Real-time disaster situation multi-dimensional verification alarm method and system based on large model

Abstract

The invention provides a real-time disaster situation multi-dimensional verification alarm method and system based on a large model, and relates to the technical field of disaster situation response, and the method comprises the steps: carrying out the backtracking and calling of multi-source disaster situation associated data after a disaster situation early warning signal is received, and obtaining a heterogeneous disaster situation data set; starting a dynamic task chain scheduling mechanism, executing multi-priority thread trend analysis, and outputting a multi-dimensional index analysis result; executing multi-dimensional index cross verification, and outputting a disaster verification conclusion; disaster decision making is carried out, and a disaster decision instruction is generated; carrying out township-level disaster situation grid positioning, and outputting a disaster boundary coordinate set; after the thermal rescue target spot is framed, geological disaster avoidance analysis is carried out, and a safety buffer area is positioned; and executing an alarm verification response. The technical problems that in the prior art, a disaster monitoring and early warning system mainly depends on a step-by-step reporting mechanism, so that disaster information transmission is lagged, accuracy is lost, and timeliness and accuracy of post-disaster rescue work are affected are solved.

Description

Technical Field

[0001] This invention relates to the field of disaster response technology, and in particular to a method and system for real-time multi-dimensional verification and alarm of disasters based on a large model. Background Technology

[0002] The current disaster monitoring and early warning system mainly relies on a hierarchical reporting mechanism, the core flaw of which lies in the serious lag and lack of accuracy in the transmission of disaster information.

[0003] On the one hand, although the red alerts for rainstorms and mudslides issued by the meteorological department can be located down to the city and county level, they cannot verify in real time whether a disaster has actually occurred. They need to wait for on-site verification by grassroots personnel before reporting, which causes the disaster confirmation to be delayed by several hours or even several days.

[0004] On the other hand, existing technologies are highly fragmented: systems such as tower power outage monitoring, communication base station status, and video surveillance operate independently, lacking a dynamic fusion mechanism for multi-source data. For example, while a tower offline rate exceeding 60% indicates a risk of communication paralysis, it is not linked to weather warnings or population distribution data for analysis; video surveillance requires manual inspection of water accumulation or river anomalies and cannot automatically extract effective features.

[0005] More importantly, the accurate location of disaster situations at the township level is completely ineffective. Traditional methods cannot quickly pinpoint areas of population gathering and rescue coordinates, leading to the blind deployment of rescue forces.

[0006] These deficiencies essentially stem from the lack of a passive response model and intelligent verification methods, and there is an urgent need to build a multi-dimensional real-time disaster verification system that requires no manual reporting and has a closed-loop process within minutes.

[0007] It should be noted that the information disclosed in this background section is intended only to enhance the understanding of the overall background of the present invention, and should not be construed as an admission or in any way implying that the information constitutes prior art known to those skilled in the art. Summary of the Invention

[0008] In response to the above-mentioned deficiencies or improvement needs of existing technologies, this invention provides a real-time disaster multi-dimensional verification and alarm method and system based on a large model. This solves the technical problem that the existing disaster monitoring and early warning system mainly relies on a hierarchical reporting mechanism, which leads to the lag and lack of accuracy in the transmission of disaster information, affecting the timeliness and accuracy of post-disaster relief work.

[0009] The specific technical solution is as follows:

[0010] According to a first aspect of the present invention, a real-time disaster multi-dimensional verification and alarm method based on a large model is provided, the method comprising:

[0011] After receiving disaster warning signals from the monitored area, the system uses the trigger time of the warning signal as the time base to perform backtracking retrieval of multi-source disaster-related data within a preset analysis time window, obtaining a heterogeneous disaster dataset. Based on the warning type of the disaster warning signal, a dynamic task chain scheduling mechanism is activated to perform multi-priority thread trend analysis on the heterogeneous disaster dataset, outputting multi-dimensional indicator analysis results. After spatiotemporally aligning the multi-dimensional indicator analysis results, multi-dimensional indicator cross-verification is performed, outputting a disaster verification conclusion. The disaster verification conclusion is then sent to the composite rule judgment engine for further processing. The system makes disaster situation decisions and generates disaster situation decision instructions. Based on the disaster situation verification conclusions and multi-dimensional index analysis results, it performs township-level disaster situation grid positioning and outputs the disaster-affected boundary coordinate set. Based on the disaster-affected boundary coordinate set, it defines the thermal rescue target points in the pre-disaster moving signal population thermal distribution map and performs geological disaster avoidance analysis on the thermal rescue target points to locate the safety buffer zone. After packaging the disaster-affected boundary coordinate set, thermal rescue target points, and safety buffer zones into a structured rescue instruction package, it executes the alarm verification response of the structured rescue instruction package according to the disaster situation level of the disaster situation decision instructions.

[0012] In one implementation, the heterogeneous disaster data set includes tower power supply status data, communication base station load data, and full-domain monitoring video streams.

[0013] In one implementation, based on the warning type of the disaster warning signal, a dynamic task chain scheduling mechanism is activated to perform multi-priority thread trend analysis on the heterogeneous disaster dataset, and output multi-dimensional indicator analysis results, including:

[0014] The disaster type identifier code of the disaster warning signal is parsed to obtain the warning type; task chain matching is performed in the scheduling rule base according to the warning type to obtain the thread priority sequence; the activation order of the video parsing component, tower status analysis component and communication load analysis component is dynamically scheduled according to the thread priority sequence, and cross-dimensional trend collaborative analysis is performed on the tower power supply status data, communication base station load data and full-domain monitoring video stream, and the multi-dimensional indicator analysis results are output.

[0015] In one implementation, if the warning type is a rainstorm scenario warning:

[0016] S1: First, activate the video parsing component, input the full-domain monitoring video stream identification output of the surface water depth change rate and river runoff interruption status as video parsing dimension indicators; S2: If the surface water depth change rate is greater than a preset change threshold, then activate the tower status analysis component, input the tower power supply status data, calculate the tower power outage density heatmap and the unit time offline rate increase as tower power supply dimension indicators; S3: If the tower power outage rate of the tower power outage density heatmap is greater than a preset power outage ratio, and the unit time offline rate increase is greater than a preset increase threshold, then activate the communication load analysis component, input the communication base station load data, calculate the channel pressure cumulative increase and the overloaded base station ratio as communication load dimension indicators, wherein the tower power supply dimension indicators, communication load dimension indicators, and video parsing dimension indicators constitute the multi-dimensional indicator analysis results.

[0017] In one implementation, the video parsing component is activated first, and the surface water depth change rate and river runoff interruption status are identified and output from the full-area monitoring video stream as video parsing dimension indicators, including:

[0018] Based on a historical disaster case database, distributed rainstorm disaster surface areas are defined. Spatial filtering of monitoring points in the overall monitoring video stream is performed to obtain distributed waterlogged area video streams. Pixel-level waterlogging depth identification and tracking are performed on the distributed waterlogged area video streams based on calibration references, and the distributed waterlogging change rate is calculated and output. The distributed waterlogging change rate is weighted and fused according to the geological permeability coefficient of the distributed rainstorm disaster surface areas to output the surface waterlogging depth change rate. The river monitoring area boundaries are retrieved from the GIS system, and monitoring points in the overall monitoring video streams are retrieved to obtain multiple fixed-point monitoring video stream sequences for multiple river branches. Water flow vector analysis based on optical flow is performed on the multiple fixed-point monitoring video stream sequences to output multiple river interruption monitoring coordinates as the river runoff interruption status.

[0019] In one implementation, after spatiotemporally aligning the results of the multidimensional index analysis, multidimensional index cross-verification is performed, and a disaster verification conclusion is output, including:

[0020] If the multidimensional indicator analysis results do not cover the tower power supply dimension indicators, communication load dimension indicators, and video analysis dimension indicators, the verification process is terminated, and the multidimensional indicator analysis results are sent to the manual decision-making end. If the multidimensional indicator analysis results cover the tower power supply dimension indicators, communication load dimension indicators, and video analysis dimension indicators, then after mapping the tower power supply dimension indicators, communication load dimension indicators, and video analysis dimension indicators based on GIS grid coding spatiotemporal alignment, the following steps are executed: s1: Extract the high-density clustering area of ​​the power outage towers from the tower power outage density heat map; s2: Perform binary partitioning of the distributed water accumulation change rate using the preset change threshold to obtain the water accumulation verification area polygon; s3: Calculate the spatial overlap area ratio of the high-density clustering area of ​​the power outage towers and the water accumulation verification area polygon; s4: Based on the cumulative increase in channel pressure and the increase in offline rate per unit time, extract the pressure increase node and the offline increase node, and calculate the increase time difference; s5: If the increase time difference meets the preset time difference threshold, and the overlap area ratio is greater than the preset area ratio threshold, then the disaster verification conclusion is a level one verification conclusion.

[0021] In one implementation, based on the disaster verification conclusion and the multi-dimensional index analysis results, township-level disaster grid positioning is performed, and a disaster boundary coordinate set is output, including:

[0022] If the disaster verification conclusion is a Level 1 verification conclusion, then construct the disaster-affected grid array of the monitoring area; after projecting the high-density clustering area of ​​the power outage tower, the polygon of the water accumulation verification area, and the coordinates of multiple river interruption monitoring onto the disaster-affected grid array, merge the disaster-affected areas based on the watershed algorithm, and output the disaster-affected boundary coordinate set.

[0023] In one implementation, after defining the thermal rescue target point based on the disaster boundary coordinate set in the pre-disaster moving signal population thermal distribution map, a geological hazard avoidance analysis is performed on the thermal rescue target point to locate a safety buffer zone, including:

[0024] The population thermal distribution map of the mobile signal is segmented using the disaster-affected boundary coordinate set to obtain a disaster-affected grid thermal distribution sub-map. After segmenting the disaster-affected grid thermal distribution sub-map based on a preset population thermal value, target point correction is performed by combining non-residential attribute thermal points from the POI database to define the thermal rescue target point. Road surface water depth identification and road collapse feature extraction are performed based on the full-domain monitoring video stream to locate the coordinates of distributed road breakpoints. The spatial restricted area mask of the thermal rescue target point is superimposed based on the coordinates of the distributed road breakpoints and the polygon of the water accumulation verification area to output the safety buffer.

[0025] In one implementation, it further includes:

[0026] If the rate of change of surface water depth is less than a preset threshold, the activation process of the tower status analysis component and the communication load analysis component is terminated, and the tower power supply dimension index is used as the result of the multidimensional index analysis; if the tower power outage rate is less than a preset power outage ratio, and / or the increase in the offline rate per unit time is less than a preset increase threshold, the activation process of the communication load analysis component is terminated, and the tower power supply dimension index and the communication load dimension index are used as the result of the multidimensional index analysis.

[0027] According to a second aspect of the present invention, a real-time disaster multi-dimensional verification and alarm system based on a large model is provided, the system comprising:

[0028] The data backtracking and retrieval module is used to retrieve multi-source disaster-related data backtracking within a preset analysis time window after receiving disaster warning signals from the monitored area, using the trigger time of the disaster warning signal as the time base, to obtain a heterogeneous disaster dataset. The thread trend analysis module is used to initiate a dynamic task chain scheduling mechanism based on the warning type of the disaster warning signal, perform multi-priority thread trend analysis on the heterogeneous disaster dataset, and output multi-dimensional indicator analysis results. The indicator cross-verification module is used to perform multi-dimensional indicator cross-verification after spatiotemporally aligning the multi-dimensional indicator analysis results, and output disaster verification conclusions. The disaster decision-making module is used to send the disaster verification conclusions to the composite rule-based decision-making system. The system comprises a disaster assessment engine for disaster decision-making and the generation of disaster decision-making instructions; a disaster grid positioning module for locating township-level disaster grids based on the disaster verification conclusions and multi-dimensional index analysis results, and outputting a disaster boundary coordinate set; a disaster avoidance analysis module for locating thermal rescue target points based on the disaster boundary coordinate set and the pre-disaster moving signal population thermal distribution map, and performing geological disaster avoidance analysis on the thermal rescue target points to locate safety buffer zones; and an alarm verification response module for packaging the disaster boundary coordinate set, thermal rescue target points, and safety buffer zones into a structured rescue instruction package, and executing the alarm verification response of the structured rescue instruction package according to the disaster level of the disaster decision-making instructions.

[0029] Beneficial effects of the embodiments of the present invention:

[0030] By receiving disaster early warning signals and using them as a time reference, multi-source disaster-related data backtracking retrieval is initiated within a preset analysis time window. This enables real-time retrieval of heterogeneous datasets across multiple information sources, avoiding the time lag problem in data reporting in traditional methods. It achieves a minute-level closed-loop response, improving the real-time performance and accuracy of disaster monitoring. Through a dynamic task chain scheduling mechanism, multiple analysis threads can be automatically activated based on the disaster early warning type, and multi-priority trend analysis can be performed, outputting multi-dimensional indicator analysis results. Intelligent task scheduling and automated analysis reduce manual intervention, achieving multi-dimensional cross-validation to ensure the accuracy of disaster information and enabling real-time detection of disaster evolution, avoiding the lag and inefficiency of manual verification. Based on disaster verification conclusions and multi-dimensional indicator analysis results, township-level disaster grid positioning is performed, and a disaster boundary coordinate set is output. This process achieves… Precise location of disaster areas can clearly identify severely affected regions, ensuring the accuracy of disaster information, improving the precision of disaster location and response speed, and ensuring that resources can be quickly and accurately allocated to the affected areas. By combining the disaster boundary coordinate set with the pre-disaster mobile signal population heat map, thermal rescue target points are defined, and geological disaster avoidance analysis is conducted to ultimately locate safety buffer zones, preventing rescue personnel from entering severely affected or impassable areas, thus improving rescue efficiency and safety. The disaster boundary coordinate set, thermal rescue target points, and safety buffer zones are packaged into a structured rescue command package, and alarm verification responses are executed according to the disaster level of the disaster decision command, ensuring the efficiency and timeliness of rescue operations. Rescue commands can not only be accurate to specific coordinates and areas, but can also be quickly adjusted according to the needs of different disaster levels, avoiding mismatches and inefficiencies in response.

[0031] Of course, implementing any product or method of the present invention does not necessarily require achieving all of the advantages described above at the same time. Attached Figure Description

[0032] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0033] Figure 1 The flowchart of the real-time disaster multi-dimensional verification and alarm method based on a large model provided by the present invention is shown.

[0034] Figure 2 The diagram shows the structure of the real-time disaster multi-dimensional verification and alarm system based on a large model provided by the present invention. Figure 3The diagram shows a schematic of the structure of an exemplary electronic device for a real-time disaster verification and alarm device based on a large model, provided by the present invention.

[0035] Figure labeling: Data backtracking retrieval module 10, thread trend analysis module 20, indicator cross-verification module 30, disaster decision-making module 40, disaster grid positioning module 50, disaster avoidance analysis module 60, alarm verification and response module 70. Detailed Implementation

[0036] To facilitate understanding of the present invention, a more complete description of the invention will be given below with reference to the accompanying drawings, which illustrate preferred embodiments of the invention. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein; rather, these embodiments are provided so that the disclosure of the invention will be more thorough and complete.

[0037] Furthermore, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0038] In the description of this invention, it should be understood that the terms "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," and "counterclockwise," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0039] Unless otherwise expressly stated, throughout the specification and claims, the term "comprising" or its variations such as "including" or "comprises" shall be understood to include the stated elements or components without excluding other elements or other components.

[0040] The present invention provides a real-time disaster multi-dimensional verification and alarm method and system based on a large model, which is used to solve the technical problem that the existing disaster monitoring and early warning system mainly relies on the hierarchical reporting mechanism, resulting in the delay and lack of accuracy in the transmission of disaster information, which affects the timeliness and accuracy of post-disaster relief work.

[0041] Example 1: See Figure 1 The present invention provides a real-time disaster multi-dimensional verification and alarm method based on a large model, the method comprising:

[0042] Y100: After receiving the disaster warning signal in the monitoring area, the system uses the trigger time of the disaster warning signal as the time reference to perform backtracking retrieval of multi-source disaster correlation data within a preset analysis time window to obtain a heterogeneous disaster dataset.

[0043] Before a disaster occurs, the monitoring area receives disaster early warning signals from sensors, meteorological equipment, and communication base stations. These signals contain information such as the disaster's occurrence, warning type, and trigger time. The trigger time of the disaster early warning signal is used as a time base, i.e., a reference time point, to determine the sequence of disaster occurrence. Based on a preset analysis time window, such as the past 30 minutes or the past hour, data is retrieved by tracing back from the disaster early warning trigger time to obtain multi-source disaster data within that time period. This data covers different areas, such as weather conditions, equipment operating status, and communication load. Disaster-related data is retrieved from multiple data sources, such as tower power supply status data, communication base station load data, and surveillance video streams, and these data are combined to form a heterogeneous disaster dataset.

[0044] Y200: Based on the warning type of the disaster warning signal, activate the dynamic task chain scheduling mechanism, perform multi-priority thread trend analysis on the heterogeneous disaster dataset, and output multi-dimensional indicator analysis results.

[0045] Disaster warning signals include warning types, such as rainstorms and mudslides. Different data processing strategies are selected based on the warning type. For example, in rainstorm scenarios, the analysis of water depth in the video stream is prioritized, while in mudslide scenarios, the analysis of communication base station load status is prioritized. A dynamic task chain scheduling mechanism is activated based on the disaster warning type. This means that based on the current disaster warning type, it automatically determines which tasks need to be executed and dynamically arranges their execution order. Task chain scheduling considers factors such as task priority and dependencies, ensuring that high-priority tasks are executed first. Different data sources are used as input and analyzed through multi-priority threads. The tasks analyzed by each thread are dynamically scheduled according to their priority. For example, in rainstorm warnings, the video analysis component has a higher priority and first identifies and analyzes water depth. Trend analysis is performed on different data sources, identifying the changing trends of these data over time, including changes in signal strength, load increases, and water depth. The final result of the analysis is multi-dimensional indicators that reflect different aspects of disaster development. For example, for rainstorm warnings, these include the rate of change in water depth, the rate of power outages of power towers, and the load of communication base stations.

[0046] Y300: After aligning the spatiotemporal analysis results of the multidimensional indicators, perform cross-verification of the multidimensional indicators and output the disaster verification conclusion.

[0047] Based on the results of multidimensional indicator analysis, spatiotemporal alignment is performed to ensure that data from different dimensions can be compared and integrated in time and space. Cross-validation of the analysis results from different dimensions is then conducted. For example, if there are numerous power outages in the tower power supply status and the water depth in the video stream reaches the warning level, it indicates a relatively severe disaster. Cross-validation aims to improve the accuracy of disaster verification, ensuring that data from multiple dimensions mutually verify each other and preventing misjudgments. After cross-validation, a disaster verification conclusion is output. This conclusion, based on the comprehensive analysis of multidimensional data, represents the current verification result of the disaster situation and is used for subsequent decision-making.

[0048] Y400: Send the disaster verification conclusion to the composite rule judgment engine for disaster decision-making and generate disaster decision instructions.

[0049] The disaster verification results are sent to a composite rule-based decision engine. This engine, based on a rule base including disaster type, severity, and historical disaster cases, makes decisions about the disaster. These rules can be those developed by an expert system or based on machine learning algorithms. The composite rule-based decision engine then generates disaster decision instructions based on the input verification results. These instructions include specific disaster handling plans, such as dispatching rescue teams, evacuating personnel, and activating an emergency response.

[0050] Y500: Based on the disaster verification conclusions and multi-dimensional index analysis results, perform township-level disaster grid positioning and output the disaster boundary coordinate set.

[0051] Based on the disaster assessment conclusions and multi-dimensional indicator analysis results, disaster grid positioning was implemented at the township level. Disaster grid positioning divides the affected area into small grids, each representing a specific geographical region. Each grid exhibits different disaster characteristics, such as water depth and power outages. Grid positioning allows for more precise disaster assessment, facilitating subsequent response and resource allocation. Through grid positioning, a set of disaster boundary coordinates was calculated. These coordinates describe the specific geographical extent of the affected area and serve as the foundational data for subsequent analysis and decision-making.

[0052] Y600: After defining the thermal rescue target point based on the disaster boundary coordinate set in the pre-disaster moving signal population thermal distribution map, perform geological disaster avoidance analysis on the thermal rescue target point and locate the safety buffer zone.

[0053] Before a disaster, a population heat map is constructed based on mobile signaling data, such as mobile phone location information, reflecting densely populated areas. Based on the disaster boundary coordinate set, the population heat map is combined with the disaster-stricken area to identify key thermal rescue targets. These targets are densely populated areas and are prioritized for rescue. Based on the location of these targets, a geological hazard avoidance analysis is conducted to determine which areas pose a disaster risk. Safety buffer zones are then established around these areas to avoid the dangers posed by the disaster. Safety buffer zones are areas designated within the disaster's impact area to protect people and infrastructure. These areas include shelters, areas with disrupted transportation, and severely affected zones. Based on the disaster type and the location of the thermal rescue targets, the coordinate range of the safety buffer zones is calculated and output to ensure that rescue resources can avoid dangerous areas.

[0054] Y700: After packaging the disaster boundary coordinate set, thermal rescue target point and safety buffer into a structured rescue instruction package, execute the alarm verification response of the structured rescue instruction package according to the disaster level of the disaster decision instruction.

[0055] Data on the disaster boundary coordinates, thermal rescue targets, and safety buffer zones are packaged into a structured rescue instruction package. This package is a standardized data packet containing key information needed for disaster response. Based on the disaster level (e.g., Level 1, Level 2) in the disaster decision-making instructions, an alarm verification response is initiated. This response sends alerts to relevant personnel or departments, notifying them to take necessary emergency measures. If the disaster is severe, an emergency response mechanism is triggered, arranging resources for timely rescue. Following the alarm verification response, relevant personnel and equipment will conduct rescue work according to the information in the structured rescue instruction package, including deploying rescue teams, mobilizing resources, arranging traffic control, and implementing evacuation and other emergency response measures.

[0056] In one implementation, the heterogeneous disaster data set includes tower power supply status data, communication base station load data, and full-domain monitoring video streams.

[0057] Tower power supply status data is used to detect the impact of disasters on infrastructure, especially the power supply status of communication towers. If a large-scale power outage occurs, it means that the disaster has caused serious damage to the infrastructure. Communication base station load data can reflect the communication situation in the disaster area. If the base station is overloaded or has excessive load, it may be a signal that people are using the communication network extensively to call for help or exchange information after the disaster. The whole-area monitoring video stream is used to monitor visual changes in the disaster area in real time. For example, in the case of heavy rain, the video stream can be used to detect changes in the depth of surface water or to identify disaster signs such as river channel interruption.

[0058] In one implementation, based on the warning type of the disaster warning signal, a dynamic task chain scheduling mechanism is initiated to perform multi-priority thread trend analysis on the heterogeneous disaster dataset, outputting multi-dimensional indicator analysis results, including:

[0059] Y210: Parse the disaster type identifier code of the disaster warning signal to obtain the warning type; Y220: Perform task chain matching in the scheduling rule base according to the warning type to obtain the thread priority sequence; Y230: According to the thread priority sequence, dynamically schedule the activation order of the video parsing component, the tower status analysis component, and the communication load analysis component, perform cross-dimensional trend collaborative analysis on the tower power supply status data, communication base station load data, and full-domain monitoring video stream, and output the multi-dimensional indicator analysis results.

[0060] Disaster warning signals are generated by meteorological, communication, or other monitoring equipment and contain information indicating the occurrence of a disaster. First, the disaster type identifier code in the warning signal is parsed. This identifier code is a unique identifier for the disaster type; it is either a numerical value or a string. By parsing this identifier code, the specific type of disaster is determined. For example, if the identifier code corresponds to a rainstorm, it indicates that the current disaster is related to a rainstorm. Once the identifier code is parsed, it is converted into a specific disaster type according to a predefined rule base or mapping table for subsequent data processing and task scheduling.

[0061] The scheduling rule base is a collection of rules for disaster response tasks. This base contains task chains and priority arrangements for various disaster types. For example, in the case of a rainstorm disaster, priority should be given to handling water accumulation in video streams, while in the case of a mudslide disaster, communication load and tower power supply status should be analyzed first. Based on the determined disaster type, the corresponding task chain is searched in the scheduling rule base. The task chain defines the tasks to be executed under a specific disaster type and the order in which they are executed. For example, under a rainstorm warning, the task chain includes video stream analysis (water accumulation detection), tower power supply status analysis, and communication load analysis. Each task in the task chain has a different priority. A priority is assigned to each task according to the priority configuration in the rule base. The priority sequence indicates which tasks should be executed first and which can be executed later. Based on these priorities, a thread priority sequence is generated.

[0062] Based on thread priority sequences, a dynamic scheduling mechanism is initiated. Dynamic scheduling means that the order of task execution is adjusted according to the actual situation and predefined priorities. The activation and execution of the video analysis component, tower status analysis component, and communication load analysis component are determined by thread priorities. Specifically, the video analysis component processes video stream data, such as surface water accumulation change rates and river interruption status; the tower status analysis component analyzes the power supply status of towers to check for large-scale power outages; and the communication load analysis component analyzes the load of communication base stations to determine if there is overload or communication interruption. When each task component is activated, cross-dimensional analysis is performed on the data. Cross-dimensional analysis refers to combining data from different sources to analyze the temporal trends of these data and collaboratively analyze the correlations between them. After cross-dimensional trend collaborative analysis, multi-dimensional indicator analysis results are output. These results represent a comprehensive assessment of the disaster situation, covering multiple aspects and providing quantitative evidence for subsequent disaster verification and decision-making.

[0063] In one implementation, if the warning type is a rainstorm scenario warning: S1: The video analysis component is activated first, and the surface water depth change rate and river runoff interruption status are input from the full-domain monitoring video stream identification output as video analysis dimension indicators; S2: If the surface water depth change rate is greater than a preset change threshold, the tower status analysis component is activated, and the tower power supply status data is input to calculate the tower power outage density heatmap and the offline rate increase per unit time as tower power supply dimension indicators; S3: If the tower power outage rate in the tower power outage density heatmap is greater than a preset power outage ratio, and the offline rate increase per unit time is greater than a preset increase threshold, the communication load analysis component is activated, and the communication base station load data is input to calculate the cumulative increase in channel pressure and the proportion of overloaded base stations as communication load dimension indicators, wherein the tower power supply dimension indicators, communication load dimension indicators, and video analysis dimension indicators constitute the multi-dimensional indicator analysis results.

[0064] If the warning type is a rainstorm warning, the video analysis component is given a higher priority and activated first. The main task of this component is to process data from the city-wide monitoring video stream, which comes from devices such as urban surveillance cameras, drone footage, and satellite imagery. The video analysis component receives the city-wide monitoring video stream, and the images and video frames in the stream contain real-time image data from different locations. Based on image processing technologies, such as deep learning image recognition, the video analysis component identifies the surface water depth in the video stream. The surface water depth change rate refers to the degree of change in the water depth at a certain location within a certain time. By identifying pixel changes in the water, the water depth at different locations is calculated, and the change rate, i.e., the speed at which the water depth changes, is further calculated. The water depth can be accurately estimated by comparing consecutive images in the video frames and combining known reference points, such as road markings and building heights.

[0065] In addition to the depth of water accumulation, the video analysis component also analyzes the water flow status of the river and identifies whether there is a river interruption. Heavy rain may cause the river flow to be interrupted or overflow, affecting the safety of the surrounding area. By analyzing the water flow movement in the video stream, such as using optical flow method and optical flow estimation, it can detect whether there are signs of blockage or alteration of the water flow in the river.

[0066] The identified surface water depth change rate and river runoff interruption status are output as video analysis dimensional indicators. These indicators provide basic data for subsequent disaster analysis and can help determine the severity of the disaster and the areas that may be affected.

[0067] If the rate of change of surface water depth output by the video analysis component is greater than the preset change threshold, for example, if the rate of change exceeds a certain water depth or speed, then subsequent steps are triggered. This threshold setting ensures that the tower status analysis component will only be activated when the disaster is severe enough.

[0068] When the rate of change in water accumulation reaches a preset threshold, the tower status analysis component is activated to begin analyzing the damage to disaster-related infrastructure. The goal is to assess whether communication infrastructure has been affected by the rainstorm. The tower status analysis component receives tower power supply status data, including real-time power supply conditions, tower operational status, and outage status. Based on this data, it calculates a tower power outage density heatmap. This heatmap is a spatial distribution map representing the density of tower power outages within a specific area, indicating which areas are most severely affected by the rainstorm and experience significant power supply disruptions. The heatmap identifies areas with severe power outages, which could impact communication services and post-disaster emergency response.

[0069] Another key indicator is the increase in the offline rate per unit time. This increase represents the change in the proportion of communication base stations or towers that are offline (unable to provide services) within a certain period of time. By analyzing the offline data of towers in different time periods, the increase in the offline rate is calculated to determine whether there is a large-scale communication interruption.

[0070] The calculated heat map of power outage density of iron towers and the increase in offline rate per unit time are used as indicators of iron tower power supply. These indicators are output together with video analysis indicators. These data can help determine whether there are communication gaps in the disaster area and whether communication services need to be restored as a priority.

[0071] The analysis determines whether the proportion of power outage towers in the tower outage density heatmap exceeds a preset outage ratio threshold, such as 30% or 50%. If this condition is met, it indicates a significant impact of the disaster on infrastructure (especially communication towers). The analysis also checks whether the increase in the offline rate of communication base stations per unit time exceeds a preset increase threshold. This threshold is determined by historical disaster data or system-set safety standards. For example, if the offline rate of a base station increases by 50% within one hour, an analysis is triggered.

[0072] If both of the above conditions are met, the communication load analysis component is activated. This component's task is to assess the load of communication base stations and detect the pressure on the communication network. The communication load analysis component acquires communication base station load data, including current traffic, number of connections, and the number of users supported by the base station. This data allows for the analysis of the current pressure on the communication network.

[0073] The cumulative increase in channel pressure refers to the increase in channel pressure on communication base stations over a certain period of time. Channel pressure comes from multiple factors, such as an increase in the number of simultaneously connected users and a sharp rise in data traffic. This indicator is used to assess the network's carrying capacity and whether there is a risk of overload or network congestion. The overloaded base station ratio refers to the proportion of base stations that cannot provide normal service due to excessive load during a disaster. If the overloaded base station ratio is high, it indicates that the communication network in the disaster area may face the risk of collapse. Overloaded base stations are usually unable to handle sufficient user requests, which may lead to communication interruptions.

[0074] The calculated cumulative increase in channel pressure and the proportion of overloaded base stations will be output as communication load dimension indicators, reflecting the pressure level and potential fault points of the communication network in the disaster area. Finally, the tower power supply dimension indicators, communication load dimension indicators, and video analysis dimension indicators will be combined to form multi-dimensional indicator analysis results. These results comprehensively demonstrate multiple aspects of the disaster situation, providing detailed data support for disaster verification and decision-making.

[0075] Similarly, if the warning type is a debris flow scenario, the tower status analysis component will be activated first, followed by the communication load analysis component, and then the video parsing component.

[0076] In one implementation, the video parsing component is activated first, and the surface water depth change rate and river runoff interruption status are identified and output from the full-domain monitoring video stream as video parsing dimension indicators, including:

[0077] S11: Based on the historical disaster case database, a distributed rainstorm disaster surface area is set, and the monitoring point spatial filtering of the full-area monitoring video stream is performed to obtain a distributed water accumulation area video stream; S12: The distributed water accumulation area video stream is subjected to pixel-level water accumulation depth identification and tracking based on calibration reference objects, and the distributed water accumulation change rate is calculated and output; S13: The distributed water accumulation change rate is weighted and fused according to the geological permeability coefficient of the distributed rainstorm disaster surface area, and the surface water accumulation depth change rate is output; S14: The boundary of the river monitoring area is retrieved from the GIS system, and the monitoring points of the full-area monitoring video stream are retrieved to obtain multiple fixed-point monitoring video stream sequences of multiple river branches; S15: The multiple fixed-point monitoring video stream sequences are subjected to water flow motion vector analysis based on optical flow method, and multiple river interruption monitoring coordinates are output as the river runoff interruption status.

[0078] Referring to a historical disaster case database, which contains past cases of rainstorm disasters, analysis of these cases can identify common surface waterlogging areas during rainstorm disasters. The historical cases provide information such as the spatial distribution and depth of waterlogged areas under different rainstorm scenarios. Based on historical data, standardized surface areas prone to rainstorm disasters are defined. These areas are frequently identified as having waterlogged conditions during past rainstorm disasters and are used as examples to locate potential waterlogging areas in the current disaster situation.

[0079] Spatial filtering is applied to monitoring points in the overall surveillance video stream to identify areas or locations affected by rainstorm disasters. These filtered monitoring points can then be further analyzed for surface water accumulation and river conditions. Spatial filtering also yields surveillance video streams of distributed waterlogged areas, focusing primarily on regions that have historically experienced waterlogging, thus helping to identify the extent of waterlogging in the current disaster situation.

[0080] By using calibration references, such as road signs, buildings, and lane markings, the scale of waterlogged areas in the video stream is calibrated. These references convert pixel-level information in the video images into actual physical units, such as centimeters or meters. Image processing algorithms, such as deep learning or traditional computer vision algorithms, are used to process the video stream of distributed waterlogged areas. Waterlogged areas in each frame of the video image are identified and tracked through pixel-level analysis. By comparing image changes in the video stream, the water depth at different locations is calculated. For example, changes in water level and the degree of water accumulation can be detected through pixel density and color differences. Based on the pixel-level water depth identification results, the rate of change of water accumulation at each monitoring point is calculated. The rate of change of water accumulation refers to the rate at which the water depth changes over a certain period of time, indicating whether the water is rapidly increasing.

[0081] Geological permeability coefficient is a parameter that measures the water permeability of soil or ground materials. Different regions have different soil and ground structures, such as sand, clay, and rock, with varying permeability coefficients, which affect the accumulation and infiltration rate of surface water during heavy rainfall. Using historical and current geological data, the geological permeability coefficient of each distributed rainstorm disaster surface area is obtained. A higher geological permeability coefficient indicates a faster rate of water infiltration and less water accumulation. The water accumulation change rate of each distributed water accumulation area is weighted and fused. Specifically, the water accumulation change rate is based on the change in water depth over time as identified from the video stream. The weighted fusion is adjusted according to the geological permeability coefficient of each area; if a region has a high geological permeability coefficient, its water accumulation change rate is relatively low; conversely, if the geological permeability coefficient is low, the water accumulation change rate is high. After weighted fusion, the surface water depth change rate is output, which integrates the region's geological permeability and the water accumulation changes in the video stream.

[0082] GIS (Geographic Information System) systems can provide various geographic location-related data, including information on the basins, branches, and dam locations of different rivers. The boundary of the river monitoring area is retrieved from the GIS system. This boundary information includes the distribution of the river basin, the topology of the river channel, and areas potentially threatened by floods. The monitoring area boundary helps determine which areas require special attention regarding water levels. Based on the retrieved river monitoring area boundary, corresponding monitoring points are determined. These monitoring points refer to actual monitoring equipment deployed in or around the river channel, such as cameras and drones. From these monitoring points, corresponding fixed-point monitoring video stream sequences are obtained. Each monitoring point represents a fixed point in the river, and the video stream contains key information such as water flow status and potential blockages in the river channel.

[0083] Optical flow is a computer vision technique used in video stream processing to analyze the motion of objects in images. It calculates the motion vector of each pixel by analyzing pixel changes between consecutive video frames. In river video streams, optical flow is used to perform vector analysis of water flow. This analysis obtains the speed, direction, and flow pattern of the water flow, determining whether there are blockages, backflows, or interruptions. Water flow vector analysis can reveal potential interruption points in the river, such as places where the flow stops or reverses, indicating catastrophic disruptions like dam failures or debris flows. Based on the analysis results, multiple river interruption monitoring coordinates are output. These coordinates represent the specific locations of abnormal water flow in the river, such as stagnation points and backflow points. These coordinates help identify which river branches have been interrupted or blocked, providing crucial information for subsequent emergency decision-making.

[0084] In one implementation, after spatiotemporally aligning the results of the multidimensional indicator analysis, multidimensional indicator cross-verification is performed, and a disaster verification conclusion is output, including:

[0085] Y310: If the multi-dimensional indicator analysis results do not cover the tower power supply dimension indicators, communication load dimension indicators, and video parsing dimension indicators, then terminate the verification process and send the multi-dimensional indicator analysis results to the manual decision-making terminal; Y320: If the multi-dimensional indicator analysis results cover the tower power supply dimension indicators, communication load dimension indicators, and video parsing dimension indicators, then after mapping the tower power supply dimension indicators, communication load dimension indicators, and video parsing dimension indicators based on GIS grid coding spatiotemporal alignment, execute: s1: Extract the height of the power outage tower from the tower power outage density heat map. s2: Binarize the distributed water accumulation change rate using the preset change threshold to obtain the water accumulation verification area polygon; s3: Calculate the spatial overlap area ratio of the high-density clustering area of ​​the power outage tower and the water accumulation verification area polygon; s4: Based on the cumulative increase in channel pressure and the increase in offline rate per unit time, extract the pressure increase node and the offline increase node, and calculate the increase time difference; s5: If the increase time difference meets the preset time difference threshold, and the overlap area ratio is greater than the preset area ratio threshold, then the disaster verification conclusion is a level one verification conclusion.

[0086] Check whether the current multidimensional indicator analysis results cover all the key indicators mentioned above. If these dimensions are not fully covered, it means that the disaster analysis is incomplete or inaccurate, and therefore the verification process cannot continue. At this point, the verification process is terminated, meaning that a complete verification conclusion on the disaster situation cannot be automatically reached. Then, the current multidimensional indicator analysis results are sent to the human decision-making end, where professionals (such as emergency commanders or disaster recovery experts) conduct further analysis and decision-making. These professionals make manual judgments based on existing data and experience to ensure that the disaster response can be initiated as early as possible.

[0087] If the multidimensional indicator analysis results have covered all key dimensions, proceed to the next step: spatiotemporal alignment. Spatiotemporal alignment unifies the various data dimensions in time and space, enabling data from different sources to be compared within the same time window and geographic area. Using GIS (Geographic Information System) grid coding, the disaster area is divided into different grids, each representing a specific geographic region. This helps to accurately map the data to actual geographical locations; for example, each grid might represent a one-kilometer square area. Once spatiotemporal alignment is complete, a series of analytical tasks will be performed based on this aligned data.

[0088] The heat map of power outage density shows the distribution density of power outage towers in different areas. High-density areas indicate that the power outage situation in these areas is more severe and requires priority for power restoration. High-density clusters of power outage towers can be extracted from the heat map; these clusters represent the areas most severely affected during a disaster and should be prioritized for recovery and relief efforts.

[0089] The distributed water accumulation change rate is binarized and partitioned using a preset change threshold. This process divides the waterlogged areas in the disaster area into two categories: water accumulation verification areas, which are areas with a change rate exceeding the threshold and are considered severely flooded, requiring close attention; and non-waterlogged areas, which have a change rate below the threshold and have less water accumulation, and can be disregarded as emergency rescue priorities. After binarization, the severely flooded areas are identified as water accumulation verification areas and converted into polygons to represent the specific boundaries of the disaster area.

[0090] The overlapping area refers to the portion of the polygonal area where power outages and flooding verification zones overlap. This represents a severe situation where both power outages and flooding have occurred simultaneously within the same geographical area. The overlapping area percentage refers to the proportion of the overlapping area to the total area. By calculating this percentage, the importance of these overlapping disaster areas can be quantified, determining whether a priority response is needed. A large overlapping area percentage indicates a severe cumulative effect of the disaster within the same area, meaning that flooding and power outages are significantly impacting the same region, necessitating priority rescue and recovery efforts for that area.

[0091] The cumulative increase in channel pressure indicates a gradual increase in pressure on the communication network over a period of time. For example, a gradual increase in base station load means more user connections, leading to a surge in network traffic. This increase reflects whether the communication network is approaching overload. The increase in offline rate per unit time indicates the rate of change in the offline status of communication base stations. If the offline rate of communication base stations increases significantly within a certain period, it means that a large number of base stations have lost signal or are not functioning properly, indicating the severity of the disaster. Pressure increase nodes and offline increase nodes represent the critical moments of changes in channel pressure and base station offline rate, respectively. These nodes indicate the moment when the impact of the disaster begins to appear significantly and are important bases for disaster assessment. The increase time difference refers to the time difference between the pressure increase node and the offline increase node. This time difference reflects the response speed and deterioration speed of the communication network under the influence of the disaster. If the time difference is small, it indicates that the disaster is spreading rapidly and communication interruptions are intensifying.

[0092] The preset time difference threshold is used to determine whether the time difference of the increase is small enough, indicating how quickly the communication network is affected after a disaster occurs. The preset area ratio threshold is used to determine whether the proportion of overlapping areas exceeds a certain standard, indicating the spatial concentration of the disaster. If the time difference of the increase meets the preset time difference threshold, i.e., the disaster is spreading rapidly, and the proportion of overlapping area is greater than the preset area ratio threshold, i.e., the disaster has a high degree of spatial overlap, then the disaster is determined to be very serious, and the verification conclusion is a level one verification conclusion, i.e., 100% confirmation of the disaster.

[0093] Furthermore, the Level 1 verification conclusion requires the simultaneous fulfillment of both spatial consistency confirmation criteria and causal correlation confirmation criteria. Spatial consistency confirmation criteria indicate a certain degree of overlap in area, suggesting a strong spatial correlation between the disaster and the disaster. Causal correlation confirmation criteria indicate a small time difference in the rate of increase, suggesting a strong temporal and spatial correlation between the disaster and the impact on the communication network, possibly caused by flooding or other disaster-related factors. If only one of these criteria is met, it indicates a possible disaster, but further manual verification is required. In this case, the verification conclusion is Level 2, meaning the severity of the disaster needs to be manually assessed. If neither spatial consistency nor causal correlation confirmation criteria are triggered, it indicates insufficient disaster impact or inconsistent analysis results, leading to a conclusion of no disaster (a rejection conclusion), and an emergency response may be postponed or not initiated.

[0094] In one implementation, township-level disaster grid positioning is performed based on the disaster verification conclusions and multi-dimensional index analysis results, outputting a disaster boundary coordinate set, including:

[0095] Y510: If the disaster verification conclusion is a Level 1 verification conclusion, then construct the disaster-affected grid array of the monitoring area; Y520: After projecting the high-density clustering area of ​​the power outage tower, the polygon of the water accumulation verification area, and the coordinates of multiple river interruption monitoring onto the disaster-affected grid array, merge the disaster-affected areas based on the watershed algorithm and output the disaster-affected boundary coordinate set.

[0096] When a Level 1 verification conclusion is reached, it indicates that the disaster situation is very serious. Based on the Level 1 verification conclusion, a disaster-affected grid array is constructed. The disaster-affected grid array is implemented through a GIS system, which divides the disaster area into multiple grids. Each grid represents a relatively independent area, which facilitates more accurate location of the disaster situation and provides data support for subsequent rescue decisions.

[0097] The high-density clustering area of ​​power outage towers, the polygonal area of ​​waterlogged verification, and the coordinates of multiple river interruption monitoring points are projected onto a disaster-affected grid array. The watershed algorithm, an image processing algorithm used in geographic information systems for region segmentation, separates or merges regions based on spatial variations (e.g., the severity or density of the disaster). Based on this projected data, the watershed algorithm is used to merge regions with different disaster data. For example, if the power outage area and the waterlogged area overlap, and there are other disasters near the river interruption points, these areas are merged into a larger, more catastrophic area. The merged disaster-affected area more accurately reflects the overall disaster situation, ensuring that relief resources are efficiently allocated to the most severely affected areas. A disaster boundary coordinate set is generated based on the merged area. This coordinate set describes the specific boundaries of the disaster area, providing detailed geographic information for subsequent emergency response and resource allocation.

[0098] In one implementation, after defining the thermal rescue target point based on the disaster-affected boundary coordinate set in the pre-disaster moving signal population thermal distribution map, a geological hazard avoidance analysis is performed on the thermal rescue target point to locate a safety buffer zone, including:

[0099] Y610: The population thermal distribution map of the mobile signal is segmented using the disaster-affected boundary coordinate set to obtain a disaster-affected grid thermal distribution sub-map; Y620: After segmenting the disaster-affected grid thermal distribution sub-map based on the preset population thermal value, target point correction is performed by combining non-residential attribute thermal points from the POI database to define the thermal rescue target point; Y630: Road surface water depth identification and road collapse feature extraction are performed based on the full-domain monitoring video stream to locate the coordinates of distributed road breakpoints; Y640: Spatial restricted area mask overlay is performed on the thermal rescue target point based on the coordinates of the distributed road breakpoints and the polygon of the water accumulation verification area to output the safety buffer.

[0100] By combining the disaster boundary coordinate set with a population heat map based on mobile signals (such as mobile phone location data), which shows the density of population in different areas, the population heat map is divided into disaster-affected grid heat map sub-maps. Each grid represents a small area. The resulting sub-maps show the disaster situation and population density within different grid areas. These sub-maps help prioritize the treatment of severely affected and densely populated areas and develop targeted relief plans.

[0101] Based on the preset population heat map value, the heat map of the disaster grid is binarized and segmented. Binarization distinguishes hot spots from non-hot spots by comparing the population density in the heat map with the set threshold, which facilitates subsequent analysis and rescue decisions. This process divides the areas in the map into two categories: severely affected areas, such as densely populated areas with severe disasters, and areas with less severe disasters or less impact.

[0102] The POI (Point of Interest) database contains key location data across different regions, such as commercial areas, residential areas, schools, and hospitals. The POI data also includes heatmaps of non-residential areas, such as shopping malls and industrial zones, which typically have lower rescue needs. By combining these non-residential heatmaps with the disaster-affected grid heatmap sub-map, heat-based rescue targets are refined, avoiding excessive concentration of resources in non-residential areas and thus more effectively allocating resources to severely affected areas that actually require rescue. Finally, by combining population heatmaps and the refined POI data, heat-based rescue targets are defined. These targets are densely populated and severely affected areas, prioritizing resource allocation, evacuation, and rescue efforts.

[0103] By monitoring the entire area via video stream, and leveraging image processing and computer vision technologies, the depth of water accumulation on roads can be identified. By analyzing pixel changes in the ground within the video, the water depth in different areas can be calculated. This helps identify which roads are impassable due to water accumulation and which areas are most severely flooded. In addition to water accumulation, features of road collapses or other disasters are extracted. If there are ground subsidence or road breaks, corresponding features such as cracks and obstructions are detected in the video stream and identified using image analysis algorithms. Through video stream analysis, distributed road breakpoint coordinates are located—points where roads are impassable due to water accumulation, landslides, or other disasters. These breakpoint coordinates provide precise geographic information for subsequent rescue and emergency traffic planning.

[0104] Combining the obtained distributed road breakpoint coordinates and the polygon of the flood verification area (the flood verification area refers to areas severely flooded due to disasters such as rainstorms), and the road breakpoint coordinates representing areas with traffic obstructions caused by the disaster, the combination of these two areas indicates which areas, due to traffic disruptions, flooding, or other factors, cannot serve as rescue routes or action zones and must be designated as restricted areas. Mask overlay refers to adding a spatial restricted area mask to the thermal rescue target area. This mask identifies areas that cannot be entered or passed through due to road breakpoints or flooding. Finally, a safety buffer zone is output. This zone refers to the safe area that needs priority protection and evacuation during a disaster. The buffer zone ensures that rescue resources do not enter areas that are already impassable, while prioritizing rescue efforts in densely populated and severely affected areas. The safety buffer zone includes accessible areas and can also be designated as a refuge and resettlement area, ensuring that rescue personnel and resources can safely and efficiently enter the disaster area.

[0105] One implementation also includes:

[0106] S131: If the rate of change of surface water depth is less than a preset change threshold, the activation process of the tower status analysis component and the communication load analysis component is terminated, and the tower power supply dimension index is used as the result of the multidimensional index analysis; S132: If the tower power outage rate is less than a preset power outage ratio, and / or the increase in the offline rate per unit time is less than a preset increase threshold, the activation process of the communication load analysis component is terminated, and the tower power supply dimension index and the communication load dimension index are used as the result of the multidimensional index analysis.

[0107] If the rate of change in surface water accumulation is less than a preset threshold, the activation process of the tower status analysis component and the communication load analysis component will be terminated. This means that the system determines that, under the current circumstances, the water accumulation problem will not have a significant impact on power supply or communication load, and therefore further tower status and communication load analysis is not required. In this case, the tower power supply dimension index is used as the result of multi-dimensional index analysis because the rate of change in water accumulation is low and insufficient to affect other disaster indicators. Therefore, the tower power supply status remains the primary dimension to be monitored, and other dimensions (such as communication load and video analysis) will not be further analyzed.

[0108] The following criteria are set: the tower power outage rate is less than a preset power outage ratio, and the increase in the offline rate per unit time is less than a preset threshold. When either or both of these criteria are triggered simultaneously, the communication load analysis component is no longer activated. In this case, communication load analysis is considered unnecessary because there are no significant problems with the power supply to the towers and the stability of the communication base stations. Under these circumstances, both the tower power supply dimension indicators and the communication load dimension indicators are output as part of the multi-dimensional indicator analysis results.

[0109] Example 2: Based on the same inventive concept as the real-time disaster multi-dimensional verification and alarm method based on a large model in the foregoing examples, this invention provides a real-time disaster multi-dimensional verification and alarm system based on a large model. See [link to example]. Figure 2 As shown, the system includes:

[0110] The data backtracking and retrieval module 10 is used to retrieve multi-source disaster-related data backtracking within a preset analysis time window after receiving a disaster warning signal from the monitoring area, using the trigger time of the disaster warning signal as the time base, to obtain a heterogeneous disaster dataset. The thread trend analysis module 20 is used to activate a dynamic task chain scheduling mechanism according to the warning type of the disaster warning signal, perform multi-priority thread trend analysis on the heterogeneous disaster dataset, and output multi-dimensional indicator analysis results. The indicator cross-verification module 30 is used to perform multi-dimensional indicator cross-verification after spatiotemporally aligning the multi-dimensional indicator analysis results, and output disaster verification conclusions. The disaster decision module 40 is used to send the disaster verification conclusions to the composite rule. The judgment engine makes disaster situation decisions and generates disaster situation decision instructions; the disaster situation grid positioning module 50 is used to perform township-level disaster situation grid positioning based on the disaster situation verification conclusion and multi-dimensional index analysis results, and output the disaster-affected boundary coordinate set; the disaster avoidance analysis module 60 is used to define the thermal rescue target point based on the disaster-affected boundary coordinate set in the pre-disaster moving signal population thermal distribution map, and perform geological disaster avoidance analysis of the thermal rescue target point to locate the safety buffer zone; the alarm verification response module 70 is used to package the disaster-affected boundary coordinate set, thermal rescue target point and safety buffer zone into a structured rescue instruction package, and execute the alarm verification response of the structured rescue instruction package according to the disaster level of the disaster situation decision instructions.

[0111] In one implementation, the heterogeneous disaster data set includes tower power supply status data, communication base station load data, and full-domain monitoring video streams.

[0112] In one implementation, the thread trend analysis module 20 is used for:

[0113] The disaster type identifier code of the disaster warning signal is parsed to obtain the warning type; task chain matching is performed in the scheduling rule base according to the warning type to obtain the thread priority sequence; the activation order of the video parsing component, tower status analysis component and communication load analysis component is dynamically scheduled according to the thread priority sequence, and cross-dimensional trend collaborative analysis is performed on the tower power supply status data, communication base station load data and full-domain monitoring video stream, and the multi-dimensional indicator analysis results are output.

[0114] In one implementation, if the warning type is a rainstorm scenario warning:

[0115] S1: First, activate the video parsing component, input the full-domain monitoring video stream identification output of the surface water depth change rate and river runoff interruption status as video parsing dimension indicators; S2: If the surface water depth change rate is greater than a preset change threshold, then activate the tower status analysis component, input the tower power supply status data, calculate the tower power outage density heatmap and the unit time offline rate increase as tower power supply dimension indicators; S3: If the tower power outage rate of the tower power outage density heatmap is greater than a preset power outage ratio, and the unit time offline rate increase is greater than a preset increase threshold, then activate the communication load analysis component, input the communication base station load data, calculate the channel pressure cumulative increase and the overloaded base station ratio as communication load dimension indicators, wherein the tower power supply dimension indicators, communication load dimension indicators, and video parsing dimension indicators constitute the multi-dimensional indicator analysis results.

[0116] In one implementation:

[0117] Based on a historical disaster case database, distributed rainstorm disaster surface areas are defined. Spatial filtering of monitoring points in the overall monitoring video stream is performed to obtain distributed waterlogged area video streams. Pixel-level waterlogging depth identification and tracking are performed on the distributed waterlogged area video streams based on calibration references, and the distributed waterlogging change rate is calculated and output. The distributed waterlogging change rate is weighted and fused according to the geological permeability coefficient of the distributed rainstorm disaster surface areas to output the surface waterlogging depth change rate. The river monitoring area boundaries are retrieved from the GIS system, and monitoring points in the overall monitoring video streams are retrieved to obtain multiple fixed-point monitoring video stream sequences for multiple river branches. Water flow vector analysis based on optical flow is performed on the multiple fixed-point monitoring video stream sequences to output multiple river interruption monitoring coordinates as the river runoff interruption status.

[0118] In one implementation, the indicator cross-verification module 30 is used for:

[0119] If the multidimensional indicator analysis results do not cover the tower power supply dimension indicators, communication load dimension indicators, and video analysis dimension indicators, the verification process is terminated, and the multidimensional indicator analysis results are sent to the manual decision-making end. If the multidimensional indicator analysis results cover the tower power supply dimension indicators, communication load dimension indicators, and video analysis dimension indicators, then after mapping the tower power supply dimension indicators, communication load dimension indicators, and video analysis dimension indicators based on GIS grid coding spatiotemporal alignment, the following steps are executed: s1: Extract the high-density clustering area of ​​the power outage towers from the tower power outage density heat map; s2: Perform binary partitioning of the distributed water accumulation change rate using the preset change threshold to obtain the water accumulation verification area polygon; s3: Calculate the spatial overlap area ratio of the high-density clustering area of ​​the power outage towers and the water accumulation verification area polygon; s4: Based on the cumulative increase in channel pressure and the increase in offline rate per unit time, extract the pressure increase node and the offline increase node, and calculate the increase time difference; s5: If the increase time difference meets the preset time difference threshold, and the overlap area ratio is greater than the preset area ratio threshold, then the disaster verification conclusion is a level one verification conclusion.

[0120] In one implementation, the disaster grid positioning module 50 is used for:

[0121] If the disaster verification conclusion is a Level 1 verification conclusion, then construct the disaster-affected grid array of the monitoring area; after projecting the high-density clustering area of ​​the power outage tower, the polygon of the water accumulation verification area, and the coordinates of multiple river interruption monitoring onto the disaster-affected grid array, merge the disaster-affected areas based on the watershed algorithm, and output the disaster-affected boundary coordinate set.

[0122] In one implementation, the disaster avoidance analysis module 60 is used for:

[0123] The population thermal distribution map of the mobile signal is segmented using the disaster-affected boundary coordinate set to obtain a disaster-affected grid thermal distribution sub-map. After segmenting the disaster-affected grid thermal distribution sub-map based on a preset population thermal value, target point correction is performed by combining non-residential attribute thermal points from the POI database to define the thermal rescue target point. Road surface water depth identification and road collapse feature extraction are performed based on the full-domain monitoring video stream to locate the coordinates of distributed road breakpoints. The spatial restricted area mask of the thermal rescue target point is superimposed based on the coordinates of the distributed road breakpoints and the polygon of the water accumulation verification area to output the safety buffer.

[0124] In one implementation:

[0125] If the rate of change of surface water depth is less than a preset threshold, the activation process of the tower status analysis component and the communication load analysis component is terminated, and the tower power supply dimension index is used as the result of the multidimensional index analysis; if the tower power outage rate is less than a preset power outage ratio, and / or the increase in the offline rate per unit time is less than a preset increase threshold, the activation process of the communication load analysis component is terminated, and the tower power supply dimension index and the communication load dimension index are used as the result of the multidimensional index analysis.

[0126] The above description is merely a preferred embodiment of the present invention, but the scope of protection of the present invention 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 the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

[0127] The foregoing description of specific exemplary embodiments of the invention is for illustrative and explanatory purposes. These descriptions are not intended to limit the invention to the precise forms disclosed, and it will be apparent that many changes and variations can be made in accordance with the foregoing teachings. The exemplary embodiments were chosen and described in order to explain the specific principles of the invention and its practical application, thereby enabling those skilled in the art to implement and utilize various different exemplary embodiments of the invention, as well as various different choices and variations. The scope of the invention is intended to be defined by the claims and their equivalents.

Claims

1. A real-time disaster multi-dimensional verification and alarm method based on a large model, characterized in that, The method includes: After receiving the disaster warning signal from the monitoring area, the multi-source disaster association data is retrieved back within a preset analysis time window, using the trigger time of the disaster warning signal as the time reference, to obtain a heterogeneous disaster dataset. Based on the warning type of the disaster warning signal, a dynamic task chain scheduling mechanism is activated to perform multi-priority thread trend analysis on the heterogeneous disaster dataset and output multi-dimensional indicator analysis results. After aligning the spatiotemporal analysis results of the multidimensional indicators, perform cross-verification of the multidimensional indicators and output the disaster verification conclusion. The disaster verification conclusion is sent to the composite rule judgment engine for disaster decision-making, and a disaster decision instruction is generated. Based on the disaster verification conclusions and multi-dimensional index analysis results, township-level disaster grid positioning is performed, and the disaster boundary coordinate set is output. After defining the thermal rescue target points based on the disaster boundary coordinate set in the pre-disaster moving signal population thermal distribution map, a geological disaster avoidance analysis of the thermal rescue target points is performed to locate the safety buffer zone; After packaging the disaster boundary coordinate set, thermal rescue target point and safety buffer into a structured rescue instruction package, the alarm verification response of the structured rescue instruction package is executed according to the disaster level of the disaster decision instruction.

2. The real-time disaster multi-dimensional verification and alarm method based on a large model as described in claim 1, characterized in that, The heterogeneous disaster data set includes tower power supply status data, communication base station load data, and full-domain monitoring video streams.

3. The real-time disaster multi-dimensional verification and alarm method based on a large model as described in claim 2, characterized in that, Based on the warning type of the disaster warning signal, a dynamic task chain scheduling mechanism is activated to perform multi-priority thread trend analysis on the heterogeneous disaster dataset, and output multi-dimensional indicator analysis results, including: Parse the disaster type identifier code of the disaster warning signal to obtain the warning type; Based on the warning type, task chain matching is performed in the scheduling rule base to obtain the thread priority sequence; Based on the thread priority sequence, the activation order of the video parsing component, tower status analysis component, and communication load analysis component is dynamically scheduled to perform cross-dimensional trend collaborative analysis on the tower power supply status data, communication base station load data, and full-domain monitoring video stream, and output the multi-dimensional indicator analysis results.

4. The real-time disaster multi-dimensional verification and alarm method based on a large model as described in claim 3, characterized in that, If the warning type is a rainstorm scenario warning: S1: Prioritize activating the video parsing component, input the full-domain monitoring video stream to identify and output the surface water depth change rate and river runoff interruption status as video parsing dimension indicators; S2: If the rate of change of surface water depth is greater than the preset change threshold, the tower status analysis component is activated, the tower power supply status data is input, and the tower power outage density heat map and the increase in offline rate per unit time are calculated as the tower power supply dimension indicators. S3: If the tower power outage rate in the tower power outage density heatmap is greater than the preset power outage ratio, and the increase in the offline rate per unit time is greater than the preset increase threshold, then the communication load analysis component is activated, and the cumulative increase in channel pressure and the proportion of overloaded base stations are calculated by inputting the communication base station load data as communication load dimension indicators. The tower power supply dimension indicator, the communication load dimension indicator, and the video parsing dimension indicator constitute the multi-dimensional indicator analysis result.

5. The real-time disaster multi-dimensional verification and alarm method based on a large model as described in claim 4, characterized in that, The video parsing component is activated first, and the surface water depth change rate and river runoff interruption status are identified and output from the full-domain monitoring video stream as video parsing dimension indicators, including: Based on the historical disaster case database, a distributed rainstorm disaster surface area is set up, and the monitoring point spatial filtering of the full-area monitoring video stream is performed to obtain the distributed water accumulation area video stream; For the video stream of the distributed water accumulation area, pixel-level water accumulation depth recognition and tracking are performed based on the calibration reference object, and the distributed water accumulation change rate is calculated and output. The distributed water accumulation change rate is weighted and fused based on the geological permeability coefficient of the surface area of ​​the distributed rainstorm disaster, and the surface water depth change rate is output. The boundary of the river monitoring area is retrieved from the GIS system, and the monitoring points of the full-area monitoring video stream are retrieved to obtain multiple fixed-point monitoring video stream sequences of multiple river branches; Perform water flow motion vector analysis based on optical flow on the multiple fixed-point monitoring video stream sequences, and output multiple river interruption monitoring coordinates as the river runoff interruption status.

6. The real-time disaster multi-dimensional verification and alarm method based on a large model as described in claim 5, characterized in that, After aligning the spatiotemporal analysis results of the multidimensional indicators, perform cross-validation of the multidimensional indicators and output disaster verification conclusions, including: If the multidimensional indicator analysis results do not cover the tower power supply dimension indicators, communication load dimension indicators, and video parsing dimension indicators, the verification process will be terminated and the multidimensional indicator analysis results will be sent to the manual decision-making terminal. If the multidimensional index analysis results cover the tower power supply dimension index, communication load dimension index, and video parsing dimension index, then after mapping the tower power supply dimension index, communication load dimension index, and video parsing dimension index based on GIS grid coding spatiotemporal alignment, the following steps are executed: s1: Extract high-density clusters of power-off towers from the heat map of power-off density of the towers; s2: The distributed water accumulation rate is binarized and partitioned using the preset change threshold to obtain a polygon of the water accumulation verification area; s3: Calculate the spatial overlap area ratio of the polygons in the high-density clustering area of ​​the power outage tower and the water accumulation verification area; s4: Based on the cumulative increase in channel pressure and the increase in offline rate per unit time, extract the pressure increase node and the offline increase node, and calculate the increase time difference; s5: If the increase time difference meets the preset time difference threshold and the overlap area ratio is greater than the preset area ratio threshold, then the disaster verification conclusion is a level one verification conclusion.

7. The real-time disaster multi-dimensional verification and alarm method based on a large model as described in claim 6, characterized in that, Based on the disaster verification conclusions and multi-dimensional indicator analysis results, township-level disaster grid positioning is performed, and the disaster boundary coordinate set is output, including: If the disaster verification conclusion is a Level 1 verification conclusion, then construct a disaster-affected grid array for the monitoring area; After projecting the high-density clustering area of ​​the power outage tower, the polygon of the water accumulation verification area, and the coordinates of multiple river interruption monitoring areas onto the disaster-affected grid array, the disaster-affected areas are merged based on the watershed algorithm, and the disaster-affected boundary coordinate set is output.

8. The real-time disaster multi-dimensional verification and alarm method based on a large model as described in claim 7, characterized in that, After defining the thermal rescue target points based on the disaster boundary coordinate set and the pre-disaster moving signal population thermal distribution map, a geological hazard avoidance analysis is performed on the thermal rescue target points to locate the safety buffer zone, including: The population thermal distribution map of the moving signal is segmented using the disaster-affected boundary coordinate set to obtain a disaster-affected grid thermal distribution sub-map; After binarizing the disaster-affected grid thermal distribution sub-map based on the preset population thermal value, the target point is corrected by combining the non-residential attribute thermal points in the POI database, and the thermal rescue target point is defined. Based on the full-domain monitoring video stream, the depth of water accumulation on the road surface and the feature extraction of road collapse are performed to locate the coordinates of distributed road breakpoints. Based on the coordinates of the distributed road breakpoints and the polygon of the water accumulation verification area, the spatial restricted area mask of the thermal rescue target point is superimposed, and the safety buffer is output.

9. The real-time disaster multi-dimensional verification and alarm method based on a large model as described in claim 4, characterized in that, Also includes: If the rate of change of surface water depth is less than a preset threshold, the activation process of the tower status analysis component and the communication load analysis component will be terminated, and the tower power supply dimension index will be used as the result of the multidimensional index analysis. If the tower power outage rate is less than the preset power outage ratio, and / or the increase in the offline rate per unit time is less than the preset increase threshold, then the activation process of the communication load analysis component will be terminated, and the tower power supply dimension index and the communication load dimension index will be used as the results of the multi-dimensional index analysis.

10. A real-time disaster multi-dimensional verification and alarm system based on a large model, characterized in that, For implementing the method steps of any one of claims 1 to 9, including: The data backtracking and retrieval module is used to retrieve multi-source disaster-related data backtracking within a preset analysis time window after receiving a disaster warning signal from the monitoring area, using the trigger time of the disaster warning signal as the time reference, to obtain a heterogeneous disaster dataset. The thread trend analysis module is used to activate a dynamic task chain scheduling mechanism based on the warning type of the disaster warning signal, perform multi-priority thread trend analysis on the heterogeneous disaster dataset, and output multi-dimensional indicator analysis results. The indicator cross-verification module is used to perform multi-dimensional indicator cross-verification after spatiotemporally aligning the analysis results of the multi-dimensional indicators, and output the disaster verification conclusion. The disaster decision-making module is used to send the disaster verification conclusion to the composite rule judgment engine for disaster decision-making and generate disaster decision instructions; The disaster grid positioning module is used to perform township-level disaster grid positioning based on the disaster verification conclusion and multi-dimensional index analysis results, and output the disaster boundary coordinate set. The disaster avoidance analysis module is used to perform geological disaster avoidance analysis on the thermal rescue target points after the thermal rescue target points are delineated based on the disaster boundary coordinate set in the pre-disaster moving signal population thermal distribution map, and to locate the safety buffer zone. The alarm verification and response module is used to package the disaster boundary coordinate set, thermal rescue target point and safety buffer into a structured rescue instruction package, and then execute the alarm verification and response of the structured rescue instruction package according to the disaster level of the disaster decision instruction.

Images