A disaster-specific emergency evacuation method based on spatial analysis algorithm

Abstract

A disaster-specific emergency evacuation method based on a spatial analysis algorithm belongs to the technical field of evacuation route planning. The invention method includes generating a dynamic risk distribution map; establishing a road structure damage prediction model; determining the gathering areas and needs of the affected population to be transferred; defining a search range with the gathering point of the affected people as the center to determine the refuge places that meet the evacuation needs; generating a spatial topological relationship map; constructing a dynamic evacuation model for disaster events; and continuously accessing the base station positioning data stream to dynamically trigger route replanning. The emergency evacuation route selection scheme designed by the present invention based on the background of multiple disasters can not only effectively cope with complex and changeable multi-disaster superposition scenarios, but is also fully applicable to emergency responses under the background of a single disaster, realizing evacuation priority grading of evacuation routes to ensure that evacuees can safely reach the evacuation destination. Optimization calculations based on dynamic personnel data can effectively improve the dynamic and disaster adaptability of the evacuation route planning model.

Description

Technical Field

[0001] The present invention relates to the technical field of evacuation route planning, and in particular to a disaster-based emergency evacuation method based on a spatial analysis algorithm. Background Art

[0002] While cities sustain direct damage from earthquakes, they are often accompanied by secondary disasters such as urban flooding caused by ruptured underground pipelines and landslides and mudslides caused by geological structural instability. Traditional disaster preparedness plans focused on a single type of disaster are no longer adequate to meet these demands. A lack of precise evacuation route guidance during the evacuation of affected residents to emergency shelters can lead them into areas threatened by secondary hazards, exacerbating the consequences. Therefore, establishing a system for optimizing emergency evacuation routes for residents, tailored to each type of disaster, has become a key measure for enhancing urban disaster resilience.

[0003] After searching, the Chinese patent with the publication number CN117196128A: A method, device and electronic equipment for planning emergency evacuation routes, its technical solution Figure 1 As shown in the figure, first, the initial evacuation path is determined based on the evacuation network. Secondly, real-time environmental information is obtained to determine whether the initial path is affected by the disaster and a feasible path is searched according to the disaster situation. Finally, the feasible path is used to replace and update the initial path to generate an optimized path.

[0004] The following deficiencies exist in existing emergency evacuation methods: existing evacuation route planning is mostly limited to a single disaster scenario, and does not consider dynamic factors such as road failure and secondary risks caused by the superposition of multiple disasters. It also relies too much on the shortest path or time optimization, and ignores the real-time changes in road capacity thresholds, congestion probability and environmental risks during the evacuation process; existing evacuation route planning mostly adopts a point-to-point model, and does not consider the upper limit of the capacity of emergency shelters. Continuing evacuation when overloaded can easily cause safety hazards, and can easily lead to inefficiency and imbalance in resource allocation under large-scale disasters; existing technologies use simulation and analysis of historical data or static data to carry out evacuation route planning. When a disaster occurs, road conditions often change rapidly, and the lag of static data makes it difficult to adjust the planning scheme in a timely manner, affecting evacuation efficiency and safety.

[0005] To address the above problems, a disaster-specific emergency evacuation method based on spatial analysis algorithm was proposed. Summary of the Invention

[0006] The purpose of the present invention is to provide a disaster-specific emergency evacuation method based on a spatial analysis algorithm, and the present device is used to solve the problems in the above background.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a disaster-specific emergency evacuation method based on a spatial analysis algorithm, comprising the following steps:

[0008] S1: Determine the types of disasters and their interaction patterns, obtain disaster characteristic parameters, quantify the comprehensive risk level of single disasters or multiple disasters based on geographic information system spatial analysis, and generate dynamic risk distribution maps;

[0009] S2: Collect primary indicator parameters such as geological structure, road grade, construction age, and seismic resistance level, use the analytic hierarchy process to determine the weight coefficients of secondary indicator parameters, establish a road structure damage prediction model, derive a road structure damage degree score, and output the road residual capacity coefficient;

[0010] S3: Integrates mobile phone and base station data to monitor population distribution and migration in real time, identify the affected population concentration areas and scale, and prioritize evacuations based on risk maps to determine the affected population concentration areas and needs.

[0011] S4: Determine the search area centered on the gathering point of disaster victims, screen surrounding emergency shelters, match the number of disaster victims through capacity demand analysis, and ultimately determine the shelter that meets the evacuation needs;

[0012] S5: Based on the preset vector geographic data and the evacuation road distribution of the emergency shelter, a spatial topology analysis algorithm is used to generate a spatial topology relationship diagram;

[0013] S6: Based on the spatial topology relationship diagram, real-time road capacity data, disaster risk probability and shelter capacity are introduced, and a dynamic evacuation model of disaster events is constructed using a five-fold cost dynamic evacuation model;

[0014] S7: Continuously access base station positioning data streams to monitor actual evacuation progress and population distribution changes, and dynamically trigger path replanning.

[0015] Furthermore, the comprehensive hazard level calculation method described in step S1 is: integrating historical disaster data and real-time crowd intelligence perception information, combining the hierarchical analysis method to quantify the weights of the three major indicators of "disaster hazard-vulnerability of disaster-bearing bodies-disaster reduction capabilities", constructing a multi-disaster coupling evaluation model, and rasterizing the affected area based on the spatial analysis technology of the geographic information system, superimposing disaster site data, geographic environment parameters and dynamic disaster indicators, generating a disaster hazard zoning map, and finally outputting the regional comprehensive hazard level.

[0016] Furthermore, the secondary indicators corresponding to the geological structure in step S2 are rock type and terrain characteristics, the secondary indicators corresponding to the road grade are road length, road width, road surface condition and road surface material, the secondary indicator corresponding to the construction age is the construction year, and the secondary indicator corresponding to the seismic resistance grade is the road surface seismic resistance grade.

[0017] Furthermore, the method for constructing the road structure damage prediction model in step S2 is: first, integrate and standardize data such as geological structure, road attributes, construction age and seismic resistance level, then determine the weight of each indicator through the hierarchical analysis method, then use the graph neural network to extract the node characteristics of the geological structure and the edge characteristics of the road attributes, and combine the time series model to process the dynamic change characteristics of the construction age and seismic resistance level, and finally fuse the weighted features to construct a road structure damage prediction model, and output the damage score and residual traffic capacity coefficient.

[0018] Furthermore, the calculation formula of the road residual capacity coefficient in step S2 is: road residual capacity coefficient=1-road structure damage degree / maximum bearing capacity.

[0019] Furthermore, the method for identifying disaster-stricken population gathering areas in step S3 is: building a deep learning model, using mobile phone user density data and geographic spatial information as input, and accurately classifying the population thermal distribution. After training and optimization with a large amount of sample data, the model can accurately mark high-density areas as "disaster-stricken population gathering areas."

[0020] Furthermore, the spatial topological relationship diagram in step S5 includes network nodes composed of gathering points of disaster-stricken people, emergency shelters and intersections in the road network, and edges composed of road sections in the road network. The predicted damaged sections are marked as disabled edges, and the edge weights of the damaged sections are dynamically adjusted according to the residual traffic capacity.

[0021] Furthermore, the method for constructing the dynamic evacuation model of disaster events in step S6 is: taking minimizing evacuation time, minimizing road disaster risk probability, maximizing shelter utilization capacity, maximizing path structure redundancy and dynamic adaptation weight adjustment as optimization objectives, and taking road traffic efficiency, disaster risk probability and shelter capacity limit as constraints, according to the optimization objectives and constraints, constructing a multi-objective function F(X)=min(T(N), R(N), Cc(N), Cr(N), Cy(N)), where N is the total number of evacuees, F(X) is the evacuation path optimization objective function, T(N) is the total evacuation time objective function, R(N) is the disaster risk probability objective function, Cc(N) is the evacuation flow objective function (avoiding exceeding the shelter capacity), Cr(N) is the shelter utilization capacity objective function, and Cy(N) is the path structure redundancy objective function.

[0022] Compared with the prior art, the present invention has the following beneficial effects:

[0023] 1. Taking multiple disaster types into consideration and comprehensively considering the impact of different disaster types on evacuation routes, the resulting evacuation plan and system are suitable for both single disaster scenarios and multi-disaster complex situations, and have wide applicability and practicality.

[0024] 2. Use road capacity, shelter capacity and path safety risk factors as evacuation road priority ranking indicators to achieve evacuation priority grading of evacuation paths, ensure smooth traffic and road safety during the evacuation process, and ensure that evacuees have sufficient accommodation space, greatly improving evacuation efficiency and safety.

[0025] 3. Taking into account the impact of earthquakes and other disasters on the road's damage degree and accessibility, a multi-objective comprehensive optimization of the evacuation route can be given.

[0026] 4. Mobile phone hotspot data within the evacuation area is used as shelter demand data. This data can more clearly reflect the actual characteristics of the crowd, providing more accurate data support for evacuation route planning. At the same time, using personnel positioning data from communication base stations, optimization calculations based on personnel dynamic data are performed, effectively improving the dynamic and disaster adaptability of the evacuation route planning model.

[0027] 5. The system constructed according to the method of the present invention can automatically calculate the optimal evacuation path plan by simply inputting initial data, which can significantly improve emergency response speed and decision-making efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0029] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0030] A disaster-specific emergency evacuation method based on a spatial analysis algorithm includes the following steps:

[0031] Step 1: Determine the disaster types and their interaction patterns, obtain disaster characteristic parameters, including disaster type, hazard level, and intensity index, quantify the comprehensive hazard level of a single disaster or multiple disasters based on geographic information system spatial analysis, and generate a dynamic risk distribution map;

[0032] Step 2: Collect primary indicator parameters such as geological structure, road grade, construction age, and seismic resistance level. Use the analytic hierarchy process to determine the weight coefficients of secondary indicator parameters, establish a road structure damage prediction model, derive a road structure damage degree score, and output the road residual capacity coefficient.

[0033] Step 3: Integrate mobile phone and base station data to monitor population distribution and migration in real time, identify the affected population concentration areas and scale, and prioritize evacuations based on risk maps to determine the affected population concentration areas and needs.

[0034] Step 4: Define a search range with the gathering point of disaster victims as the center, screen surrounding emergency shelters, match the number of disaster victims through capacity demand analysis, and finally determine the shelter that meets the evacuation needs. You can use the new location allocation tool of Network Analyst in ArcGIS to perform location allocation solution. In the model parameter setting, use the actual travel distance of the road network as the impedance parameter, adopt the "maximize capacity-constrained coverage" allocation mode, set the impedance cutoff value, and ensure that the path length between the demand point and the facility point does not exceed the service radius threshold. The specific location allocation solution process is as follows:

[0035] (1) Constructing a network dataset: Integrating road data with municipal traffic control information to construct a road network dataset that includes travel directions, one-way restrictions, and dynamic risk attributes;

[0036] (2) Loading constraints: Associating the maximum number of effective shelters at the facility point and demographic data of the demand point into the attribute table and embedding the location allocation tool;

[0037] (3) Run the solver: Using the “maximize capacity-constrained coverage” mode, with the actual travel distance as the impedance parameter, generate an allocation plan that maximizes the total population coverage through iterative optimization;

[0038] Step 5: Based on the preset vector geographic data and the evacuation road distribution of the emergency shelter, a spatial topology analysis algorithm is used to generate a spatial topology relationship diagram;

[0039] Step 6: Based on the spatial topology diagram, we introduce real-time road capacity data, disaster risk probability, and shelter capacity, and use the five-fold cost dynamic evacuation model to build a dynamic evacuation model for disaster events.

[0040] Step 7: Continuously access base station positioning data streams to monitor actual evacuation progress and population distribution changes, and dynamically trigger path replanning.

[0041] The calculation method for the comprehensive disaster risk level in step one is: integrating historical disaster data and real-time crowd intelligence perception information, combining the hierarchical analysis method to quantify the weights of the three major indicators of "disaster risk-vulnerability of disaster-bearing bodies-disaster reduction capabilities", building a multi-disaster coupling evaluation model, and rasterizing the affected areas based on the spatial analysis technology of the geographic information system. The disaster site data, geographic environment parameters and dynamic disaster indicators are superimposed to generate a disaster risk zoning map, and finally output the regional comprehensive risk level.

[0042] Specifically, this calculation model comprehensively considers the risk levels of multiple disasters, including "earthquake-urban flooding", "earthquake-geological disaster", and "earthquake-urban flooding-geological disaster". The calculation method for the comprehensive disaster risk level is:

[0043] Obtain regional historical disaster data, including disaster type, level, location, impact range, number of casualties, etc.;

[0044] By integrating the natural and socio-economic attributes of disasters and using the analytic hierarchy process, we can derive the influencing factors of "disaster hazard - vulnerability of disaster-bearing bodies - disaster reduction and relief capabilities" and further construct a multi-hazard assessment model.

[0045] Disaster site data collection based on crowd-sensing includes information such as disaster type, rescue situation at the work site, search and rescue and casualties, on-site evacuation, and infrastructure damage. It also collects disaster-related data and geographic information fed back by online groups.

[0046] From the disaster site information data fed back by the network, three indicators are selected to represent the disaster situation, namely disaster hazard, vulnerability of hazard-bearing bodies and disaster reduction and relief capabilities. The evaluation process is to divide the disaster-affected area into multiple grid cells, and carry out disaster hazard, vulnerability of hazard-bearing bodies and disaster reduction and relief capabilities evaluation respectively. ArcGIS software can be used to overlay the various indicator data and conduct spatial analysis. Taking into account the results of the three indicators, a disaster hazard zoning map of the affected area can be generated.

[0047] Based on geographic information, indicators representing the disaster-affected area are selected, and the geographic information is superimposed on the disaster risk zoning map through the geographic information system to obtain the comprehensive disaster risk level of the area. The indicators representing the disaster-affected area refer to the natural conditions of the disaster-stricken area, such as the topography, geological structure, and water environment.

[0048] By identifying earthquakes and the coupling of earthquakes with other secondary hazards, obtaining disaster characteristic parameters, and conducting network analysis of the affected area, we can quantify the comprehensive risk level of a single hazard or the combined effects of multiple hazards. This multi-hazard-based emergency evacuation route selection plan not only effectively addresses complex and changing multi-hazard scenarios, but is also fully applicable to emergency responses under a single hazard.

[0049] The secondary indicators corresponding to the geological structure in step 2 are rock type and terrain characteristics, the secondary indicators corresponding to the road grade are road length, road width, road surface condition and road surface material, the secondary indicator corresponding to the construction year is the construction year, and the secondary indicator corresponding to the seismic resistance level is the road surface seismic resistance level.

[0050] The method for constructing the road structure damage prediction model in step 2 is as follows: first, integrate and standardize data such as geological structure, road attributes, construction age and seismic resistance level, then determine the weight of each indicator through the hierarchical analysis method, then use the graph neural network to extract the node characteristics of the geological structure and the edge characteristics of the road attributes, and combine the time series model to process the dynamic change characteristics of the construction age and seismic resistance level, finally fuse the weighted features to construct the road structure damage prediction model, and output the damage score and residual capacity coefficient.

[0051] The calculation formula for the road residual capacity coefficient in step 2 is: road residual capacity coefficient = 1-road structure damage degree / maximum bearing capacity.

[0052] Specifically, first collect data on geological structure, road grade, construction age, and seismic resistance grade, and perform data standardization and normalization. Then, use the hierarchical analysis method to determine the weight coefficients of each secondary indicator parameter. Then, use the geological structure, road attributes, construction age and other parameters as input features of the model. Use the graph neural network and the graph convolution layer to extract node features from the geological structure data and edge features from the road attribute data. Use time series processing to process parameters such as construction age and seismic resistance grade as time coding features. Use the time-gated network to extract long-term dependency information, assign the determined weight coefficients to each layer of features, and establish a road structure damage prediction model. Finally, by inputting the four indicator data of geological structure, road attributes, construction age, and seismic resistance grade, the road structure damage degree score and the road residual capacity coefficient are obtained.

[0053] The method for identifying disaster-affected population clusters in step three is to build a deep learning model, use mobile phone user density data and geographic spatial information as input, and accurately classify the population thermal distribution. After training and optimization with a large amount of sample data, the model can accurately mark high-density areas as "disaster-affected population clusters."

[0054] Specifically, the team obtained mobile phone hotspot data from relevant departments, including timestamps, specific locations (latitude and longitude), and the number of mobile phone users. They also obtained base station location and coverage data from telecommunications operators, including timestamps, base station geographic locations (latitude and longitude), and the number of mobile phone users. The mobile phone hotspot data and communication base station positioning data were aligned according to the same time interval, and invalid or duplicate data was deleted to ensure data accuracy and uniqueness. The aligned data was processed using kernel density estimation methods. By setting bandwidth parameters within a certain range, the locations of affected population clusters were identified. Combined with spatial analysis techniques, the population density of each affected area was calculated. Mobile phone user movement path data, i.e., trajectory information from one base station to another, was used. Time series analysis was used to track mobile phone users' base station switching within different time periods, extracting complete migration trajectories. Using mobile phone user density data and geospatial information as input, the team accurately classified the population thermal distribution. After training and optimizing on a large amount of sample data, the model can accurately label high-density areas as "disaster-affected population clusters." It can assess regional population mobility by counting the number of mobile phone users entering and leaving an area within a certain time interval. It can also assess regional population stability by observing how mobile phone users stay in an area over a longer period of time. Combined with the population migration trajectory data extracted in the third step, time series analysis can be used to calculate population density trends for each time period. Using historical disaster population clusters, migration data, and population distribution data as training samples, a population migration model based on a long-short-term memory network is constructed, and model parameters are optimized through cross-validation.

[0055] The spatial topology relationship diagram in step five includes network nodes composed of gathering points of disaster-stricken people, emergency shelters, and intersections in the road network, and edges composed of road sections in the road network. The predicted damaged sections are marked as disabled edges, and the edge weights of the damaged sections are dynamically adjusted according to the residual traffic capacity.

[0056] Specifically, the ArcGIS Network Analyst module can be used to dynamically adjust the edge weights of damaged road sections according to the residual capacity, focusing on repairing breaks, overlaps, and connectivity errors in the road network. Based on the needs of urban traffic diversion, the topological relationship of vulnerable road sections is reconstructed and associated using a method combining spatial connection and field calculation. Dynamic risk parameters are dynamically updated according to the simulation model to update the traffic status and impedance weight of the road network and establish a multi-state road network topology (normal / partially damaged / completely damaged). The algorithm for the spatial topological relationship diagram of the emergency evacuation network is as follows:

[0057] Determine evacuation areas in the event of a disaster;

[0058] Obtain the number of people to be evacuated in the evacuation area and the distribution of gathering points of the affected people, and determine the evacuation traffic distribution of the evacuation area based on the distribution of gathering points of the affected people;

[0059] Based on the spatial analysis system, a path network spatial topology model is generated according to the preset vector geographic data and the evacuation traffic distribution of the evacuation area. The path network spatial topology model includes network nodes composed of disaster-stricken population gathering points, emergency shelter points, and intersections in the road network, and edges composed of road segments in the road network.

[0060] Using the ArcGIS Network Analyst module, and based on information on disaster risk points such as earthquakes, geological disasters, and urban flooding, predicted damaged road sections can be marked as disabled edges. Edge weights for damaged sections can be dynamically adjusted based on residual capacity, and topological relationships for vulnerable sections can be reconstructed based on urban traffic diversion needs.

[0061] The association is performed by combining spatial connection and field calculation. The dynamic risk parameters are updated dynamically according to the simulation model to dynamically update the traffic status and impedance weight of the road network and establish a multi-state road network topology (normal / partially damaged / completely damaged).

[0062] The construction method of the dynamic evacuation model for disaster events in step six is: taking minimizing evacuation time, minimizing road disaster risk probability, maximizing shelter utilization capacity, maximizing path structure redundancy and dynamic adaptation weight adjustment as optimization objectives, and taking road traffic efficiency, disaster risk probability and shelter capacity limit as constraints. According to the optimization objectives and constraints, a multi-objective function F(X)=min(T(N),R(N),Cc(N),Cr(N),Cy(N)) is constructed. In the multi-objective function, N is the total number of evacuees, F(X) is the evacuation path optimization objective function, T(N) is the total evacuation time objective function, R(N) is the disaster risk probability objective function, Cc(N) is the evacuation flow objective function (avoiding exceeding the shelter capacity), Cr(N) is the shelter utilization capacity objective function, and Cy(N) is the path structure redundancy objective function.

[0063] Specifically, considering the requirements for timeliness and safety in the actual process of emergency evacuation, the objectives are to minimize evacuation time, minimize the probability of road disaster risks, maximize the capacity of shelters, maximize path structure redundancy and dynamic adaptive weight adjustment, and to use road traffic efficiency (not exceeding the preset maximum value), disaster risk probability (not exceeding the preset threshold), and shelter capacity limit (not exceeding the number of people the shelter can accommodate) as constraints.

[0064] Collect data on population distribution, shelter distribution and capacity, and disaster risk. Based on the spatial topology model of the path network generated by the GIS system, set capacity attributes for emergency shelter points and set four attributes for edges consisting of road segments in the road network: capacity, transit time, disaster risk probability, and path structure redundancy.

[0065] A non-dominated sorting genetic algorithm II is introduced to randomly generate an initial population that meets the constraints. The performance of each individual in the five objective functions is calculated, and the population is sorted according to the Pareto optimal principle. All non-dominated solutions are retained, and crossover and mutation operations are performed to generate a new offspring population. The fitness calculation, non-dominated sorting and genetic operations are repeated until the convergence conditions are met. All Pareto optimal solutions are collected, the trade-off relationship between each objective function is analyzed, and the optimal or suboptimal path is selected according to actual needs.

[0066] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "includes," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0067] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A disaster-specific emergency evacuation method based on a spatial analysis algorithm, characterized in that: The following steps are involved: S1: Determine the types of disasters and their interaction patterns, obtain disaster characteristic parameters, quantify the comprehensive risk level of single disasters or multiple disasters based on geographic information system spatial analysis, and generate dynamic risk distribution maps; S2: Collect the first-level indicator parameters of geological structure, road grade, construction age, and seismic resistance level, use the hierarchical analysis method to determine the weight coefficients of the second-level indicator parameters, establish a road structure damage prediction model, derive a road structure damage degree score, and output the road residual capacity coefficient; S3: Integrates mobile phone and base station data to monitor population distribution and migration in real time, identify the affected population concentration areas and scale, and prioritize evacuations based on risk maps to determine the affected population concentration areas and needs. S4: Determine the search area centered on the gathering point of disaster victims, screen surrounding emergency shelters, match the number of disaster victims through capacity demand analysis, and ultimately determine the shelter that meets the evacuation needs; S5: Based on the preset vector geographic data and the evacuation road distribution of the emergency shelter, a spatial topology analysis algorithm is used to generate a spatial topology relationship diagram; S6: Based on the spatial topology relationship diagram, real-time road capacity data, disaster risk probability and shelter capacity are introduced, and a dynamic evacuation model of disaster events is constructed using a five-fold cost dynamic evacuation model; S7: Continuously access base station positioning data streams to monitor actual evacuation progress and population distribution changes, and dynamically trigger route replanning; The comprehensive hazard level calculation method described in step S1 is as follows: integrating historical disaster data and real-time crowd intelligence perception information, combining the analytic hierarchy process to quantify the weights of three indicators: disaster hazard, vulnerability of disaster-bearing bodies, and disaster mitigation capacity, constructing a multi-hazard coupling evaluation model, rasterizing the affected area based on geographic information system spatial analysis technology, superimposing disaster site data, geographic environment parameters, and dynamic disaster indicators, generating a disaster hazard zoning map, and finally outputting the comprehensive regional hazard level; The method for constructing the road structure damage prediction model described in step S2 is as follows: first, the geological structure, road attributes, construction age and seismic resistance level data are integrated and standardized, and then the weight of each indicator is determined by the hierarchical analysis method. Then, the node characteristics of the geological structure and the edge characteristics of the road attributes are extracted using the graph neural network, and the dynamic change characteristics of the construction age and seismic resistance level are processed in combination with the time series model. Finally, the weighted features are integrated to construct the road structure damage prediction model, and the damage score and residual traffic capacity coefficient are output.

2. The method for emergency evacuation based on disaster type according to claim 1, characterized in that: The secondary indicators corresponding to the geological structure in step S2 are rock type and terrain characteristics, the secondary indicators corresponding to the road grade are road length, road width, road surface condition and road surface material, the secondary indicator corresponding to the construction age is the construction year, and the secondary indicator corresponding to the seismic resistance grade is the road surface seismic resistance grade.

3. The method for emergency evacuation based on disaster type according to claim 1, characterized in that: The calculation formula of the road residual capacity coefficient in step S2 is: road residual capacity coefficient = 1 - road structure damage degree / maximum bearing capacity.

4. The method for emergency evacuation based on spatial analysis algorithm according to claim 1, characterized in that: The method for identifying disaster-affected population clusters in step S3 is to build a deep learning model, use mobile phone user density data and geographic spatial information as input, and accurately classify the population thermal distribution. After training and optimization with a large amount of sample data, the model can accurately mark high-density areas as disaster-affected population clusters.

5. The method for emergency evacuation based on disaster type and spatial analysis algorithm according to claim 1, characterized in that: The spatial topological relationship diagram described in step S5 includes network nodes composed of gathering points of disaster-stricken people, emergency shelters, and intersections in the road network, and edges composed of road sections in the road network. The predicted damaged sections are marked as disabled edges, and the edge weights of the damaged sections are dynamically adjusted according to the residual traffic capacity.

6. The method for emergency evacuation based on disaster type and spatial analysis algorithm according to claim 1, characterized in that: The method for constructing the dynamic evacuation model for disaster events in step S6 is: with the goals of minimizing evacuation time, minimizing road disaster risk probability, maximizing the utilization capacity of shelters, maximizing path structure redundancy and dynamic adaptation weight adjustment, and with road traffic efficiency, disaster risk probability and shelter capacity limit as constraints, according to the goals and constraints, construct a multi-objective function F(X)=min(T(N),R(N),Cc(N),Cr(N),Cy(N)).

7. The method for emergency evacuation based on disaster type based on spatial analysis algorithm according to claim 6, characterized in that: In the multi-objective function, N is the total number of evacuees, F(X) is the evacuation path optimization objective function, T(N) is the total evacuation time objective function, R(N) is the disaster risk probability objective function, Cc(N) is the evacuation flow objective function, Cr(N) is the shelter capacity objective function, and Cy(N) is the path structure redundancy objective function.

Images