecast result corresponding to each node in the spatiotemporal graph structure at the current time;

[0010] Inputting the demand forecast result corresponding to the current time into the resource allocation model, and using the resource allocation model to predict the resource allocation plan corresponding to the current time;

[0011] According to the resource allocation plan, emergency evacuation resources are dispatched for the emergency places corresponding to each of the nodes.

[0012] Optionally, the step of inputting the spatiotemporal graph structure into a pre-trained demand forecasting model, and predicting the demand forecasting result corresponding to each node in the spatiotemporal graph structure at the current time by the spatiotemporal graph convolutional network and the variational autoencoder in the demand forecasting model, includes:

[0013] Inputting the spatiotemporal graph structure into the spatiotemporal graph convolutional network, and performing convolution processing in sequence by each convolutional layer in the spatiotemporal graph convolutional network to obtain a demand prediction matrix of the spatiotemporal graph structure, wherein each convolutional layer includes a graph convolution sublayer and a time convolution sublayer;

[0014] The demand prediction matrix is ​​input into the variational autoencoder, and the encoder in the variational autoencoder performs encoding processing to obtain an encoding vector, and the encoding vector is input into the decoder in the variational autoencoder, and the decoder performs decoding processing to obtain the demand prediction result corresponding to each node in the space-time graph structure at the current time.

[0015] Optionally, inputting the demand forecast result corresponding to the current time into a resource allocation model, and obtaining the resource allocation scheme corresponding to the current time by prediction by the resource allocation model, comprises:

[0016] The resource allocation model is run in a preset time period, and the current demand forecast results of each emergency site in the current time period and the resource inventory are input into the resource allocation model. The resource allocation model uses the allocation quantity and total inventory of each emergency site as constraints, and determines the resource allocation plan for each emergency site according to the current demand forecast results and resource inventory.

[0017] Optionally, the resource allocation model uses the allocation quantity and total inventory of each emergency site as constraints, and determines the demand priority and resource allocation plan of each emergency site according to the current demand forecast result and resource inventory, including:

[0018] Determining the demand priority of each of the emergency sites based on a preset weight coefficient, the current demand forecast result, and the resource inventory;

[0019] The allocation quantity of each emergency site is determined based on the demand priority of each emergency site, the current demand forecast result and the current resource inventory.

[0020] Optionally, the obtaining of actual data of each emergency location at the current time includes:

[0021] Obtaining actual demand data corresponding to at least one historical time and a baseline demand forecast result corresponding to each of the historical times;

[0022] Comparing the actual demand data corresponding to the historical time with the benchmark demand forecast result corresponding to the historical time to obtain a comparison result;

[0023] Analyzing and processing the comparison result to determine whether to start rolling prediction;

[0024] If so, obtain the actual data of each emergency location at the current time.

[0025] Optionally, the process of determining the benchmark demand forecast result includes:

[0026] Obtaining historical data of each emergency site at a historical time, the historical data including: the personnel density of the site, the remaining capacity of the site, and the resource inventory of the site;

[0027] Constructing a historical graph structure according to the historical data, wherein the historical graph structure includes a plurality of nodes and edges connecting the nodes, wherein the nodes are used to represent emergency sites, and the attribute information of the nodes includes: the personnel density of the sites, the remaining capacity of the sites, and the resource stock of the sites, and the edges are used to represent the traffic information between the emergency sites corresponding to the two nodes connected by the edges;

[0028] The historical graph structure is input into a pre-trained demand forecasting model, and the spatiotemporal graph convolutional network and variational autoencoder in the demand forecasting model predict the benchmark demand forecast results corresponding to each node in the historical graph structure.

[0029] Optionally, the method further comprises:

[0030] According to the resource inventory and demand forecast results of each of the emergency sites at the current time, a shared resource pool is constructed, wherein the shared resource pool includes: the surplus resources of each of the emergency sites;

[0031] Determining whether to trigger regional sharing based on demand forecast results of each of the emergency sites and preset thresholds;

[0032] If so, determining the sharing priority of each of the emergency sites, determining at least one target emergency site from each of the emergency sites according to the sharing priority, determining the resources to be scheduled of each of the target emergency sites from the shared resource pool based on the automatic scheduling algorithm, and generating a resource scheduling plan for each of the resources to be scheduled;

[0033] The resources to be scheduled are scheduled according to the resource scheduling schemes.

[0034] Optionally, after scheduling the resource allocation amount of each node according to the resource allocation scheme, the method further includes:

[0035] Acquire actual demand data, and obtain a prediction error based on the actual demand data and each of the demand prediction results;

[0036] If the prediction error is greater than a preset error threshold, the demand prediction model is trained according to the actual demand data and each of the demand prediction results to obtain an optimized demand prediction model.

[0037] In a second aspect, an embodiment of the present application further provides an electronic device, comprising: a processor, a storage medium and a bus, wherein the storage medium stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor communicates with the storage medium through the bus, and the processor executes the machine-readable instructions to perform the steps of an emergency evacuation information processing method as described in any one of the first aspects.

[0038] In a third aspect, an embodiment of the present application further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of an emergency evacuation information processing method as described in any one of the first aspects are executed.

[0039] The beneficial effects of the present application are as follows: by collecting various types of actual data from multiple emergency sites and realizing the fusion of multiple data sources, evacuation decisions and resource allocation decisions can be made more reliable in complex environments. By constructing a spatiotemporal graph structure for actual data, the spatiotemporal relationship between emergency sites can be more comprehensively reflected, thereby providing a global perspective for demand forecasting and resource allocation. By using a demand forecasting model to perform demand forecasting and obtain the demand distribution of each emergency site in the future, the characteristics of personnel flow and fluctuations in material demand in each emergency site can be better reflected, thereby improving the system response speed and resource allocation efficiency. By generating a resource allocation plan based on the demand forecasting results through a resource allocation model, it is possible to ensure balanced resource use in multiple emergency sites and avoid overload or underutilization of a single emergency site. Compared with the traditional method of equally distributing resources, the present application achieves the maximization of resource utilization in emergency sites.

[0040] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, preferred embodiments are specifically cited below and described in detail with reference to the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0042] Figure 1 A schematic diagram of the architecture of an emergency evacuation information processing system provided by an embodiment of the present application is shown;

[0043] Figure 2 A flowchart of an emergency evacuation information processing method provided by an embodiment of the present application is shown;

[0044] Figure 3 An example diagram of a space-time graph structure provided by an embodiment of the present application is shown;

[0045] Figure 4 A flow chart for determining demand forecast results provided by an embodiment of the present application is shown;

[0046] Figure 5 A flow chart for determining the allocation amount of each emergency site provided by an embodiment of the present application is shown;

[0047] Figure 6 A flowchart of obtaining actual data provided by an embodiment of the present application is shown;

[0048] Figure 7 A flowchart for determining a baseline demand forecast result provided by an embodiment of the present application is shown;

[0049] Figure 8 A flow chart of resource scheduling provided by an embodiment of the present application is shown;

[0050] Fig. 9 A flow chart for optimizing a demand forecasting model provided by an embodiment of the present application is shown;

[0051] Fig.10 A schematic diagram of the structure of an emergency evacuation information processing device provided in an embodiment of the present application is shown;

[0052] Fig.11 A schematic structural diagram of an electronic device provided in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0053] To make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments. The components of the embodiments of the present application generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application claimed for protection, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present application.

[0054] It should be noted that the term "comprising" will be used in the embodiments of the present application to indicate the existence of the features declared thereafter, but does not exclude the addition of other features.

[0055] When a city encounters an emergency, how to effectively evacuate the crowd and ensure the supply of materials to various emergency sites becomes an important issue.

[0056] Currently, materials can be distributed through the evacuation system. By analyzing data such as building layout and personnel situation, the evacuation system can plan evacuation routes according to the location of each emergency site and allocate resources according to the resource needs of each emergency site.

[0057] However, existing evacuation systems generally analyze a single data source and cannot effectively integrate multiple real-time data for accurate prediction. When planning a route, the Dijkstra algorithm and A* algorithm are often used to plan the shortest path, but these algorithms cannot be dynamically adjusted in emergencies and are difficult to cope with changes in demand. In addition, most emergency resource allocations are based on static demand and lack the ability to update in real time. In the case of a surge in resource demand or uneven distribution, serious resource shortages or waste may occur.

[0058] In summary, the existing evacuation system has problems such as lack of dynamic prediction capabilities, low resource utilization efficiency, and inability to flexibly respond to complex dynamic demand changes.

[0059] Based on this, this application proposes an emergency evacuation information processing method, which can be applied to Figure 1 In the emergency evacuation information processing system shown. Figure 1The system includes electronic equipment, data acquisition devices installed in multiple emergency sites, and data acquisition devices installed on the passages between the emergency sites. Each data acquisition device sends the collected real-time data to the electronic equipment. The electronic equipment analyzes and processes the acquired real-time data and generates the current resource allocation plan, so that users can complete the evacuation of personnel and resource scheduling of the emergency sites based on the resource allocation plan.

[0060] This application can first perform demand forecasting and pre-allocation based on historical data, analyze the data to obtain the benchmark demand forecast results for the target time period, and when the plan generation and material scheduling for each sub-period are generated and completed in sequence based on the benchmark demand forecast results, the actual demand after scheduling each sub-period can be compared with the benchmark demand forecast results to determine whether it is necessary to re-forecast the demand. If so, the current real-time data is obtained and re-forecasted to obtain the demand forecast results, and the plan for the current sub-period is regenerated based on the demand forecast results. In response to emergencies during the evacuation scheduling process, this application can realize dynamic adjustments based on real-time data, improve resource utilization, and can also flexibly respond to complex dynamic demand changes.

[0061] Next, combine Figure 2 The emergency evacuation information processing method of the present application is described. The execution subject of the method can be Figure 1 The electronic device having processing capability shown in FIG. Figure 2 , the method comprising:

[0062] S201. Acquire actual data of each emergency site at the current time, where the actual data includes: personnel density of the site, remaining capacity of the site, and resource inventory of the site.

[0063] Multiple data collection devices can be pre-deployed in emergency sites, including sensors, cameras, and GPS (Global Position System) positioning devices, etc., to collect information about population flow, remaining capacity, and resource inventory in emergency sites. Multiple data collection devices can also be deployed between emergency sites to collect information about road traffic conditions between emergency sites. Each data collection device can be used to collect information about the flow of people in the emergency site, the remaining capacity of the site, and the resource inventory of the site. Figure 1 The electronic device shown is connected and the collected data is transmitted to the electronic device in real time.

[0064] The density of people in the venue can be determined by monitoring data from multiple cameras and GPS positioning devices installed in the venue. The remaining capacity of the venue indicates the capacity of people that the emergency venue can still accommodate. The resource stock of the venue can be the remaining material stock of the venue.

[0065] The electronic device can pre-generate the demand forecast results of each emergency site in the future as the benchmark demand forecast results, and generate the scheduling plan for the sub-period in the future based on the benchmark demand forecast results. Each time the scheduling plan for each emergency site is generated and the scheduling is completed, the actual demand of each emergency site in this round can be compared with the benchmark demand forecast results. If the actual demand differs greatly from the benchmark demand forecast results, step S201 can be executed to obtain the actual data of the current time. If the actual demand differs little from the benchmark demand forecast results, the scheduling plan for the next sub-period can be generated based on the benchmark demand forecast results.

[0066] For example, assuming that the benchmark demand forecast results for the future T0-T6 period have been generated in advance, the scheduling plan for the T0-T1 period can be generated first based on the benchmark demand forecast results for the T0-T1 period in the benchmark demand forecast results. After the scheduling plan is executed, the actual demand of each emergency site in the T0-T1 period is compared with the benchmark demand forecast results in the T0-T1 period. If there is a large gap between the actual demand and the benchmark demand forecast results, real-time data can be obtained, and the demand forecast results for the T1-T7 period can be regenerated. The newly generated demand forecast results are used as new benchmark forecast results, and a scheduling plan for the T1-T2 period is generated based on the new benchmark forecast results.

[0067] S202. Construct a spatiotemporal graph structure based on actual data. The spatiotemporal graph structure includes multiple nodes and edges connecting the nodes. The nodes are used to represent emergency sites. The attribute information of the nodes includes: the personnel density of the site, the remaining capacity of the site, and the resource stock of the site. The edges are used to represent the traffic information between the emergency sites corresponding to the two nodes connected by the edge.

[0068] Optionally, the spatiotemporal graph structure includes a node set V = {v1, v2, ..., v n}, edge set E = {(v i , v j )}, node attribute matrix X and edge weight matrix W. Each element in the node set is used to represent an emergency site, each element in the edge set is used to represent the traffic information between two connected emergency sites, each element in the node attribute matrix is ​​used to represent the attribute information of the node, and each element in the edge weight matrix is ​​used to represent the attribute information of the edge.

[0069] The attribute information of the node includes: the personnel density of the site, the remaining capacity of the site, and the resource stock of the site. The personnel density of the site can be the current population of the site, the remaining capacity of the site can be the number of people that the site can accommodate, and the resource stock of the site can be the stock of food, water, medicine and other materials in the site. The edge between two nodes is used to represent the travel distance or time between the emergency sites represented by the nodes. The attribute information of the edge includes: road travel time, real-time traffic status, etc.

[0070] The actual data includes data of multiple emergency sites and road traffic data between emergency sites. Relevant data of emergency sites are extracted from the actual data, and each node and attribute information of each node are generated. Edges are generated for the nodes of connected emergency sites according to the road traffic data, and attribute information of the edges is generated based on the road traffic data, thereby obtaining a spatiotemporal graph structure.

[0071] The spatiotemporal graph structure can characterize the spatiotemporal relationship between emergency sites, thus providing a global perspective for demand forecasting and resource scheduling. Figure 3 As shown in the figure, it is an example of a time-space graph structure. Assuming that the system is connected to 6 emergency places, refer to Figure 3 ,The time-space graph structure includes 6 nodes, each node is used to represent an emergency place, and the nodes are connected by edges, which are used to represent the traffic connection information of the two connected emergency places, including the path status of the traffic connection and the travel time or distance, etc. The path status includes: inaccessible and accessible. The existence of edges between nodes indicates that there are accessible paths between the emergency places represented by the nodes, and whether the paths are accessible needs to be determined based on the path status. The absence of edges between nodes indicates that there is no accessible path between the emergency places represented by the nodes.

[0072] S203, input the spatiotemporal graph structure into a pre-trained demand forecasting model, and use the spatiotemporal graph convolutional network and variational autoencoder in the demand forecasting model to predict the demand forecast results corresponding to each node in the spatiotemporal graph structure at the current time.

[0073] Optionally, the demand forecast result may characterize the demand distribution of emergency sites, including the predicted number of evacuees, the predicted amount of material demand, etc. Based on the demand forecast result, the number of people to be evacuated to each emergency site and the amount of materials to be dispatched to each emergency site may be determined.

[0074] The demand forecasting model includes Spatial Temporal Graph Convolutional Networks (ST-GCN network), Variational Autoencoder (VAE encoder) and decoder. Among them, the spatial temporal graph convolutional network can predict the input spatial temporal graph structure and obtain the demand forecast matrix corresponding to the spatial temporal graph structure. The VAE encoder can encode the demand forecast matrix to generate resource requirements under different demand scenarios, and can realize uncertainty modeling of demand forecasting and multi-scenario demand generation to meet different forecasting scenarios. The decoder can decode the encoding vector output by the VAE encoder to obtain the demand forecast results corresponding to each node in the spatial temporal graph structure at the current time.

[0075] It should be noted that the demand forecasting model can generate demand forecast results for the first period in the future based on the actual data at the current time. For example, the demand forecast results for each node in the next several hours, including the forecast results for the number of personnel at each emergency site at each time point in the next several hours and the predicted demand for materials.

[0076] S204: Input the demand forecast result corresponding to the current time into the resource allocation model, and use the resource allocation model to predict and obtain the resource allocation plan corresponding to the current time.

[0077] The demand forecasting model can forecast the demand in the first future period, and the resource allocation model can use the demand forecasting result as input to forecast the resource allocation plan in the sub-period of the second future period, where the second period can be a sub-period of the first period.

[0078] It should be understood that in the process of resource scheduling, if an emergency occurs, the number of personnel or the inventory of materials at the emergency site will be inconsistent with the predicted demand forecast results. Therefore, in this application, after predicting the demand forecast model for a period of time in the future, a resource allocation plan is generated for the sub-periods of the future period. If the number of personnel or the inventory of materials at the site is inconsistent with the predicted demand forecast results, the demand forecast results can be regenerated and the plan can be regenerated. Otherwise, the resource allocation plan for each sub-period will continue to be generated in sequence according to the demand forecast results.

[0079] For example, assuming that the demand forecasting model can forecast the demand for the next six hours and obtain the demand forecast results for the next six hours, the resource allocation model can generate the resource allocation plan for the first hour based on the demand forecast results for the first hour.

[0080] The resource allocation model may be a Mixed Integer Linear Programming (MILP) resource allocation model, which determines the resource allocation plan for each emergency site by constructing a cost minimization objective function and taking the allocation quantity and total inventory quantity of each emergency site as constraints.

[0081] The resource allocation plan is used to describe the resource allocation amount of emergency sites and the resource allocation method of each emergency site. The resource allocation method includes the source of resources (resource supply point or other emergency sites), the receiving place of resources, and the route of resource transportation. As a possible implementation method, the resource allocation model can also generate a personnel evacuation plan based on the demand forecast results to describe the number of personnel evacuated from the emergency site, which can specifically include the number of personnel entering the emergency site and the number of personnel evacuated from the emergency site.

[0082] S205. According to the resource allocation plan, emergency evacuation resources are dispatched for the emergency sites corresponding to each node.

[0083] Emergency evacuation resources are dispatched to the emergency places corresponding to each node, which can be to dispatch materials to each emergency place according to the dispatch plan indicated by the resource allocation plan, and evacuate personnel to each emergency place according to the dispatch plan indicated by the resource allocation plan.

[0084] It is worth noting that after the dispatch is completed, the actual material demand and the actual number of evacuees at each emergency site can be obtained, and the actual material demand can be compared with the predicted material demand in the demand forecast result, and the actual number of evacuees can be compared with the predicted number of evacuees in the demand forecast result, so as to obtain the difference between the actual value of the emergency site after this dispatch and the demand forecast result, and determine whether to update the demand forecast result and the resource allocation plan based on the difference. If the demand forecast result and the resource allocation plan need to be updated, the above steps S201-S204 can be re-executed to generate a resource allocation plan that meets the current actual needs.

[0085] In the embodiment of the present application, by collecting various types of actual data from multiple emergency sites and realizing the fusion of multiple data sources, evacuation decisions and resource allocation decisions can be made more reliable in complex environments. By constructing a spatiotemporal graph structure for actual data, the spatiotemporal relationship between emergency sites can be more comprehensively reflected, thereby providing a global perspective for demand forecasting and resource allocation. By performing demand forecasting through a demand forecasting model, the demand distribution of each emergency site in the future can be obtained, which can better reflect the personnel flow characteristics and material demand fluctuations in each emergency site, thereby improving the system response speed and resource allocation efficiency. By generating a resource allocation plan based on the demand forecast results through a resource allocation model, it is possible to ensure balanced resource use in multiple emergency sites and avoid overload or underutilization of a single emergency site. Compared with the traditional method of evenly dividing resources, this application achieves the maximization of resource utilization in emergency sites.

[0086] The following is a further explanation of the above-mentioned input of the spatiotemporal graph structure into the pre-trained demand forecasting model, and the prediction of the demand forecast results corresponding to each node in the spatiotemporal graph structure at the current time by the spatiotemporal graph convolutional network and variational autoencoder in the demand forecasting model. Figure 4 As shown, the above step S203 includes:

[0087] S401. Input the spatiotemporal graph structure into the spatiotemporal graph convolutional network. Each convolutional layer in the spatiotemporal graph convolutional network performs convolution processing in sequence to obtain a demand prediction matrix of the spatiotemporal graph structure. Each convolutional layer includes a graph convolution sublayer and a time convolution sublayer.

[0088] Optionally, the node attribute matrix and edge weight matrix in the spatiotemporal graph structure can be input into the spatiotemporal graph convolutional network. The spatiotemporal graph convolutional network includes N graph convolution sublayers, N time convolution sublayers, and an output layer. Each graph convolution sublayer is connected to a time convolution sublayer, and N is a positive integer. The graph convolution sublayer is used to model the spatial relationship of the spatiotemporal graph structure, and the time convolution sublayer is used to capture the changes in the time dimension, superimposing the temporal features on the spatial features, thereby constructing the spatiotemporal relationship of each emergency site.

[0089] The formula for convolution processing of the graph convolution sublayer can be expressed as the following formula (1).

[0090]

[0091] Among them, H (l) is the node feature of the l-th layer of graph convolution, A is the adjacency matrix, D is the degree matrix, W (l) is the weight matrix.

[0092] The time convolution sublayer performs convolution processing on the convolution results output by the graph convolution sublayer, thereby superimposing time series features on the convolution results. After the last time convolution sublayer sends the convolution results to the output layer, the output layer can output the demand forecast matrix for a period of time in the future, including the predicted number of evacuees and the predicted material demand for each emergency site.

[0093] S402, input the demand prediction matrix into the variational autoencoder, the encoder in the variational autoencoder performs encoding processing to obtain a coding vector, and input the coding vector into the decoder, the decoder performs decoding processing to obtain the demand prediction result corresponding to each node in the space-time graph structure at the current time.

[0094] The mathematical representation of the variational autoencoder can be shown as follows: The variational autoencoder can receive the demand prediction matrix and generate the distribution of the latent variable Z for modeling the uncertainty of demand. Where μ(X)σ(X) is the mean and standard deviation generated by the encoder.

[0095] Z~q(Z|X)=N(μ(X),σ(X) 2 ) (2)

[0096] The loss function of the variational autoencoder can be shown as follows: Equation (3). The loss function includes reconstruction loss and KL divergence loss, which are used to balance prediction accuracy and uncertainty. q(z|X) [logp(X|z)] represents the reconstruction loss, D KL (q(z|X)||p(z)) denotes the KL divergence loss.

[0097] L=E q(z|X) [log(X|z)]-D KL (q(z|X)||p(z)) (3)

[0098] The decoder can reconstruct the demand distribution of each emergency site in the future by sampling the latent variable Z. This distribution generates resource requirements under different demand scenarios to meet different prediction scenarios.

[0099] The following is a further explanation of the above-mentioned inputting the demand forecast result corresponding to the current time into the resource allocation model, and obtaining the resource allocation plan corresponding to the current time by the resource allocation model prediction, including:

[0100] The resource allocation model is run in a preset time period, and the current demand forecast results and resource inventory of each emergency site in the current time period are input into the resource allocation model. The resource allocation model uses the allocation quantity and total inventory of each emergency site as constraints, and determines the resource allocation plan for each emergency site according to the current demand forecast results and resource inventory.

[0101] Optionally, the second time period may be used as a preset time period, and the current demand forecast result in the current time period may be the demand forecast result of the corresponding time of the current time period in the first time period. Exemplarily, the preset time period may be half an hour, and the demand forecast results for the next six hours are generated in the above step S203. When generating a resource allocation plan, the resource allocation model may be run every half an hour, and the demand forecast results for the next half an hour and the current resource inventory are input into the resource allocation model to generate a resource allocation plan for the current time period, i.e., the next half an hour.

[0102] The allocation of emergency sites can be the amount of materials allocated to the emergency sites, and the total inventory can be the remaining amount of materials in all emergency sites and material supply points. It should be understood that in order to ensure the rationality of material allocation and the supply of materials to each emergency site, the resource demand of each emergency site should be less than the allocation, and the total allocation of resources to all emergency sites should be less than the total inventory.

[0103] The current demand forecast results and resource inventory are input into the resource allocation model. The resource allocation model can predict the resource allocation plan within the current time period. The resource allocation plan is used to indicate the material dispatch plan and personnel evacuation plan for each emergency site within the current time period.

[0104] Among them, the resource allocation model is used to allocate resources for predicted demand to ensure that the allocation cost is optimized to the greatest extent when resources are sufficient.

[0105] The following is a further explanation of the resource allocation model that uses the allocation amount and total inventory of each emergency site as constraints to determine the resource allocation plan for each emergency site based on the current demand forecast results and resource inventory. Figure 5 As shown, the step includes:

[0106] S501. Determine the demand priority of each emergency location based on a preset weight coefficient, current demand forecast results, and resource inventory.

[0107] The demand priority is used to describe the demand level and urgency of the emergency site. The higher the demand priority, the higher the demand for the emergency site and the higher the urgency of the demand. The calculation method of the demand priority can be shown in the following formula (4).

[0108] priority(v i ) = α·demand + β·remaining resources (4)

[0109] Wherein, α and β may be constants, indicating preset weight coefficients. The demand may be the demand indicated by the current demand forecast result, and the remaining resources may be the resource stock at the emergency point.

[0110] S502: Determine the allocation quantity of each emergency site based on the demand priority of each emergency site, the current demand forecast result and the current resource inventory.

[0111] The resource allocation model can construct a cost minimization objective function and determine the allocation amount of each emergency site. The cost minimization objective function is shown in the following formula (5).

[0112]

[0113] Among them, c i is the unit transportation cost of distributing materials to shelters, x i For the allocation amount.

[0114] The constraints of the resource allocation model include demand satisfaction constraints and inventory constraints. The demand satisfaction precondition is to ensure that the demand for each emergency site is d i Satisfy the allocation x i , which can be specifically expressed as the following formula (6). The inventory constraint is to ensure that the total allocation of all emergency sites cannot exceed the current total inventory S, which can be specifically expressed as the following formula (7).

[0115]

[0116] The resource allocation model is constrained by equations (6) and (7). Based on the minimization cost objective function, a resource scheduling plan can be generated. The resource scheduling plan indicates the allocation amount of each emergency site and the resource allocation method of each emergency site.

[0117] The following is a further explanation of the above-mentioned acquisition of actual data of each emergency location at the current time. Figure 6 As shown, the above step S201 includes:

[0118] S601. Obtain actual demand data corresponding to at least one historical time and benchmark demand forecast results corresponding to each historical time.

[0119] If an emergency occurs, such as an increase in the number of people to be evacuated or a reduction in supplies, the accuracy of the baseline demand forecast results will decrease. In order to ensure that changes in demand are responded to in a timely manner, the actual demand data of each emergency site can be obtained after the scheduling is completed, and the actual demand data can be compared with the baseline demand forecast results for the same time.

[0120] Exemplarily, actual demand data at time T1 after scheduling is completed may be obtained, and the actual demand data may be compared with the baseline demand forecast result at time T1 to determine whether an emergency occurs.

[0121] S602: Compare the actual demand data corresponding to the historical time with the benchmark demand forecast result corresponding to the historical time to obtain a comparison result.

[0122] Optionally, the actual evacuee personnel data in the actual demand data may be compared with the predicted number of evacuees in the benchmark demand forecast result, and the actual material demand in the actual demand data may be compared with the predicted material demand in the benchmark demand forecast result to obtain a comparison result.

[0123] The comparison result is used to indicate a first difference between the actual number of evacuees and the predicted number of evacuees, and a second difference between the actual material demand and the predicted material demand.

[0124] S603: Analyze and process the comparison result to determine whether to start rolling prediction.

[0125] If the absolute value of the first difference is greater than the first difference threshold, and / or the absolute value of the second difference is greater than the second difference threshold, rolling prediction is enabled.

[0126] Among them, the absolute value of the first difference is greater than the first difference threshold, indicating that the number of evacuated personnel has suddenly increased or decreased, and the absolute value of the second difference is greater than the second difference threshold, indicating that the demand for materials has increased or decreased significantly. At this time, it is necessary to re-forecast the demand and update the resource allocation plan.

[0127] S604: If yes, obtain the actual data of each emergency location at the current time.

[0128] By comparing actual demand data with baseline demand forecast data, we can respond promptly when emergencies occur, update demand forecast results and resource allocation plans for each emergency site, and improve the efficiency of the system's response to dynamic demand.

[0129] As another possible implementation method, the user may also report the emergency to the system. Once the system receives the emergency report information, it starts rolling forecasting, re-acquires the actual data of the emergency site at the current time, and re-executes the above steps S201-S205 to timely update the current resource allocation plan and achieve timely response to the emergency.

[0130] The following is a further explanation of the process of determining the above baseline demand forecast results. Figure 7 As shown, the process includes:

[0131] S701. Obtain historical data of each emergency site at a historical time, the historical data including: personnel density of the site, remaining capacity of the site, and resource inventory of the site.

[0132] When the system starts running, it can obtain historical data of all historical times before the current moment. The historical data include: the density of people in each emergency site at historical time, the remaining capacity at each historical time, and the resource inventory.

[0133] S702. Construct a historical graph structure based on historical data. The historical graph structure includes multiple nodes and edges connecting the nodes. The nodes are used to represent emergency sites. The attribute information of the nodes includes: the personnel density of the site, the remaining capacity of the site, and the resource stock of the site. The edges are used to represent the traffic information between the emergency sites corresponding to the two nodes connected by the edge.

[0134] Optionally, the historical graph structure includes a node set, an edge set, a node attribute information set, and an edge attribute information set. The process of constructing the historical graph structure can be the same as the process of constructing the spatiotemporal graph structure in the above step S202, extracting each emergency place as a node from the historical data, extracting the data of each emergency place to obtain the attribute information of the node, extracting the traffic information of each emergency place to obtain the edges between the nodes and the attribute information of the edges, and the specific process is not repeated here.

[0135] S703: Input the historical graph structure into a pre-trained demand forecasting model, and use the spatiotemporal graph convolutional network and variational autoencoder in the demand forecasting model to predict the benchmark demand forecast results corresponding to each node in the historical graph structure.

[0136] The node attribute information set and the edge attribute information set in the historical graph structure are input into the demand forecasting model, and the spatiotemporal graph convolutional network in the demand forecasting model performs convolution processing to obtain the demand forecasting matrix, and the demand forecasting matrix is ​​input into the variational autoencoder, and the variational autoencoder performs encoding processing to obtain the encoded vector, and finally the encoded vector is input into the decoder of the model to obtain the benchmark demand forecasting results corresponding to each node in the historical graph structure.

[0137] Among them, the benchmark demand prediction result corresponding to each node in the historical graph structure can be the demand distribution of each emergency site in the first period in the future, including the predicted number of evacuees and the predicted number of required materials.

[0138] It should be noted that after obtaining the benchmark demand forecast results, a resource allocation plan for that time can be generated based on the demand forecast results of each node at each time in the demand forecast results, and the actual demand data at that moment can be obtained when that moment is reached. The actual demand data is compared with the benchmark demand forecast results. If the difference is large, the actual data is re-acquired and the benchmark demand forecast data is generated. If the difference is not large, the resource allocation plan for the next second time period in the future can be generated based on the benchmark demand forecast data.

[0139] This application also includes a regional sharing mechanism. When demand surges or exceeds the forecast value, the system can allocate resources between emergency sites to ensure the timely supply of key resources and avoid resource shortages in high-demand emergency sites. The following is a further explanation of the regional sharing mechanism, such as Figure 8 As shown, including:

[0140] S801. Construct a shared resource pool based on the resource inventory and demand forecast results of each emergency site at the current time. The shared resource pool includes: surplus resources of each emergency site.

[0141] Among them, the initial resources in the shared resource pool come from the surplus allocation of each emergency site. Each emergency site can report excess resources to the shared pool based on its material inventory and demand status, forming a centrally managed allocable resource pool.

[0142] Specifically, if the demand forecast result indicates that the material is not in short supply, and the material inventory indicates that the material has surplus, the material can be reported to the shared resource pool. If there is surplus material after reserving the material based on the demand forecast result, the surplus material can be reported to the shared resource pool.

[0143] S802: Determine whether to trigger regional sharing based on demand forecast results of each emergency location and a preset threshold.

[0144] As a possible implementation method, if the demand forecast results indicate that the material demand exceeds the preset allocation threshold, it can be determined that regional sharing needs to be triggered, or if the actual demand data of the emergency site indicates that the material demand exceeds n% of the current inventory, it can also be determined that regional sharing needs to be triggered, where n is any positive integer set by the user.

[0145] S803. If so, determine the sharing priority of each emergency site, determine at least one target emergency site from each emergency site according to the sharing priority, determine the resources to be scheduled for each target emergency site from the shared resource pool based on the automatic scheduling algorithm, and generate a resource scheduling plan for each resource to be scheduled.

[0146] Among them, the sharing priority is used to indicate the order of resource scheduling after the regional sharing mechanism is turned on. The sharing priority can be determined based on factors such as the urgency of the emergency site, the demand, and the distance between the emergency site and the preset point. The following formula (8) is a calculation formula for the sharing priority, where γ and δ are settable constants, representing the control coefficient. The target emergency site can be an emergency site with a higher sharing priority.

[0147] priority(v i )=γ·demand+δ·distance (8)

[0148] If regional sharing is triggered, the allocation scheme of the shared resource pool can be determined based on the automatic scheduling algorithm shown in the following formula (9). The goal of the algorithm is to give priority to shelters with scarce resources and minimize the transportation time and allocation cost of resources.

[0149]

[0150] Among them, T i It represents the amount of resources allocated to the i-th shelter in the shared resource pool.

[0151] S804: Schedule each resource to be scheduled according to each resource scheduling scheme.

[0152] As a possible implementation method, the resource scheduling problem between emergency sites can be modeled as a network flow problem, and the objective function is shown in the following equation (10), where f i , j For resources from v i to v j The amount of preparation, c i , j The unit transportation cost.

[0153]

[0154] After determining the scheduling scheme for each resource to be scheduled based on the above objective function, the resources can be scheduled according to the resource quantity and scheduling path indicated by the scheduling scheme.

[0155] After completing the evacuation task, the application can also optimize the model based on the actual scheduling results and the prediction results, such as Fig. 9 As shown, specifically including:

[0156] S901. Acquire actual demand data, and obtain prediction errors based on the actual demand data and each demand prediction result.

[0157] The actual demand data includes the actual number of evacuees and the actual number of materials required. An evacuation mission includes multiple dispatches. After each evacuation mission is completed, the prediction error can be obtained based on the actual demand data of each emergency site during the evacuation mission and the demand prediction results.

[0158] S902: If the prediction error is greater than a preset error threshold, the demand prediction model is trained according to the actual demand data and each demand prediction result to obtain an optimized demand prediction model.

[0159] If the prediction error is greater than the preset error threshold, the actual demand and prediction difference data can be used as training samples to update the parameters of the spatiotemporal graph convolutional network and variational autoencoder in the demand forecasting model through retraining. The reconstruction error in the variational autoencoder is used to adjust the modeling accuracy of the demand distribution. The convolution parameters of the spatiotemporal graph convolutional network are updated to improve the ability to capture dynamic demand characteristics. For example, a higher weight is given to the error part that exceeds the tolerance range, so that the model can adapt to abnormal demand changes more quickly.

[0160] This application optimizes the model based on actual demand data and demand forecast results. The optimized model can better reflect the personnel flow characteristics and material demand fluctuations under various demand scenarios. After multiple retrainings, the model can gradually adapt to different evacuation scenarios and demand patterns. And after the system applies historical feedback information to the model optimization, it provides support for demand forecasting of a new round of evacuation events, improving system response speed and resource allocation efficiency.

[0161] Based on the same inventive concept, the embodiment of the present application also provides an emergency evacuation information processing device corresponding to the emergency evacuation information processing method, and the device is deployed in Figure 1 In the electronic device, since the principle of solving the problem by the device in the embodiment of the present application is similar to the above-mentioned emergency evacuation information processing method in the embodiment of the present application, the implementation of the device can refer to the implementation of the method, and the repeated parts will not be repeated.

[0162] Fig.10 A schematic diagram of the structure of an emergency evacuation information processing device provided in an embodiment of the present application is shown, and the device includes: an acquisition module 1001, a construction module 1002, a demand prediction module 1003, a solution prediction module 1004 and a scheduling module 1005.

[0163] The acquisition module 1001 is used to acquire the actual data of each emergency site at the current time, and the actual data includes: the personnel density of the site, the remaining capacity of the site, and the resource inventory of the site;

[0164] A construction module 1002 is used to construct a spatiotemporal graph structure according to actual data. The spatiotemporal graph structure includes multiple nodes and edges connecting the nodes. The nodes are used to represent emergency sites. The attribute information of the nodes includes: the personnel density of the site, the remaining capacity of the site, and the resource stock of the site. The edges are used to represent the traffic information between the emergency sites corresponding to the two nodes connected by the edge.

[0165] The demand forecasting module 1003 is used to input the spatiotemporal graph structure into a pre-trained demand forecasting model, and the spatiotemporal graph convolutional network and variational autoencoder in the demand forecasting model predict the demand forecast results corresponding to each node in the spatiotemporal graph structure at the current time;

[0166] The solution prediction module 1004 is used to input the demand prediction result corresponding to the current time into the resource allocation model, and the resource allocation model predicts the resource allocation solution corresponding to the current time;

[0167] The scheduling module 1005 is used to schedule emergency evacuation resources for the emergency sites corresponding to each node according to the resource allocation plan.

[0168] Optionally, the demand forecasting module 1003 is specifically used for:

[0169] The spatiotemporal graph structure is input into the spatiotemporal graph convolutional network, and each convolutional layer in the spatiotemporal graph convolutional network performs convolution processing in sequence to obtain the demand prediction matrix of the spatiotemporal graph structure. Each convolutional layer includes a graph convolution sublayer and a time convolution sublayer respectively.

[0170] The demand prediction matrix is ​​input into the variational autoencoder, which is encoded by the encoder in the variational autoencoder to obtain a coding vector, and the coding vector is input into the decoder in the variational autoencoder to be decoded by the decoder to obtain the demand prediction result corresponding to each node in the spatiotemporal graph structure at the current time.

[0171] Optionally, the solution prediction module 1004 is specifically used for:

[0172] The resource allocation model is run in a preset time period, and the current demand forecast results and resource inventory of each emergency site in the current time period are input into the resource allocation model. The resource allocation model uses the allocation quantity and total inventory of each emergency site as constraints, and determines the resource allocation plan for each emergency site according to the current demand forecast results and resource inventory.

[0173] Optionally, the solution prediction module 1004 is specifically used for:

[0174] Determine the demand priority of each emergency site based on the preset weight coefficient, current demand forecast results and resource inventory;

[0175] The allocation quantity for each emergency site is determined based on the demand priority of each emergency site, the current demand forecast results and the current resource inventory.

[0176] Optionally, the device further comprises a comparison module, which is used to:

[0177] Obtaining actual demand data corresponding to at least one historical time and benchmark demand forecast results corresponding to each historical time;

[0178] Compare the actual demand data corresponding to the historical time with the benchmark demand forecast result corresponding to the historical time to obtain a comparison result;

[0179] Analyze and process the comparison results to determine whether to start rolling forecast;

[0180] If so, obtain the actual data of each emergency location at the current time.

[0181] Optionally, the acquisition module 1001 is further used for:

[0182] Obtain historical data of each emergency site at historical time, including: personnel density of the site, remaining capacity of the site, and resource inventory of the site;

[0183] The construction module 1002 is further used to: construct a historical graph structure according to historical data, wherein the historical graph structure includes a plurality of nodes and edges connecting the nodes, wherein the nodes are used to represent emergency sites, and the attribute information of the nodes includes: the personnel density of the site, the remaining capacity of the site, and the resource stock of the site, and the edges are used to represent the traffic information between the emergency sites corresponding to the two nodes connected by the edge;

[0184] The demand forecasting module 1003 is also used to: input the historical graph structure into a pre-trained demand forecasting model, and obtain the benchmark demand forecasting results corresponding to each node in the historical graph structure by the spatiotemporal graph convolutional network and variational autoencoder in the demand forecasting model.

[0185] Optionally, the device further includes a sharing module, configured to:

[0186] According to the resource inventory and demand forecast results of each emergency site at the current time, a shared resource pool is constructed, which includes: the surplus resources of each emergency site;

[0187] Determine whether to trigger regional sharing based on demand forecast results of each emergency location and preset thresholds;

[0188] If so, determine the sharing priority of each emergency site, determine at least one target emergency site from each emergency site according to the sharing priority, determine the resources to be scheduled for each target emergency site from the shared resource pool based on the automatic scheduling algorithm, and generate a resource scheduling plan for each resource to be scheduled;

[0189] Schedule each resource to be scheduled according to each resource scheduling plan.

[0190] Optionally, the device further comprises an optimization module, configured to:

[0191] Obtain actual demand data, and obtain prediction errors based on the actual demand data and each demand prediction result;

[0192] If the prediction error is greater than the preset error threshold, the demand prediction model is trained according to the actual demand data and each demand prediction result to obtain an optimized demand prediction model.

[0193] In the embodiment of the present application, by collecting various types of actual data from multiple emergency sites and realizing the fusion of multiple data sources, evacuation decisions and resource allocation decisions can be made more reliable in complex environments. By constructing a spatiotemporal graph structure for actual data, the spatiotemporal relationship between emergency sites can be more comprehensively reflected, thereby providing a global perspective for demand forecasting and resource allocation. By performing demand forecasting through a demand forecasting model, the demand distribution of each emergency site in the future can be obtained, which can better reflect the personnel flow characteristics and material demand fluctuations in each emergency site, thereby improving the system response speed and resource allocation efficiency. By generating a resource allocation plan based on the demand forecast results through a resource allocation model, it is possible to ensure balanced resource use in multiple emergency sites and avoid overload or underutilization of a single emergency site. Compared with the traditional method of evenly dividing resources, this application achieves the maximization of resource utilization in emergency sites.

[0194] Fig.11 A structural schematic diagram of an electronic device provided in an embodiment of the present application is shown, including: a processor 1101, a storage medium 1102 and a bus 1103, wherein the storage medium 1102 stores machine-readable instructions executable by the processor 1101, and when the electronic device runs an emergency evacuation information processing method such as the one in the embodiment, the processor 1101 communicates with the storage medium 1102 via the bus 1103, and the processor 1101 executes the machine-readable instructions, and the processor 1101 executes the preamble of the method item to execute the steps in the above-mentioned emergency evacuation information processing method.

[0195] An embodiment of the present application further provides a computer-readable storage medium, on which a computer program is stored. The computer program is executed when a processor is running, and the processor executes the steps in the above-mentioned emergency evacuation information processing method.

[0196] In the embodiment of the present application, the computer program can also execute other machine-readable instructions when run by the processor to execute other methods described in the embodiment. For the specific execution method steps and principles, please refer to the description of the embodiment, which will not be repeated here.

[0197] In the embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.

[0198] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0199] In addition, each functional unit in the embodiments provided in the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0200] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0201] It should be noted that similar numbers and letters represent similar items in the following figures. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. In addition, the terms "first", "second", "third", etc. are only used to distinguish the description and are not to be understood as indicating or implying relative importance.

[0202] Finally, it should be noted that the above-described embodiments are only specific implementation methods of the present application, which are used to illustrate the technical solutions of the present application, rather than to limit them. The protection scope of the present application is not limited thereto. Although the present application is described in detail with reference to the above-mentioned embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the above-mentioned embodiments within the technical scope disclosed in the present application, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application. They should all be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be based on the protection scope of the claims.

Claims

1. A method for processing emergency evacuation information, characterized in that: include: Acquire actual data of each emergency site at the current time, the actual data including: personnel density of the site, remaining capacity of the site, and resource inventory of the site; Constructing a spatiotemporal graph structure according to the actual data, wherein the spatiotemporal graph structure includes a plurality of nodes and edges connecting the nodes, wherein the nodes are used to represent emergency sites, and the attribute information of the nodes includes: the personnel density of the sites, the remaining capacity of the sites, and the resource stock of the sites, and the edges are used to represent the traffic information between the emergency sites corresponding to the two nodes connected by the edges; Input the spatiotemporal graph structure into a pre-trained demand forecasting model, and use the spatiotemporal graph convolutional network and variational autoencoder in the demand forecasting model to predict the demand forecast result corresponding to each node in the spatiotemporal graph structure at the current time; Inputting the demand forecast result corresponding to the current time into the resource allocation model, and using the resource allocation model to predict the resource allocation plan corresponding to the current time; According to the resource allocation plan, emergency evacuation resources are dispatched for the emergency places corresponding to each of the nodes.

2. The method according to claim 1, characterized in that The step of inputting the spatiotemporal graph structure into a pre-trained demand forecasting model, and predicting the demand forecasting result corresponding to each node in the spatiotemporal graph structure at the current time by the spatiotemporal graph convolutional network and the variational autoencoder in the demand forecasting model includes: Inputting the spatiotemporal graph structure into the spatiotemporal graph convolutional network, and performing convolution processing in sequence by each convolutional layer in the spatiotemporal graph convolutional network to obtain a demand prediction matrix of the spatiotemporal graph structure, wherein each convolutional layer includes a graph convolution sublayer and a time convolution sublayer; The demand prediction matrix is ​​input into the variational autoencoder, and the encoder in the variational autoencoder performs encoding processing to obtain an encoding vector, and the encoding vector is input into the decoder in the variational autoencoder, and the decoder performs decoding processing to obtain the demand prediction result corresponding to each node in the space-time graph structure at the current time.

3. The method according to claim 1, characterized in that: The step of inputting the demand forecast result corresponding to the current time into a resource allocation model, and obtaining the resource allocation scheme corresponding to the current time by prediction by the resource allocation model, comprises: The resource allocation model is run in a preset time period, and the current demand forecast results of each emergency site in the current time period and the resource inventory are input into the resource allocation model. The resource allocation model uses the allocation quantity and total inventory of each emergency site as constraints, and determines the resource allocation plan for each emergency site according to the current demand forecast results and resource inventory.

4. The method according to claim 3, characterized in that: The resource allocation model uses the allocation amount and total inventory of each emergency site as constraints, and determines the demand priority and resource allocation plan of each emergency site according to the current demand forecast result and resource inventory, including: Determining the demand priority of each of the emergency sites based on a preset weight coefficient, the current demand forecast result, and the resource inventory; The allocation quantity of each emergency site is determined based on the demand priority of each emergency site, the current demand forecast result and the current resource inventory.

5. The method according to claim 1, characterized in that The obtaining of actual data of each emergency location at the current time includes: Obtaining actual demand data corresponding to at least one historical time and a baseline demand forecast result corresponding to each of the historical times; Comparing the actual demand data corresponding to the historical time with the benchmark demand forecast result corresponding to the historical time to obtain a comparison result; Analyzing and processing the comparison result to determine whether to start rolling prediction; If so, obtain the actual data of each emergency location at the current time.

6. The method according to claim 5, characterized in that The process of determining the baseline demand forecast result includes: Obtaining historical data of each emergency site at a historical time, the historical data including: the personnel density of the site, the remaining capacity of the site, and the resource inventory of the site; Constructing a historical graph structure according to the historical data, wherein the historical graph structure includes a plurality of nodes and edges connecting the nodes, wherein the nodes are used to represent emergency sites, and the attribute information of the nodes includes: the personnel density of the sites, the remaining capacity of the sites, and the resource stock of the sites, and the edges are used to represent the traffic information between the emergency sites corresponding to the two nodes connected by the edges; The historical graph structure is input into a pre-trained demand forecasting model, and the spatiotemporal graph convolutional network and variational autoencoder in the demand forecasting model predict the benchmark demand forecast results corresponding to each node in the historical graph structure.

7. The method according to claim 1, characterized in that The method further comprises: According to the resource inventory and demand forecast results of each of the emergency sites at the current time, a shared resource pool is constructed, wherein the shared resource pool includes: the surplus resources of each of the emergency sites; Determining whether to trigger regional sharing based on demand forecast results of each of the emergency sites and preset thresholds; If so, determining the sharing priority of each of the emergency sites, determining at least one target emergency site from each of the emergency sites according to the sharing priority, determining the resources to be scheduled of each of the target emergency sites from the shared resource pool based on the automatic scheduling algorithm, and generating a resource scheduling plan for each of the resources to be scheduled; The resources to be scheduled are scheduled according to the resource scheduling schemes.

8. The method according to claim 1, characterized in that: After scheduling the resource allocation amount of each node according to the resource allocation scheme, the method further comprises: Acquire actual demand data, and obtain a prediction error based on the actual demand data and each of the demand prediction results; If the prediction error is greater than a preset error threshold, the demand prediction model is trained according to the actual demand data and each of the demand prediction results to obtain an optimized demand prediction model.

9. An electronic device, characterized in that: include: A processor, a storage medium and a bus, wherein the storage medium stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor communicates with the storage medium via the bus, and the processor executes the machine-readable instructions to perform the steps of the emergency evacuation information processing method as described in any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the emergency evacuation information processing method according to any one of claims 1 to 8 are executed.

Images