Intelligent management and control system and method for emergency shelter based on multi-source information fusion

Abstract

The application discloses an emergency shelter intelligent management and control system and method based on multi-source information fusion, and relates to the technical field of geographic data processing.The application carries out data fusion on real-time scene data, establishes a dynamic scene monitoring model according to the data fusion result, establishes a historical disaster evolution model according to historical case data, compares the historical disaster evolution model with the dynamic scene monitoring model, selects a comparison case according to the comparison result, simulates the disaster trend in the dynamic scene monitoring model according to the comparison case, covers the simulation disaster trend result on the dynamic scene monitoring model, simulates crowd evacuation with the emergency shelter as a simulation end point on the dynamic scene monitoring model, generates a personnel evacuation route according to the crowd evacuation simulation result, and dynamically adjusts according to real-time scene data during the execution of the personnel evacuation route.

Description

Technical Field

[0001] This invention relates to the field of geographic data processing technology, specifically to an intelligent management and control system and method for emergency shelters based on multi-source information fusion. Background Technology

[0002] In today's society, various natural disasters and emergencies occur frequently, such as earthquakes, floods, and fires, posing a huge threat to people's lives and property. Emergency shelters, as crucial infrastructure for protecting the lives of disaster victims, require scientific and effective management.

[0003] Traditional methods for managing emergency shelters have many limitations. On the one hand, in terms of data acquisition, the past reliance on a single data source makes it difficult to comprehensively and accurately reflect the actual situation of emergency shelters and surrounding monitoring scenarios. For example, relying solely on camera images cannot capture data such as temperature, humidity, and gas concentration in the environment, resulting in an incomplete understanding of the scenario and thus affecting subsequent decision-making and management.

[0004] On the other hand, in disaster response, the lack of accurate prediction and simulation of disaster evolution trends means that traditional methods often rely on experience-based judgments, making it difficult to accurately simulate disaster trajectories based on real-time scenarios and historical case data. This results in a lack of scientific basis when formulating evacuation plans, easily leading to problems such as unreasonable evacuation routes and low evacuation efficiency. Furthermore, during evacuation, traditional methods cannot dynamically adjust evacuation routes based on real-time changing scenario data. In the event of emergencies such as road blockages or secondary disasters, evacuation work can become chaotic, increasing the risk of casualties and property damage. Therefore, this paper proposes a smart management system and method for emergency shelters based on multi-source information fusion. Summary of the Invention

[0005] The purpose of this invention is to provide an intelligent management and control system and method for emergency shelters based on multi-source information fusion, so as to solve the problems in the background technology.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] The intelligent management and control method for emergency shelters based on multi-source information fusion includes the following steps:

[0008] Step S1: Set up data sensing devices in various emergency shelters and monitoring scenarios, and obtain real-time scene data of various emergency shelters and monitoring scenarios through the data sensing devices. Perform data fusion on the real-time scene data from different data sensing devices, and establish a dynamic scene monitoring model based on the data fusion results.

[0009] Step S2: Obtain historical case data of various disasters, establish a historical disaster evolution model based on the historical case data, compare the historical disaster evolution model with the dynamic scene monitoring model, select comparison cases based on the comparison results, simulate the disaster trajectory in the dynamic scene monitoring model based on the comparison cases, and overlay the simulated disaster trajectory results onto the dynamic scene monitoring model.

[0010] Step S3: Based on the simulated disaster trajectory results on the dynamic scene monitoring model, conduct a population evacuation simulation with the emergency shelter as the simulation endpoint. Generate evacuation routes based on the population evacuation simulation results, and dynamically adjust them based on real-time scene data during the execution of the evacuation routes.

[0011] Furthermore, the process of setting up data sensing devices in emergency shelters and monitoring scenarios includes:

[0012] Data sensing devices are set up for emergency shelters and monitoring scenarios respectively. The data sensing devices for emergency shelters cover three data sensing directions: external environment, personnel status, and material reserves. The data sensing devices for monitoring scenarios cover three data sensing directions: external disasters, traffic conditions, and infrastructure.

[0013] The same data sensing frequency is set for the data sensing devices in emergency shelters and monitoring scenarios, and the same data sensing range is set for the data sensing devices in emergency shelters or monitoring scenarios. They are distributed in the emergency shelters or monitoring scenarios at the same spatial distance, so that the combined result of the overlapping part of the data sensing range of any data sensing device with all the data sensing devices in its adjacent spatial locations covers the data sensing range of that data sensing device, so that the combined range of the data sensing ranges of all data sensing devices covers the entire emergency shelter and monitoring scenario.

[0014] Furthermore, the process of data fusion for real-time scene data includes:

[0015] After all data sensing devices synchronize their time, they acquire real-time scene data within their data sensing range. Whenever a data sensing frequency is reached, the data sensing device converts the collected real-time scene data into JSON format, with fields including device number, acquisition time, data type, value, unit, and location coordinates.

[0016] The real-time scene data for emergency shelters includes weather data, personnel data, material data, and video data;

[0017] The real-time scene data monitored includes weather data, traffic data, and video data;

[0018] A three-dimensional coordinate system is constructed, and the acquisition time, data type, and location coordinates contained in the fields of each real-time scene data are mapped into the three-dimensional coordinate system, so that each real-time scene data is distributed in the three-dimensional coordinate system according to the spatiotemporal location order and data type.

[0019] A deviation threshold is set for each type of real-time scene data. If the data are in an overlapping state and the difference between the same real-time scene data from different data sensing devices is less than or equal to the deviation threshold, the average value of the two sets of real-time scene data is accumulated and used as the real-time scene data for the corresponding scene location.

[0020] If the difference is greater than the deviation threshold, the default value of the data sensing device within the corresponding collection time is incremented by 1. When the default value is greater than or equal to 3, the real-time scene data of the corresponding data type of the data sensing device within the corresponding collection time is deleted, and the real-time scene data collected by the remaining data sensing devices is used as the real-time scene data of the corresponding scene location.

[0021] When the failure value of the data sensing device associated with any scene location is greater than or equal to 3, the real-time scene data of the data type corresponding to that scene location is temporarily ignored.

[0022] Furthermore, the process of establishing a dynamic scene monitoring model includes:

[0023] Based on the video data in the real-time scene data collected at various data sensing frequencies, a visual 3D model is established. Then, the weather data of emergency shelters and monitoring scenes are overlaid on the visual 3D model in the form of textures. Each texture represents a numerical range in the image range it occupies.

[0024] Personnel and material data are directly labeled on the corresponding parts of the visualized 3D model of emergency shelters, while traffic data is overlaid on the visualized 3D model in the form of point cloud data, thus obtaining a dynamic scene monitoring model.

[0025] Furthermore, the process of establishing a historical disaster evolution model includes:

[0026] The historical case data includes disaster type names, evolution process data, impact result data, and emergency response data;

[0027] Feature extraction dimensions are set for various disaster types. Based on the feature extraction dimensions, historical feature data sets are extracted from each historical case data. Then, historical disaster evolution models corresponding to the historical case data are generated using each historical data in the historical feature data set as independent variable data and the data affecting the outcome as dependent variable data.

[0028] The dependent variable data in the historical disaster evolution model exists in the form of an image model, while the independent variable data and the weather and environmental data in the dynamic scene monitoring model are all overlaid on the dependent variable data in the form of textures.

[0029] The lowest values ​​of each independent variable in the historical disaster evolution model corresponding to various disaster types are obtained as the induced feature values.

[0030] Furthermore, the process of selecting comparative cases includes:

[0031] Based on the spatial location and number of data sensing devices, each dynamic scene monitoring model is divided into scene monitoring areas;

[0032] The triggering characteristic values ​​of various disaster types are compared with various real-time weather data in each scene monitoring area. If there is a real-time weather data that is greater than or equal to the triggering characteristic value, then the scene monitoring area is marked as abnormal weather data according to the type of real-time weather data; otherwise, no abnormal weather data is marked.

[0033] When the abnormal weather data labels in the scene monitoring area meet all feature extraction dimensions of a disaster type, disaster labels are set for the corresponding scene monitoring area according to the disaster type name.

[0034] Then, using the scene monitoring area with disaster labels as the monitoring location, the disaster label spread process of the scene monitoring area under more than 50 subsequent data sensing frequencies is obtained, and the historical disaster evolution model is retrieved and compared with the disaster label spread process based on the disaster labels;

[0035] The initial state of the dependent variable data in each historical disaster evolution model is compared with that of the dynamic scene monitoring model. At least 100 historical disaster evolution models that are most similar to the visualized 3D model in the dynamic scene monitoring model are selected as comparison cases, and then the comparison cases are simultaneously divided into several comparison scene areas.

[0036] Furthermore, the process of simulating the course of a disaster includes:

[0037] The real-time values ​​of various feature extraction dimensions corresponding to the scene monitoring area with disaster annotation and the scene disaster area in its adjacent location are obtained at the beginning and end of each data perception frequency. Then, the disaster spread vector is established with the disaster annotation spread direction as the vector direction and the real-time values ​​of various feature extraction dimensions corresponding to the adjacent scene monitoring areas with disaster annotation as scalars.

[0038] Starting from the first abnormal comparison scene area in the comparison case, obtain the comparison spread vector of the comparison scene area with the abnormality under each data perception frequency duration. Select the comparison spread vector and the disaster spread vector with the most similar vector direction to perform cosine similarity operation. The result of the cosine similarity operation is recorded as the similarity value.

[0039] The sum of similar values ​​between the scene monitoring area and each comparison case is accumulated under each data sensing frequency, and a similarity threshold is set.

[0040] Then, starting from the first data perception frequency of the disaster labeling, the relationship between the sum of similar values ​​of each comparative case and the similarity threshold is determined, and simulated cases are selected based on the determination results;

[0041] Based on the disaster coverage, independent variable data, and dependent variable data of the simulated case at various data perception frequencies, the monitoring area of ​​the first scene with an anomaly label will be used as the center to cover the dynamic scene monitoring model.

[0042] If there are still no comparable cases to retain, and in the process of selecting simulation cases, the comparable case with the largest sum of similar values ​​is selected for each data sensing frequency period. In the subsequent three data sensing frequencies, the disaster coverage area, independent variable data, and dependent variable data are covered on the dynamic scene monitoring model.

[0043] Furthermore, the crowd evacuation simulation process includes:

[0044] Based on the traffic data marked in each scene monitoring area in the dynamic scene monitoring model, personnel evacuation target points are set in the scene monitoring area. Then, based on the update results of the dynamic scene monitoring model under the continuous data perception frequency, the movement speed and movement direction of each personnel evacuation target point are obtained.

[0045] At the same time, based on the personnel and material data of each emergency shelter in the dynamic scenario monitoring model, the remaining capacity of each emergency shelter is marked.

[0046] Based on the principle of proximity and the remaining capacity of each emergency shelter as a constraint, evacuation personnel are allocated to each emergency shelter. Then, with the evacuation target point as the initial point of the simulation and the corresponding emergency shelter as the simulation endpoint, multiple evacuation routes are simulated in the dynamic scenario monitoring model.

[0047] Based on the simulation results of the movement speed, movement direction and disaster trajectory of each person's evacuation target point, the simulation of the evacuation route at each data sensing frequency, the expected location of each person's evacuation target point and whether it is far from the disaster-covered location, and the setting of distance thresholds;

[0048] If the distance to the disaster-covered location is less than or equal to the distance threshold, the corresponding evacuation route is determined to be unusable; otherwise, the evacuation route simulation continues.

[0049] When the personnel evacuation route simulation ends, the personnel evacuation route with the shortest simulation time and not judged as unusable is selected for use.

[0050] If all evacuation routes are deemed unusable, the nearest evacuation target point will be selected as the simulation endpoint to generate multiple evacuation routes. The evacuation route simulation will then be performed again. If all evacuation routes are still deemed unusable, the evacuation route with the shortest overall simulation time and the furthest distance from the disaster-covered location will be selected and put into use.

[0051] Furthermore, the process of dynamically adjusting evacuation routes includes:

[0052] During the process of putting the personnel evacuation route into use, each time it passes through a data sensing frequency, the simulation case is rematched according to the updated dynamic scenario monitoring model, and the disaster trajectory simulation result is reset according to the simulation case.

[0053] Then, a simulation is performed on the personnel evacuation routes that have already been deployed. If the simulation results are satisfactory, the current personnel evacuation routes are retained; otherwise, new personnel evacuation routes are set up and selected for use.

[0054] The intelligent management and control system for emergency shelters based on multi-source information fusion includes a multi-source data perception module, a scene logic analysis module, and an intelligent navigation module.

[0055] The multi-source data sensing module is used to acquire real-time scene data from various emergency shelters and monitoring scenarios, fuse real-time scene data from different data sensing devices, and establish a dynamic scene monitoring model based on the data fusion results.

[0056] The scenario logic analysis module is used to acquire historical case data of various disasters, establish a historical disaster evolution model based on the historical case data, compare the historical disaster evolution model with the dynamic scenario monitoring model, select comparison cases based on the comparison results, simulate the disaster trajectory in the dynamic scenario monitoring model based on the comparison cases, and overlay the simulated disaster trajectory results onto the dynamic scenario monitoring model.

[0057] The intelligent navigation module is used to simulate crowd evacuation on a dynamic scene monitoring model, with emergency shelters as the simulated endpoint. It generates evacuation routes based on the simulation results and dynamically adjusts them according to real-time scene data during the execution of the evacuation routes.

[0058] The technical effects and advantages provided by the present invention in the above technical solution are as follows:

[0059] 1. This invention, by setting up data sensing devices in various emergency shelters and monitoring scenarios and fusing real-time scene data from different data sensing devices, can comprehensively and accurately grasp the real-time status of emergency shelters and their surrounding environment. At the same time, the various types of scene data complement each other, providing a richer and more reliable data foundation for subsequent disaster analysis and the generation of personnel evacuation routes.

[0060] 2. This invention simulates crowd evacuation using emergency shelters as the simulated endpoint, generates dynamic evacuation routes based on the simulation results, and dynamically adjusts these routes according to real-time scenario data during the evacuation process. This allows for timely adjustments to evacuation routes based on actual conditions, avoiding dangerous areas and congested sections, improving evacuation efficiency, and maximizing the protection of people's lives. Attached Figure Description

[0061] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0062] Figure 1 This is a flowchart of the intelligent management and control method for emergency shelters based on multi-source information fusion as described in this invention.

[0063] Figure 2 This is a system block diagram of the intelligent management and control system for emergency shelters based on multi-source information fusion as described in this invention. Detailed Implementation

[0064] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0065] Please see Figure 1 As shown, the intelligent management and control method for emergency shelters based on multi-source information fusion includes the following steps:

[0066] Step S1: Set up data sensing devices in various emergency shelters and monitoring scenarios, and obtain real-time scene data of various emergency shelters and monitoring scenarios through the data sensing devices. Perform data fusion on the real-time scene data from different data sensing devices, and establish a dynamic scene monitoring model based on the data fusion results.

[0067] Step S2: Obtain historical case data of various disasters, establish a historical disaster evolution model based on the historical case data, compare the historical disaster evolution model with the dynamic scene monitoring model, select comparison cases based on the comparison results, simulate the disaster trajectory in the dynamic scene monitoring model based on the comparison cases, and overlay the simulated disaster trajectory results onto the dynamic scene monitoring model.

[0068] Step S3: Based on the simulated disaster trajectory results on the dynamic scene monitoring model, conduct a population evacuation simulation with the emergency shelter as the simulation endpoint. Generate evacuation routes based on the population evacuation simulation results, and dynamically adjust them based on real-time scene data during the execution of the evacuation routes.

[0069] Step S1 is achieved through the following process:

[0070] Step S101: Set up the data sensing device. The specific process includes:

[0071] Data sensing devices are set up for emergency shelters and monitoring scenarios respectively. The data sensing devices for emergency shelters cover three data sensing directions: external environment, personnel status, and material reserves. The data sensing devices for monitoring scenarios cover three data sensing directions: external disasters, traffic conditions, and infrastructure.

[0072] The data sensing devices in the emergency shelter include weather sensing data (including meteorological monitoring stations, wind speed sensors, and barometric pressure sensors), personnel status monitoring sensors (composed of an infrared dual-beam personnel counter and an AI video camera), and material reserve sensors (composed of RFID readers and RFID tags attached to various emergency supplies).

[0073] The data sensing devices in the monitoring scenario include disaster monitoring sensors (composed of meteorological monitoring stations, wind speed sensors, and air pressure sensors, etc.) and traffic and infrastructure sensors (composed of AI video cameras and traffic flow sensors).

[0074] Each data sensing device is assigned the number a1, a2, a3, ..., a n b1, b2, b3, ..., b m , where a n and b mLet n and m represent the nth and mth data sensing devices in the emergency shelter and the monitoring scenario, respectively, where n and m are natural numbers greater than 0;

[0075] The same data sensing frequency is set for data sensing devices in emergency shelters and monitoring scenarios, with the duration of the data sensing frequency generally ranging from 10ms to 50ms.

[0076] The data sensing devices in emergency shelters or monitoring scenarios are set with the same data sensing range and distributed at the same spatial distance. The combined result of the overlapping part of the data sensing range of any data sensing device (except the boundary position) with the data sensing range of all adjacent data sensing devices covers the data sensing range of that data sensing device, so that the combined range of the data sensing range of all data sensing devices covers the entire emergency shelter and monitoring scenario.

[0077] Step S102 involves data fusion of real-time scene data, specifically including:

[0078] After all data sensing devices synchronize their time, they acquire real-time scene data within their data sensing range. Whenever a data sensing frequency is reached, the data sensing device converts the collected real-time scene data into JSON format, with fields including device number, acquisition time, data type, value, unit, and location coordinates.

[0079] The real-time scene data for emergency shelters includes weather data (rainfall, wind speed, displacement, water level, etc.), personnel data (real-time number of people, density, type, and location), material data (inventory quantity and equipment status), and video data.

[0080] The real-time scene data monitored includes weather data (rainfall, wind speed, displacement, water level, etc.), traffic data (vehicle flow, pedestrian flow, road conditions), and video data;

[0081] As can be seen from step S101, the data sensing ranges of adjacent spatial location data sensing devices have overlapping parts, and a three-dimensional coordinate system is constructed. The acquisition time, data type and location coordinates contained in the fields of each real-time scene data are mapped into the three-dimensional coordinate system, so that each real-time scene data is distributed in the three-dimensional coordinate system according to the spatiotemporal location order and data type.

[0082] A deviation threshold is set for each type of real-time scene data. If the data are in an overlapping state and the difference between the same real-time scene data from different data sensing devices is less than or equal to the deviation threshold, the average value of the two sets of real-time scene data is accumulated and used as the real-time scene data for the corresponding scene location.

[0083] If the difference is greater than the deviation threshold, then the default value of the data sensing device within the corresponding collection time is increased by 1;

[0084] When the default value is greater than or equal to 3, the real-time scene data of the corresponding data type of the data sensing device within the corresponding collection time is deleted, and the real-time scene data collected by the remaining data sensing devices is used as the real-time scene data of the corresponding scene location; otherwise, it is not deleted.

[0085] It should be noted that if the data sensing device has a failure value of 3 or more for three consecutive data sensing frequencies, the sensor associated with the corresponding data type is deemed to be damaged, and staff will be arranged to carry out repairs.

[0086] When the failure value of the data sensing device associated with any scene location is greater than or equal to 3, the real-time scene data of the data type corresponding to that scene location is temporarily ignored.

[0087] Step S103: Establish a dynamic scene monitoring model. The specific process includes:

[0088] Based on the video data in the real-time scene data collected at various data sensing frequencies, a visual 3D model is established. Then, the weather data of emergency shelters and monitoring scenes are overlaid on the visual 3D model in the form of textures. Each texture represents a numerical range in the image range it occupies.

[0089] Personnel and material data are directly labeled on the corresponding parts of the visualized 3D model of the emergency shelter, while traffic data is overlaid on the visualized 3D model in the form of point cloud data, thus obtaining a dynamic scene monitoring model;

[0090] It should be noted that the dynamic scene monitoring model is updated once every data sensing frequency duration.

[0091] Step S2 is achieved through the following process:

[0092] Step S201: Establish a historical disaster evolution model. The specific process includes:

[0093] Obtain k historical case data for several types of disasters (such as earthquakes, floods, typhoons, landslides, mudslides, etc.) through the Internet, where k is a positive integer greater than 1000;

[0094] The historical case data includes disaster type name, evolution process data (such as weather data during the flood spread process, including rainfall, wind speed, displacement, water level, etc.), impact result data (impact range, infrastructure damage data, etc.), and emergency response data (number of evacuees, amount of materials consumed, and use of shelters).

[0095] It should be noted that the weather data in the evolution process data are the values ​​confirmed at the time of the corresponding occurrence;

[0096] Feature extraction dimensions are set for various disaster types. For example, the feature extraction dimensions for floods are rainfall, water level, flow velocity, duration, and inundation range, while the feature extraction dimensions for typhoons are central pressure, maximum wind speed, movement speed, impact range, and duration.

[0097] Based on the feature extraction dimensions, a set of historical feature data is extracted from each historical case data. Then, a historical disaster evolution model corresponding to the historical case data is generated using each historical data in the set of historical feature data as independent variable data and the data affecting the outcome as dependent variable data.

[0098] The dependent variable data in the historical disaster evolution model exists in the form of an image model, while the independent variable data and the weather and environmental data in the dynamic scene monitoring model are all overlaid on the dependent variable data in the form of textures.

[0099] The lowest values ​​of each independent variable in the historical disaster evolution model corresponding to various disaster types are obtained as the induced feature values.

[0100] Step S202: Select a comparative case. The specific process includes:

[0101] Based on the spatial location and number of data sensing devices, each dynamic scene monitoring model is divided into m+n scene monitoring areas;

[0102] The triggering characteristic values ​​of various disaster types are compared with various real-time weather data in each scene monitoring area. If there is a real-time weather data that is greater than or equal to the triggering characteristic value, an abnormal weather data label is set for the scene monitoring area according to the type of the real-time weather data; otherwise, no abnormal weather data label is set.

[0103] When the abnormal weather data labels in the scene monitoring area meet all feature extraction dimensions of a disaster type, disaster labels are set for the corresponding scene monitoring area according to the disaster type name.

[0104] Because environmental disasters expand or move in real-world scenarios in the form of spread or cell division, when a scene monitoring area meets the requirements for disaster occurrence, it usually generates a chain reaction in an avalanche-like manner, producing the same environmental impact on the surrounding environment.

[0105] Then, using the scene monitoring area with disaster labels as the monitoring location, the disaster label spread process of the scene monitoring area under more than 50 subsequent data sensing frequencies is obtained, and the historical disaster evolution model is retrieved and compared with the disaster label spread process based on the disaster labels;

[0106] The initial state of the dependent variable data in each historical disaster evolution model is compared with that of the dynamic scene monitoring model. At least 100 historical disaster evolution models that are most similar to the visualized 3D model in the dynamic scene monitoring model are selected as comparison cases, and then the comparison cases are simultaneously divided into several comparison scene areas.

[0107] Step S203: Simulate the disaster trajectory. The specific process includes:

[0108] The real-time values ​​of various feature extraction dimensions corresponding to the scene monitoring area with disaster annotation and the scene disaster area in its adjacent location are obtained at the beginning and end of each data perception frequency. Then, the disaster spread vector is established with the disaster annotation spread direction as the vector direction and the real-time values ​​of various feature extraction dimensions corresponding to the adjacent scene monitoring areas with disaster annotation as scalars.

[0109] Starting from the first abnormal comparison scene area in the comparison case, obtain the comparison spread vector of the comparison scene area with the abnormality under each data perception frequency duration. Select the comparison spread vector and the disaster spread vector with the most similar vector direction to perform cosine similarity operation. The result of the cosine similarity operation is recorded as the similarity value.

[0110] The sum of similar values ​​between the scene monitoring area and each comparison case is accumulated under each data sensing frequency. A similarity threshold, a first limit number of times, and a second limit number of times are set, where the first limit number of times is less than the second limit number of times, and the values ​​are all greater than 100.

[0111] Then, starting from the first data perception frequency of the disaster label, we determine the relationship between the sum of similar values ​​of each comparative case and the similarity threshold;

[0112] If the sum of similar values ​​is less than the similarity threshold, the corresponding comparison case is removed.

[0113] If the sum of similar values ​​is greater than or equal to the similarity threshold, the corresponding comparison case is retained;

[0114] When the data perception frequency is measured for the second limited number of times, the comparison case with the largest cumulative sum of similar values ​​is selected as the simulation case.

[0115] If there are no retained comparison cases when the second limited number of data perception frequencies are reached, then the retained comparison case with the largest cumulative sum of similar values ​​when the first limited number of data perception frequencies are reached is selected and recorded as the simulated case.

[0116] Based on the disaster coverage, independent variable data, and dependent variable data of the simulated case at various data perception frequencies, the monitoring area of ​​the first scene with an anomaly label will be used as the center to cover the dynamic scene monitoring model.

[0117] If there are still no comparable cases to retain, and in the process of selecting simulation cases, the comparable case with the largest sum of similar values ​​is selected for each data sensing frequency period. In the subsequent three data sensing frequencies, the disaster coverage area, independent variable data, and dependent variable data are covered on the dynamic scene monitoring model.

[0118] Step S3 is achieved through the following process:

[0119] Step S301, crowd evacuation simulation, the specific process includes:

[0120] Based on the traffic data (vehicle flow, pedestrian flow, road conditions) marked in each scene monitoring area in the dynamic scene monitoring model, personnel evacuation target points are set in the scene monitoring area. Then, based on the update results of the dynamic scene monitoring model under the continuous data perception frequency, the movement speed and movement direction of each personnel evacuation target point are obtained.

[0121] At the same time, based on the personnel data (real-time number of people, density, type and location) and material data (inventory quantity and equipment status) of each emergency shelter in the dynamic scene monitoring model, the remaining capacity of each emergency shelter is marked.

[0122] Based on the principle of proximity and the remaining capacity of each emergency shelter as a constraint, evacuation personnel are allocated to each emergency shelter. Then, with the evacuation target point as the initial point of the simulation and the corresponding emergency shelter as the simulation endpoint, multiple evacuation routes are simulated in the dynamic scenario monitoring model.

[0123] Based on the simulation results of the movement speed, movement direction and disaster trajectory of each person's evacuation target point, the simulation of the evacuation route at each data sensing frequency, the expected location of each person's evacuation target point and whether it is far from the disaster-covered location, and the setting of distance thresholds;

[0124] It should be noted that, based on the dependent variable data in the disaster trajectory simulation results, the dynamic scene monitoring model is used to determine whether each road is passable under each data perception frequency.

[0125] If the distance to the disaster-covered location is less than or equal to the distance threshold, the corresponding evacuation route is determined to be unusable; otherwise, the evacuation route simulation continues.

[0126] When the personnel evacuation route simulation ends, the personnel evacuation route with the shortest simulation time and not judged as unusable is selected for use.

[0127] If all evacuation routes are deemed unusable, the nearest evacuation target point will be selected as the simulation endpoint to generate multiple evacuation routes. The evacuation route simulation will then be performed again. If all evacuation routes are still deemed unusable, the evacuation route with the shortest overall simulation time and the furthest distance from the disaster-covered location will be selected and put into use.

[0128] Step S302: Dynamically adjust the evacuation route for personnel. The specific process includes:

[0129] During the process of putting the personnel evacuation route into use, each time it passes through a data sensing frequency, the simulation case is rematched according to the updated dynamic scenario monitoring model, and the disaster trajectory simulation result is reset according to the simulation case.

[0130] Then, a simulation is performed on the personnel evacuation routes that have already been deployed. If the simulation results are satisfactory, the current personnel evacuation routes are retained; otherwise, new personnel evacuation routes are set up and selected for use.

[0131] Please see Figure 2 As shown, the intelligent management and control system for emergency shelters based on multi-source information fusion includes a multi-source data perception module, a scene logic analysis module, and an intelligent navigation module.

[0132] The multi-source data sensing module is used to acquire real-time scene data from various emergency shelters and monitoring scenarios, fuse real-time scene data from different data sensing devices, and establish a dynamic scene monitoring model based on the data fusion results.

[0133] The scenario logic analysis module is used to acquire historical case data of various disasters, establish a historical disaster evolution model based on the historical case data, compare the historical disaster evolution model with the dynamic scenario monitoring model, select comparison cases based on the comparison results, simulate the disaster trajectory in the dynamic scenario monitoring model based on the comparison cases, and overlay the simulated disaster trajectory results onto the dynamic scenario monitoring model.

[0134] The intelligent navigation module is used to simulate crowd evacuation on a dynamic scene monitoring model, with emergency shelters as the simulated endpoint. It generates evacuation routes based on the simulation results and dynamically adjusts them according to real-time scene data during the execution of the evacuation routes.

[0135] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. An emergency shelter intelligent management and control method based on multi-source information fusion, characterized in that, The method comprises the following steps: Step S1, setting up data sensing devices in each emergency shelter and monitoring scene, and obtaining real-time scene data of each emergency shelter and monitoring scene through the data sensing devices, performing data fusion on the real-time scene data from different data sensing devices, and establishing a dynamic scene monitoring model according to the data fusion result; Step S2, obtaining historical case data of various disasters, establishing a historical disaster evolution model according to the historical case data, comparing the historical disaster evolution model with the dynamic scene monitoring model, selecting a comparison case according to the comparison result, simulating the disaster trend in the dynamic scene monitoring model according to the comparison case, and covering the simulation disaster trend result on the dynamic scene monitoring model; The process of simulating the disaster trend comprises: Obtaining real-time values of various feature extraction dimensions of the scene disaster area with disaster annotation and its adjacent position, taking the disaster annotation spreading direction as the vector direction, and taking the real-time values of various feature extraction dimensions of the adjacent scene monitoring area with disaster annotation as the scalar, to establish a disaster spreading vector; Starting from the comparison scene area where the first anomaly appears in the comparison case, obtaining a comparison spreading vector with the anomaly, selecting the comparison spreading vector and the disaster spreading vector with the closest vector direction to perform cosine similarity operation, and recording the result as a similarity value; Setting a data sensing frequency, accumulating the similarity value sum of the scene monitoring area and each comparison case under each data sensing frequency, and setting a similarity threshold; Starting from the first data sensing frequency where the disaster annotation appears, judging the size relationship between the similarity value sum of the comparison case and the similarity threshold, selecting a simulation case according to the judgment result, covering the disaster coverage range, independent variable data and dependent variable data of the simulation case at each data sensing frequency period on the dynamic scene monitoring model, and taking the scene monitoring area where the first anomaly annotation appears as the center; Step S3, according to the simulation disaster trend result on the dynamic scene monitoring model, taking the emergency shelter as the simulation end point to simulate the crowd evacuation, generating the personnel evacuation route according to the crowd evacuation simulation result, and dynamically adjusting the personnel evacuation route according to the real-time scene data during the execution of the personnel evacuation route; The process of simulating the crowd evacuation comprises: Setting a personnel evacuation target point, obtaining the moving speed and moving direction of the personnel evacuation target point, and marking the remaining accommodation capacity of the emergency shelter according to the personnel data and material data of the emergency shelter; According to the near principle and the remaining accommodation capacity of the emergency shelter as the constraint condition, distributing the evacuation personnel to the emergency shelter, taking the personnel evacuation target point as the simulation initial point, and taking the distributed emergency shelter as the simulation end point to simulate multiple personnel evacuation routes; According to the moving speed, moving direction and disaster trend simulation result of each personnel evacuation target point, simulating the expected position of each personnel evacuation target point and whether it is away from the disaster covered position, and setting a distance threshold; According to the size relationship between the disaster covered position distance and the distance threshold, selecting the personnel evacuation route with the shortest simulation time and not being judged as unable to use for use.

2. The multi-source information fusion-based intelligent management and control method of emergency shelter according to claim 1, characterized in that, The process of setting up the data sensing device in the emergency shelter and the monitoring scene includes: The data sensing device in the emergency shelter and the monitoring scene is set with the same data sensing frequency, the same data sensing range, and the same spatial distance distribution, so that the data sensing range of any data sensing device and the data sensing range of all adjacent data sensing devices overlap to cover the data sensing range of the data sensing device, and the combined range of the data sensing range of all data sensing devices covers the entire emergency shelter and the monitoring scene.

3. The multi-source information fusion-based intelligent management and control method of emergency shelter according to claim 2, characterized in that, The process of data fusion of real-time scene data includes: After time synchronization between all data sensing devices, real-time scene data within the data sensing range is obtained, and when a data sensing frequency ends, the data sensing device converts the collected real-time scene data into JSON format, including device number, collection time, data type, value, unit, and position coordinates; The real-time scene data of the emergency shelter includes weather data, personnel data, material data, and video data, and the real-time scene data of the monitoring scene includes weather data, traffic data, and video data; For each type of real-time scene data, a deviation threshold is set, and if the same real-time scene data from different data sensing devices is in an overlapping state and the difference is less than or equal to the deviation threshold, the two sets of real-time scene data are averaged to obtain the real-time scene data of the corresponding scene position; If the difference is greater than the deviation threshold, the data sensing device at the corresponding collection time is set with a loss of credit value of 1, and when the loss of credit value is greater than or equal to 3, the real-time scene data of the corresponding data type of the data sensing device at the corresponding collection time is deleted, and the real-time scene data collected by the remaining data sensing devices is used as the real-time scene data of the corresponding scene position; When the loss of credit value of any data sensing device associated with the scene position is greater than or equal to 3, the real-time scene data of the corresponding data type of the scene position is temporarily ignored.

4. The multi-source information fusion-based intelligent management and control method of emergency shelter according to claim 3, characterized in that, The process of establishing a dynamic scene monitoring model includes: According to the video data collected in each data sensing frequency, a visual three-dimensional model is established, and then the weather data of the emergency shelter and the monitoring scene is covered on the visual three-dimensional model in the form of texture, the personnel data and the material data are directly marked on the corresponding part of the emergency shelter in the visual three-dimensional model, and the traffic data is covered on the visual three-dimensional model in the form of point cloud data, and then the dynamic scene monitoring model is obtained.

5. The multi-source information fusion-based intelligent management and control method of emergency shelter according to claim 4, characterized in that, The process of establishing a historical disaster evolution model includes: Setting feature extraction dimensions for various disaster types, extracting historical feature data sets from each historical case data according to the feature extraction dimensions, and then using each historical data in the historical feature data set as the independent variable data and the influence result data as the dependent variable data to generate the historical disaster evolution model of the corresponding historical case data; The dependent variable data in the historical disaster evolution model exists in the form of an image model, and the independent variable data and the weather data and environmental data in the dynamic scene monitoring model are all covered on the dependent variable data in the form of textures; The minimum value of each item of independent variable data in the historical disaster evolution model corresponding to each disaster type is obtained as an inducing characteristic value.

6. The multi-source information fusion-based intelligent management and control method of emergency shelter according to claim 5, characterized in that, The selection process of the comparative case includes: Each dynamic scene monitoring model is divided into a scene monitoring area, and the inducing characteristic value of each disaster type is compared with each real-time weather data of each scene monitoring area. If there is one real-time weather data greater than or equal to the inducing characteristic value, the scene monitoring area is set with an abnormal weather data label, otherwise no abnormal weather data label is set; When the abnormal weather data label carried by the scene monitoring area meets all the feature extraction dimensions of one disaster type, the scene monitoring area is set with a disaster label according to the disaster type name; The scene monitoring area with the disaster label is taken as a monitoring location, the disaster label spread process of the scene monitoring area under the subsequent data sensing frequency is obtained, and the disaster label spread process is compared with the historical disaster evolution model according to the disaster label; The initial state of the dependent variable data in each historical disaster evolution model is compared with the dynamic scene monitoring model, and the historical disaster evolution model is selected as a comparative case according to the comparison result, and then the comparative case is divided into several comparative scene areas.

7. The multi-source information fusion-based intelligent management and control method of emergency shelter according to claim 6, characterized in that, The process of dynamically adjusting the personnel evacuation route includes: During the use of the personnel evacuation route, every time a data sensing frequency is passed, a simulation case is matched again according to the updated dynamic scene monitoring model, and the disaster trend simulation result is set again according to the simulation case; Then the personnel evacuation route that has been put into use is simulated once. If the simulation result is passed, the current personnel evacuation route is retained, otherwise the personnel evacuation route is set and selected again for use.

8. The intelligent management and control system of emergency shelter based on multi-source information fusion, used to realize the intelligent management and control method of emergency shelter based on multi-source information fusion according to any one of claims 1-7, characterized in that, The system includes a multi-source data sensing module, a scene logic analysis module, and an intelligent navigation module; The multi-source data sensing module is used to obtain real-time scene data of each emergency shelter and monitoring scene, to perform data fusion on real-time scene data from different data sensing devices, and to establish a dynamic scene monitoring model according to the data fusion result; The scene logic analysis module is used to obtain historical case data of various disasters, to establish a historical disaster evolution model according to the historical case data, to compare the historical disaster evolution model with the dynamic scene monitoring model, to select a comparative case according to the comparison result, to simulate a disaster trend in the dynamic scene monitoring model according to the comparative case, and to cover the simulation disaster trend result on the dynamic scene monitoring model; The intelligent navigation module is used to simulate crowd evacuation on the dynamic scene monitoring model with the emergency shelter as a simulation end point, to generate a personnel evacuation route according to the crowd evacuation simulation result, and to dynamically adjust the personnel evacuation route during the execution of the personnel evacuation route according to real-time scene data.

Images