Elastic integration method and system for multi-source heterogeneous disaster investigation and evaluation model

Abstract

The invention relates to the technical field of intelligent disaster assessment, in particular to a multi-source heterogeneous disaster investigation and assessment model elastic integration method and system, and the method comprises the following steps: generating an assessment reference layer based on multi-source heterogeneous data, recognizing abnormal fluctuation and influence factor deviation, screening key nodes and combinations, and extracting a propagation trajectory and a dynamic path. And analyzing trend and efficiency superposition characteristics, and outputting a disaster behavior linkage response control instruction set. According to the method, a space-time identification mechanism is introduced through structured recombination of multi-source heterogeneous data, so that multi-dimensional unified expression and time sequence continuity of disaster information are realized, abnormal intervals and factor deviations are identified, interference data are effectively eliminated, disaster identification precision and evaluation consistency are improved, and positioning capability is enhanced; the disaster diffusion direction is determined through area offset and trajectory recognition, recovery efficiency information is superposed to analyze response association, a control instruction set with spatial-temporal characteristics is output, and closed-loop regulation and control of disaster assessment, prediction and response are achieved.

Description

Technical Field

[0001] This invention relates to the field of intelligent disaster assessment technology, and in particular to a flexible integration method and system for multi-source heterogeneous disaster investigation and assessment models. Background Technology

[0002] The field of intelligent disaster assessment technology involves the effective assessment and analysis of disasters using various technical means. Through data collection, analysis, and model building, it accurately determines the degree of disaster impact, predicts disaster trends, and then formulates reasonable emergency response and post-disaster recovery measures. This technical field encompasses a variety of methods, including disaster monitoring based on remote sensing data, disaster information acquisition and processing, disaster assessment model construction, data integration and analysis, etc. The core aspects of intelligent disaster assessment mainly involve the multi-source heterogeneous integration of disaster data, the accuracy and robustness of assessment models, real-time updating and processing of disaster information, and subsequent disaster recovery and planning. By using multiple sensors and data acquisition methods, combined with intelligent algorithms and models, the efficiency and accuracy of disaster assessment are improved, especially in real-time response and early warning after a disaster occurs, playing a crucial role.

[0003] The traditional flexible integration method for multi-source heterogeneous disaster investigation and assessment models refers to the integration and analysis of data from different sources and of different types to establish a unified disaster assessment model. Traditional methods utilize data sources including remote sensing imagery, geographic information data, and meteorological data. These different data sources suffer from format differences, massive data volumes, and high timeliness requirements. Existing solutions rely on manual data preprocessing and integration, or on disaster assessment based on fixed models. Traditional methods have limitations when dealing with complex disaster scenarios, particularly in terms of the accuracy of disaster type and post-disaster recovery assessment, failing to fully leverage the potential of the data.

[0004] Existing technologies, when processing multi-source heterogeneous disaster data, rely on manual preprocessing or fixed assessment models. Faced with problems such as differences in data formats, large quantities, and time sensitivity, they lack effective data structure reconstruction capabilities, which easily leads to insufficient information integration and data redundancy. There is ambiguity in identifying the relationship between disaster impact factors and actual disaster distribution, resulting in a slow response of assessment models to disaster change trends. In particular, during the post-disaster recovery process, they cannot reflect regional change information in a timely manner, causing delays in adjusting response strategies. For example, in rapidly evolving disasters such as flash floods or earthquakes, fixed models cannot identify short-term sudden variations, which can easily lead to an imbalance in the allocation of emergency resources and reduce the accuracy and timeliness of post-disaster recovery. Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of existing technologies by proposing a flexible integration method and system for multi-source heterogeneous disaster investigation and assessment models.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a method for elastic integration of multi-source heterogeneous disaster investigation and assessment models, comprising the following steps: S1: Based on the initial input of multi-source heterogeneous data, the data from different sources in the disaster investigation and assessment model are restructured in a structured manner, and combined with time labels and spatial identifiers to generate a disaster assessment baseline layer; S2: Based on the disaster assessment benchmark layer, analyze the correlation between disaster impact factors and disaster distribution characteristics, identify abnormal disaster fluctuation ranges and impact factor deviation ranges, screen disaster units that exceed preset thresholds, remove related interference data, and generate a disaster anomaly diagnosis layer. S3: Based on the disaster anomaly diagnosis layer, extract the disaster distribution status and the change value of influencing factors, identify key disaster nodes and combinations of influencing factors, screen disaster units that conform to known disaster patterns, and overlay location codes to mark the corresponding areas to generate a disaster impact behavior distribution map; S4: Based on the continuous change information of multiple time periods in the disaster impact behavior distribution map, identify the disaster offset of adjacent areas in time periods, extract the offset direction and propagation trajectory, filter out the area segments with the same continuous offset direction, and obtain the disaster dynamic tracking path segment.

[0007] As a further embodiment of the present invention, the disaster assessment benchmark layer includes a disaster distribution feature value set, an influencing factor attribute raster, and a spatial operation parameter surface; the disaster anomaly diagnosis layer includes disaster fluctuation identification results, influencing factor deviation markers, and interference data removal masks; the disaster impact behavior distribution map includes a target disaster candidate set, key node response templates, and regional location codes; and the disaster dynamic tracking path segment includes a continuous offset trend trajectory, a path direction vector set, and regional offset nodes.

[0008] As a further aspect of the present invention, the step of obtaining the disaster assessment benchmark layer specifically includes: S111: Based on the initial input of multi-source heterogeneous data, collect data from different sources in the disaster assessment model, combine time labels to perform frame sequence matching, remove disaster units in the disaster distribution sequence whose difference between adjacent frames exceeds the fluctuation critical threshold, and generate a multi-dimensional disaster matrix; S112: Based on the multidimensional disaster matrix, extract the intensity of influencing factors and disaster distribution data under the same time and space identifiers, filter out records with missing items, and generate a valid disaster operation factor data group; S113: Call the effective disaster operation factor data group, identify the standard deviation of the influence factor intensity in the disaster unit group with the same disaster operation mode and normalize it, calculate the normalized disaster difference intensity value, reclassify the disaster units in the region, and establish a disaster assessment benchmark layer.

[0009] As a further aspect of the present invention, the steps for obtaining the disaster anomaly diagnosis layer are specifically as follows: S211: Based on the disaster assessment benchmark layer, extract the disaster distribution status and influencing factor data in the layer, match and compare the fluctuation value of the disaster unit with the interval of the influencing factor, analyze the corresponding relationship, and obtain the disaster fluctuation interval; S212: Call the disaster fluctuation range, combine it with the distribution status of disaster units in the assessment layer, determine whether it falls into the specified range, identify the difference of disaster units that are not in the range, determine the objects to be removed based on the difference, and obtain a list of disaster unit numbers to be removed. S213: Call the list of disaster unit numbers to be removed, delete the interference data associated with the corresponding numbers in the evaluation layer, reorganize the spatial distribution of the remaining disaster units, and generate a disaster anomaly diagnosis layer.

[0010] As a further aspect of the present invention, the steps for obtaining the disaster impact behavior distribution map are as follows: S311: Based on the disaster anomaly diagnosis layer, extract the distribution status and influencing factor change values ​​of the disaster units in the layer, combine the differentiated disaster nodes with the corresponding influencing factor data, identify typical disaster pattern areas, and obtain disaster pattern characterization combinations; S312: Call the disaster model characterization combination, detect the fluctuation differences of disaster units and the degree of coordinated change of influencing factors in the combination area, aggregate and compare the response values ​​of disaster units at key nodes, identify the areas with potential disaster impact behavior, and obtain the disaster impact sensitive area index set; S313: Call the disaster-affected sensitive area index set, match the disaster situation units and disaster pattern templates in the sensitive areas, filter the disaster situation units that meet the feature requirements, and mark the location coding information of the corresponding areas to generate a disaster impact behavior distribution map.

[0011] As a further aspect of the present invention, the step of obtaining the disaster dynamic tracking path segment specifically includes: S411: Based on the continuous change information of multiple time periods in the disaster impact behavior distribution map, identify the geometric center offset of adjacent areas in the time series, extract the offset direction vector and displacement length, analyze the angle difference of the consistency of the offset direction between adjacent areas, and obtain continuous offset direction data of the disaster situation. S412: Call the disaster continuous offset direction data, calculate the angle fluctuation dispersion value according to the changing trend of the angle difference, extract the area segment with stable offset direction, filter the continuous segment with angle change below the preset threshold, and obtain stable offset segment information. S413: Call the stable offset segment information, combine the offset direction angle in each segment, the intensity of disaster distribution between regions and the time series continuity index, identify the trajectory segments with strong continuity and concentrated disaster signals, and obtain the disaster dynamic tracking path segments.

[0012] As a further aspect of the present invention, the method further includes step S5: S5: Call the trend of changes in influencing factors and disaster level in the disaster dynamic tracking path segment, overlay the disaster recovery efficiency layer, compare the time synchronous change characteristics between influencing factors, identify the overlapping positions of segment coordinates and disaster distribution where turning points occur simultaneously, and output the disaster behavior linkage response control instruction set. The disaster behavior linkage response control instruction set includes the linkage change zone of influencing factors, the intersection point of disaster response, and the abnormal behavior indicator group.

[0013] As a further aspect of the present invention, the step of obtaining the disaster behavior linkage response control instruction set specifically includes: S511: Based on the changing trends of influencing factors and the level of disaster in the disaster dynamic tracking path segment, extract the daily change value of influencing factors and the daily average value of disaster in the segment, analyze the magnitude of changes in influencing factors and the degree of disaster fluctuation, and generate a group of synchronous influencing factor change characteristic values. S512: Call the synchronous impact factor change feature value group, and based on the disaster recovery efficiency layer grid efficiency sequence, combined with the impact factor change amplitude, disaster fluctuation degree and recovery efficiency, extract the turning point of the continuous period, filter the segment number of the turning point feature and summarize it to obtain the sudden change synchronous segment index list. S513: Based on the mutation synchronization section index list, match the disaster distribution layer coordinate units corresponding to the section number, filter overlapping grids with the same index and mark them, and output the disaster behavior linkage response control instruction set.

[0014] The resilient integration system for the multi-source heterogeneous disaster investigation and assessment model is used to execute the aforementioned resilient integration method for the multi-source heterogeneous disaster investigation and assessment model. The system includes: The disaster baseline construction module collects data from differentiated sources in the disaster assessment model, extracts disaster operation data, and formats the disaster data and influencing factors by combining time and spatial identifiers to generate a disaster operation parameter set. The disaster anomaly detection module, based on the disaster operation parameter set, filters disaster units that exceed a preset threshold according to the disaster distribution status and influencing factor data, removes interfering data associated with anomalies, integrates the remaining disaster distribution status and operation information, marks abnormal areas, and generates a disaster anomaly marking map. Based on the disaster anomaly marker map, the impact behavior extraction module extracts the disaster distribution status, impact factor change values, and disaster intensity, identifies disaster units that conform to the disaster pattern, matches disaster impact behavior templates, overlays location codes and time tags, performs spatial positioning, and generates disaster impact behavior distribution results. Based on the disaster impact behavior distribution results, the disaster path tracking module extracts time tags and disaster operation information, identifies the spatial distance between disaster behavior areas in adjacent time periods, determines the offset direction and continuous trajectory, integrates path segments with consistent offset directions, and obtains disaster dynamic tracking path segments. The behavior linkage analysis module, based on the disaster dynamic tracking path segment, compares the time series according to the changing trend of the influencing factors of the path segment, the disaster level and the recovery efficiency, identifies the synchronous turning point, locates the overlapping position of the turning area and the disaster behavior, and outputs the disaster behavior linkage response control instruction set.

[0015] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, a time-space collaborative identification mechanism is introduced during the structured recombination of multi-source heterogeneous data to achieve a unified multi-dimensional expression of disaster information, enhancing the integrity and temporal continuity of disaster data integration. By identifying abnormal intervals and factor deviations, abnormal interference information is effectively eliminated, improving the accuracy of disaster identification and the consistency of assessment. By comparing and screening the combined characteristics of influencing factors with known disaster patterns, the targeting and precise positioning capabilities of disaster identification are enhanced. Through regional offset and trajectory identification over continuous time periods, the dynamic evolution trend of disaster spread paths is obtained, clarifying the direction of disaster propagation and development potential. Recovery efficiency information is simultaneously superimposed in disaster linkage analysis to realize the correlation analysis between disaster evolution behavior and response efficiency. Finally, a response control instruction set with spatiotemporal correlation characteristics is generated, realizing integrated closed-loop control of disaster assessment, prediction, and response. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the workflow of the present invention; Figure 2 This is a flowchart illustrating the process of obtaining the disaster assessment baseline layer in this invention; Figure 3 This is a flowchart illustrating the process of obtaining the disaster anomaly diagnosis layer in this invention. Figure 4 This is a flowchart illustrating the process of obtaining the disaster impact behavior distribution map in this invention. Figure 5 This is a flowchart illustrating the process of obtaining disaster dynamic tracking path segments in this invention. Figure 6 This is a flowchart illustrating the acquisition of the disaster behavior linkage response control instruction set in this invention. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0018] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the 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, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0019] Example 1 Please see Figure 1 This invention provides a technical solution: a method for resilient integration of multi-source heterogeneous disaster investigation and assessment models, comprising the following steps: S1: Based on the initial input of multi-source heterogeneous data, the data from different sources in the disaster investigation and assessment model are restructured in a structured manner, and combined with time labels and spatial identifiers to generate a disaster assessment baseline layer; S2: Based on the disaster assessment baseline layer, analyze the correlation between disaster impact factors and disaster distribution characteristics, identify abnormal disaster fluctuation ranges and impact factor deviation ranges, screen disaster units that exceed preset thresholds, remove related interference data, and generate a disaster anomaly diagnosis layer. S3: Based on the disaster anomaly diagnosis layer, extract the disaster distribution status and the change value of influencing factors, identify key disaster nodes and combinations of influencing factors, screen disaster units that conform to known disaster patterns, and overlay location codes to mark the corresponding areas to generate a disaster impact behavior distribution map; S4: Based on the continuous change information of multiple time periods in the disaster impact behavior distribution map, identify the disaster offset of adjacent areas in time periods, extract the offset direction and propagation trajectory, filter out the area segments with the same continuous offset direction, and obtain the disaster dynamic tracking path segment. S5: Call the trend of changes in influencing factors and disaster level in the disaster dynamic tracking path segment, overlay the disaster recovery efficiency layer, compare the time synchronous change characteristics between influencing factors, identify the overlapping positions of segment coordinates and disaster distribution where the turning point occurs at the same time, and output the disaster behavior linkage response control instruction set.

[0020] The disaster assessment baseline layer includes a disaster distribution feature value set, an influencing factor attribute raster, and a spatial operation parameter surface. The disaster anomaly diagnosis layer includes disaster fluctuation identification results, influencing factor deviation markers, and interference data removal masks. The disaster impact behavior distribution map includes a target disaster candidate set, key node response templates, and regional location codes. The disaster dynamic tracking path segment includes a continuous offset trend trajectory, a path direction vector set, and regional offset nodes. The disaster behavior linkage response control instruction set includes an influencing factor linkage change area, disaster response intersection points, and abnormal behavior indicator groups.

[0021] Please see Figure 2 The specific steps for obtaining the disaster assessment baseline layer are as follows: S111: Based on the initial input of multi-source heterogeneous data, collect data from different sources in the disaster assessment model, combine time labels to perform frame sequence matching, remove disaster units in the disaster distribution sequence whose difference between adjacent frames exceeds the fluctuation critical threshold, and generate a multi-dimensional disaster matrix; Based on initial inputs of multi-source heterogeneous data, such as high-resolution satellite imagery, ground sensor networks, social media reports, and meteorological monitoring station data, data from differentiated sources are collected for disaster assessment models. For example, for flood disasters, Sentinel-1 radar satellite imagery, real-time water level data from ground water gauges, disaster-affected images from social media, and rainfall data released by meteorological bureaus are acquired. The data enters the preprocessing module through a unified interface, where frame sequence matching is performed using time stamps. Each data point is accompanied by a precise timestamp; for example, satellite imagery was acquired at 10:30:00 on July 15th, and water level data was read at 10:30:05. Data from different sources within similar time windows are associated based on timestamps to construct a chronologically ordered frame sequence; for example, a frame is generated every 5 minutes. Disaster units where the difference between adjacent frames in the disaster distribution sequence exceeds a critical fluctuation threshold are removed. First, a disaster unit is defined. The unit of measurement is the smallest unit of assessment, such as a street block. A fluctuation threshold of 20% is set. This threshold is based on statistical analysis of 200 similar flood disasters over the past five years, determining that the maximum reasonable rate of change of the inundated area within one hour is 15%, with a 20% tolerance margin. Specifically, the disaster index values ​​(such as inundated area) of the disaster unit at time t and t+Δt are calculated, and their rate of change is also calculated. For example, if the inundated area of ​​disaster unit A is 10,000 square meters at time t and 13,000 square meters at time t+Δt, the rate of change is 30%. Since 30% exceeds the 20% fluctuation threshold, the data for disaster unit A between frames t and t+Δt will be discarded. If the inundated area is 11,500 square meters at time t+Δt, the rate of change is 15%, which does not exceed the threshold, the data frame is retained, generating a multi-dimensional disaster matrix. This matrix includes time, spatial location, disaster type, disaster intensity index, and the intensity of various influencing factors. S112: Based on the multidimensional disaster matrix, extract the intensity of impact factors and disaster distribution data under the same time and space identifiers, filter out records with missing items, and generate a valid disaster operation factor data set; Based on the multidimensional disaster matrix, for example, for the disaster unit at coordinates (100.0, 30.0) at 10:30:00 on July 15th, we extract influencing factor data such as rainfall of 50 mm / hour and wind speed of 15 m / s under that time-space identifier, as well as disaster distribution data such as flood depth of 1.2 meters and affected population of 500 people. Records with missing items are filtered out. Missing items refer to cases where the disaster indicator or influencing factor data field under a specific time-space identifier in the matrix has a null value, a zero value, or an invalid identifier such as "N / A". Each extracted record is then processed... The process involves checking each field one by one. For example, if a record contains "rainfall: 50mm / h, wind speed: N / A, water depth: 1.2m", where "wind speed: N / A" is a missing field, the record is filtered out. If a record contains "rainfall: 50mm / h, wind speed: 15m / s, water depth: 1.2m", all fields are valid, and the record is retained, generating a valid disaster operation factor data set. This data set is a cleaned and filtered collection, and each record contains complete and valid time, spatial identifiers, disaster distribution data, and corresponding impact factor intensity data.

[0022] S113: Call upon the effective disaster operation factor data set, identify the standard deviation of the influencing factor intensity in disaster unit groups with consistent disaster operation modes, and normalize it using the following formula: ; Calculate the normalized disaster severity difference intensity value, reclassify disaster units within the region, and establish a disaster assessment benchmark layer; in, This represents the normalized value of the intensity of disaster differences. This represents the standard deviation of the intensity of the influencing factors in the i-th disaster unit. The average of the standard deviations of the intensity of the influencing factors within the disaster unit group. The intensity variability of the influence factor representing the i-th disaster unit. The number of units in a disaster unit group that represents a consistent disaster operation mode. This represents the sum of the original values ​​of the influence factors for the j-th disaster unit. This represents the average of the sum of the original values ​​of all influencing factors within the disaster unit group; The system retrieves a dataset of effective disaster operation factors, which provides filtered and complete disaster unit information. It identifies and normalizes the standard deviations of influencing factor intensity within disaster unit groups exhibiting consistent disaster operation patterns. First, cluster analysis is used to identify disaster unit groups with consistent disaster operation patterns. For example, disaster units within a region with similar disaster conditions and similar trends in influencing factor changes are grouped together. For instance, in a flood disaster, neighborhoods A, B, and C exhibit similar characteristics in flood propagation speed, water depth change rate, and population migration trends. If the patterns are consistent, they constitute a disaster unit group with a unified disaster operation mode. Then, the standard deviation of the intensity of the influence factors for each disaster unit in the group is calculated. For example, for disaster unit A, its influence factors (rainfall, wind speed, water level) are {50, 15, 1.2}, {60, 12, 1.5}, and {55, 14, 1.3}, respectively. The intensity of its influence factors is the sum of the factor values. For example, taking the sum of the influence factors for the day, the average intensity of the influence factors for unit A is (50+15+1.2) / 3=22.07, and its standard deviation is... The calculation is performed using the following steps: First, calculate the sum of the squared differences between the intensity of each influencing factor and its mean. For example, the sum of the influencing factor intensities for unit A is... Then, collect historical data for the unit over several consecutive days and calculate its standard deviation over time. Assume that the rainfall data for the unit A in the past 5 days is {50, 60, 55, 65, 70} mm / h, and the average value is (50+60+55+65+70) / 5=60 mm / h. Then its standard deviation ; Similarly, for disaster units B and C, assuming the calculation yields... ,These The value represents the volatility of the intensity of the influencing factors within each unit. Next, the average of the standard deviations of the intensity of the influencing factors within the disaster unit group is calculated. ,Right now Simultaneously, calculate the intensity variability of the influence factor for the i-th disaster unit. This indicates the relative fluctuation of the intensity of an influencing factor. For example, if the average intensity of an influencing factor in a disaster unit is 100 and the standard deviation is 10, then its variability is 0.1. Assuming... ,also, The number of units in a disaster unit group that represents a consistent disaster operation mode, in this example Calculate the sum of the original values ​​of the impact factors for the j-th disaster unit. For example, if the rainfall in disaster unit A on a certain day is 50mm, the wind speed is 15m / s, and the water depth is 1.2m, then the sum of its original values ​​is... Similarly, for units B and C, assume Calculate the average of the sum of the original values ​​of all influencing factors within the disaster unit group. ; Use the formula: Calculate the normalized disaster severity difference value; the numerator in this formula... The deviation of the standard deviation of the influence factor intensity of a single disaster unit from the standard deviation of the group mean reflects the difference between the internal volatility of that unit and the average level of the group. The square root term in the denominator... The relative variability of the unit itself was taken into account, while This represents the variance of the sum of the original values ​​of the influence factors of all units within a group, measuring the dispersion of the overall influence factor level of the group. By dividing the numerator and denominator and taking the absolute value, the normalization of the intensity of disaster difference is achieved, allowing comparison of disaster data of different dimensions and scales. Assuming the normalized intensity of disaster difference for disaster unit A is calculated... Substitute the above data into the formula: First, calculate the summation term in the denominator: ; Then calculate the square root term in the denominator: ; Final calculation : ; Similarly, it can be calculated , The innovation of this formula lies in comprehensively considering the volatility of the unit itself. ), within-group average ( ) and the overall dispersion within the group ( This method can more comprehensively and accurately quantify the intensity of differences among individual disaster units within the disaster operation mode, thereby effectively distinguishing between normal fluctuations and abnormal changes. The results show that the normalized intensity of difference for disaster unit A is approximately 0.3108, for disaster unit B it is approximately 0.2995, and for disaster unit C it is approximately 0.0055. Based on the calculated normalized intensity of difference in disaster units within the region, the disaster units within the region are reclassified. Disaster units are divided into different levels; for example, thresholds T1=0.1 and T2=0.5 are set. At that time, it was classified as a "low-difference" disaster unit; when At that time, it was classified as a "moderately differentiated" disaster unit; when At that time, it is classified as a "high-discrepancy" disaster unit, for example, disaster unit A ( Unit B was classified as a "moderately differentiated" disaster unit. ) was classified as a "moderately differentiated" disaster unit, disaster unit C ( These were classified as "low-difference" disaster units. In this way, a disaster assessment baseline layer was established. This layer is a Geographic Information System (GIS) layer that contains the geographic location information of each disaster unit and its normalized disaster difference intensity value. And the corresponding classification results.

[0023] Please see Figure 3 The specific steps for obtaining the disaster anomaly diagnosis layer are as follows: S211: Based on the disaster assessment baseline layer, extract the disaster distribution status and influencing factor data in the layer, match and compare the fluctuation value of the disaster unit with the interval of the influencing factor, analyze the correspondence, and obtain the disaster fluctuation interval; Based on the disaster assessment baseline layer, the current disaster indicators (such as inundation depth and affected population) and related influencing factor data (such as rainfall and river flow) for each disaster unit are retrieved from the layer. The fluctuation values ​​of the disaster units are matched and compared with the intervals of the influencing factors. First, the fluctuation value of the current disaster indicator for each disaster unit compared to the historical average or baseline value is calculated. For example, if the current inundation depth of a unit is 1.8 meters, the historical average is 0.5 meters, and the fluctuation value is 1.3 meters. Multiple intervals are defined for each influencing factor; for example, rainfall is divided into "mild (0-10 mm / hour)," "moderate (10-30 mm / hour)," and "severe (30-60 mm / hour)." The system uses intervals such as "rainfall (60 mm / hour)" and "heavy rain (60 mm / hour or more)" to cross-match the fluctuation values ​​of disaster units with the current intervals of each influencing factor. For example, if the water depth fluctuation of a unit is 1.3 meters and the rainfall in its area is in the "heavy rain" interval, this matching relationship is recorded and analyzed. By statistically analyzing a large amount of historical data, the correlation between disaster fluctuation values ​​and influencing factor intervals is identified. For example, it was found that within the "heavy rain" influencing factor interval, 80% of disaster units have water depth fluctuation values ​​exceeding 1.0 meter. The disaster fluctuation interval is obtained, and based on the above analysis results, a corresponding expected range of disaster fluctuation is determined for each influencing factor interval.

[0024] S212: Call the disaster fluctuation range, combine it with the distribution status of disaster units in the assessment layer, determine whether it falls into the specified range, identify the difference of disaster units that are not in the range, determine the objects to be removed based on the difference, and obtain the list of disaster unit numbers to be removed. The system retrieves the disaster fluctuation range. For each disaster unit in the assessment layer, it obtains its current disaster fluctuation value. For example, if the current inundation depth fluctuation value of disaster unit X is 1.5 meters, based on its influence factor range (e.g., rainfall is in the "heavy rain" range), it retrieves the corresponding disaster fluctuation range (e.g., the disaster fluctuation range corresponding to "heavy rain" is 1.0 meter to 2.5 meters). It compares whether 1.5 meters is between 1.0 meter and 2.5 meters. If the fluctuation value of disaster unit X is 2.8 meters, it is determined that it does not fall within the specified range. For disaster units that are not within the range, the system identifies the difference. For disaster units whose fluctuation values ​​do not fall within the specified range, it calculates their fluctuation values ​​and compares them with the specified range. The difference in the interval boundary is used as an example. If the fluctuation value of unit X is 2.8 meters and the upper limit of the specified interval is 2.5 meters, the difference is 0.3 meters (2.8-2.5). The unit is removed based on the difference. The difference threshold is set at 0.2 meters, which means that data exceeding the normal fluctuation range by more than 0.2 meters is considered abnormal data. If the calculated difference is greater than or equal to this threshold, the disaster unit is removed. For example, if the difference is 0.3 meters, which is greater than the threshold of 0.2 meters, then unit X is removed. A list of the removed disaster unit numbers is obtained, and the unique numbers of all the disaster units that are removed are summarized into a list.

[0025] S213: Call the list of disaster unit numbers to be removed, delete the interference data associated with the corresponding numbers in the assessment layer, reorganize the spatial distribution of the remaining disaster units, and generate a disaster anomaly diagnosis layer. The system retrieves a list of disaster unit numbers to be removed. For each disaster unit number listed, it searches for the corresponding record in the disaster assessment baseline layer and logically deletes or marks it as invalid. For example, if the list contains "001, 005, 012", it removes or disables the disaster distribution status and influencing factor data associated with these three numbers. The spatial distribution of the remaining disaster units is then reorganized. After removing interfering data, the remaining valid disaster units undergo spatial data reconstruction. For example, spatial interpolation techniques are used to smooth the valid data around the removed units, or the removed units are simply removed and their proximity relationships are updated, so that all valid disaster units can re-form a continuous and logically complete spatial distribution. For example, if unit 001 is removed, its surrounding units 002, 003, and 004 are considered directly adjacent, and the original position of 001 is inferred or smoothed based on the disaster status. A disaster anomaly diagnosis layer is generated. This layer is a new GIS layer that removes all data points diagnosed as anomalies based on the disaster assessment baseline layer, enabling the layer to more accurately reflect the real, interference-free disaster distribution status and influencing factor data within the region.

[0026] Please see Figure 4 The specific steps for obtaining the disaster impact behavior distribution map are as follows: S311: Based on the disaster anomaly diagnosis layer, extract the distribution status and influencing factor change values ​​of the disaster units in the layer, combine the differentiated disaster nodes with the corresponding influencing factor data, identify typical disaster pattern areas, and obtain disaster pattern characterization combinations; Based on the disaster anomaly diagnosis layer, the geographical location, current disaster level, specific disaster indicators, and current values ​​and changes in related influencing factors compared to the previous time point or historical average are obtained for each disaster unit. For example, if the rainfall in a unit changes from 20 mm / hour to 50 mm / hour, the change is 30 mm / hour. Combined with differentiated disaster nodes and corresponding influencing factor data (differentiated disaster nodes refer to geographical locations exhibiting significant characteristics in disaster development, such as areas with the fastest flood inundation or the highest earthquake intensity), the spatial distribution density of disaster units and the gradient changes in disaster indicators are analyzed to identify... The system identifies nodes and associates their data with the corresponding time-related influencing factor data. For example, it identifies river bends as differentiated nodes for flooding, with the influencing factor data being an upstream peak flow of 800 cubic meters per second. It also identifies typical disaster pattern areas by performing pattern recognition on differentiated disaster nodes and their influencing factor data. For example, it uses spatial pattern matching technology to compare the identified node groups with pre-established typical disaster pattern templates to obtain disaster pattern representation combinations. Each identified typical disaster pattern area is represented by a set of features, including its geographical range, dominant disaster type, key influencing factor set, and corresponding range of change.

[0027] S312: Call the disaster model characterization combination, detect the fluctuation differences of disaster units and the degree of coordinated change of influencing factors in the combination area, aggregate and compare the response values ​​of disaster units at key nodes, identify the areas with potential disaster impact behavior, and obtain the disaster impact sensitive area index set; The system invokes disaster model representation combinations. For each geographical area covered by a disaster model representation combination, a detailed analysis of all disaster units within that area is conducted. First, the fluctuation range of disaster indicators (such as inundation area) for each disaster unit over a period of time is calculated. For example, if the inundation area of ​​a certain block increases from 500 square meters to 1500 square meters, the fluctuation range is 1000 square meters. Second, the degree of coordinated change among various influencing factors (such as rainfall and water level) within the area is assessed. For example, by calculating the correlation coefficient between rainfall and water level, or by monitoring whether the water level rises synchronously when rainfall increases, the response values ​​of disaster units at key nodes are aggregated and compared. Key nodes refer to geographical locations that have a decisive impact on disaster changes under a specific disaster model, such as flood control dikes and reservoirs. The flood discharge outlet summarizes the response values ​​of disaster indicators of all disaster units in the region at key nodes. For example, it counts the maximum inundation depth of all disaster units in a certain region when the flood peak passes through a key water level monitoring station, and identifies areas with potential disaster impact behavior. Based on the comprehensive analysis of fluctuation differences, degree of coordinated change and key node response values, it identifies and marks those areas where disaster units exhibit specific potential disaster impact behavior under the current influencing factors. For example, if the rainfall and water level in a certain region are positively correlated and the inundation depth of its low-lying areas fluctuates significantly, then the region is marked as a high-risk area for flood impact behavior potential, resulting in a disaster-sensitive area indicator set. This indicator set contains a series of unique identifiers for areas marked as disaster-sensitive areas.

[0028] S313: Call the disaster impact sensitive area index set, match the disaster situation units and disaster pattern templates in the sensitive areas, filter the disaster situation units that meet the feature requirements, and mark the location coding information of the corresponding areas to generate a disaster impact behavior distribution map; The system calls upon a disaster-sensitive area indicator set. For each sensitive area in the indicator set, detailed data of all disaster units within that area are extracted from the disaster anomaly diagnosis layer, including their current disaster distribution status and all influencing factor data. Then, the extracted disaster unit data is compared with predefined disaster model templates. A disaster model template is a standardized set of disaster behavior features; for example, a "landslide model" template includes features such as a specific slope range, a sharp increase in water content, and an accelerated surface displacement rate. During the comparison, the similarity between the disaster unit data and each template is calculated, and disaster units that meet the feature requirements are selected. Based on the similarity calculation results, a matching threshold is set; for example, similarity... Only disaster units with a similarity score of 0.8 or higher (0.0 indicates complete dissimilarity, 1.0 indicates complete match) are considered to meet the feature requirements. Disaster units that have a similarity score exceeding this threshold with a certain disaster pattern template are selected. For example, if disaster unit D in a sensitive area has a similarity score of 0.85 with the "landslide pattern" template, which is higher than the threshold of 0.8, then unit D is selected and its corresponding location code information is labeled. For each selected disaster unit that meets the feature requirements, one or more location codes are assigned to it. The code can be geographic coordinates, administrative division code, or specific grid unit ID. A disaster impact behavior distribution map is generated, which is a visualized GIS layer.

[0029] Please see Figure 5 The specific steps for obtaining the disaster dynamic tracking path segment are as follows: S411: Based on the continuous change information of multiple time periods in the disaster impact behavior distribution map, identify the geometric center offset of adjacent areas in the time series, extract the offset direction vector and displacement length, analyze the angle difference of the consistency of the offset direction between adjacent areas, and obtain continuous offset direction data of the disaster. Based on the continuous changes in the disaster impact behavior distribution map across multiple time periods, the map is first organized into continuous time series frames, for example, one frame per hour. For each time frame, the region representing a specific disaster event is identified, and the geometric center of that region is calculated. The geometric center positions of the same disaster event region in adjacent time frames are compared, and its offset (ΔX, ΔY) in the two-dimensional coordinate system is calculated. For example, if the center of the flood area at time t1 is (110.1, 30.5) and at time t2 it is (110.2, 30.6), then the offset is (0.1, 0.1). The offset direction vector and displacement length are extracted. Based on the calculated offset (ΔX, ΔY), it is converted into a direction vector and a scalar (displacement length). For example, for the offset (0.1, 0.1), its direction vector points northeast, and the displacement length is... For each unit distance, analyze the angle difference of the offset direction consistency between adjacent areas. For three consecutive time frames (t1, t2, t3), calculate the offset direction vector V1 from t1 to t2 and the offset direction vector V2 from t2 to t3, and then calculate the angle between these two vectors, i.e., the angle difference. For example, if the direction of V1 is 45 degrees and the direction of V2 is 50 degrees, then the angle difference is 5 degrees, and obtain the continuous offset direction data of the disaster.

[0030] S412: Call up continuous offset direction data for disaster situation, and use the formula based on the changing trend of the angle difference: ; Calculate the angular fluctuation dispersion value, extract the region segment with stable offset direction, filter the continuous segment with angular change below the preset threshold, and obtain stable offset segment information; in, Represents the angular fluctuation dispersion value. The number of consecutive angle measurement points within a given area segment. This represents the angle difference in the continuous offset direction of the disaster situation at the k-th measuring point. This represents the average angle difference between all measuring points within the represented area segment. The weight representing the variation of the angle difference at the k-th measuring point. The standard deviation of the disaster offset angle difference in the area where the kth measuring point is located; The system retrieves continuous offset data of the disaster area, which contains information on the direction and angle changes of the disaster area over time. Based on the trend of the angle difference, a formula is used: The parameters for calculating the angular fluctuation dispersion value are defined as follows: The value represents the angular fluctuation dispersion, used to quantify the degree of instability in the disaster offset direction within a specific area. This represents the number of consecutive angle measurement points within a given area. Specifically, it refers to the number of measurements taken within a continuous time period to determine the angle difference in the direction of movement of the disaster-stricken area. The value range is typically from 5 to 20, depending on the length of the time window to be analyzed and the data sampling frequency. For example, if an angle difference is calculated every hour, and a continuous 24-hour period is analyzed, then... , This represents the angle difference in the continuous offset direction of the disaster situation at the k-th measuring point. It is the angle between adjacent offset direction vectors, expressed in degrees, and ranges from 0 to 180 degrees. The average angle difference of all measuring points within the represented area segment is the average of all measurements within that time period. The arithmetic mean of the values ​​is used to measure the average trend of the change in the direction of disaster shift within that paragraph. This represents the variation weight of the angle difference at the k-th measuring point. This weight reflects the degree of abnormality or importance of the angle difference at that measuring point. For example, if a certain Much larger than the average, then its A higher value can... Defined as (in It should be a small positive number to prevent the denominator from being zero (e.g., 0.01). The larger the value, the more drastic the change in direction, and the greater the contribution of that measuring point to the overall fluctuation dispersion. The standard deviation of the disaster offset angle difference in the area where the k-th measuring point is located represents the range of fluctuation of the angle difference within the local time window where measuring point k is located. If this value is large, it means that the local directional change is unstable. The advantage of this formula is that the numerator... The degree to which each angular difference deviates from the average is measured and multiplied by its variation weight, highlighting measurement points with drastic fluctuations, while the denominator... This comprehensively considers both the overall average deviation and local fluctuations. Through this proportional relationship, the Q-value can accurately quantify the degree of dispersion of the disaster's shift direction. A high Q-value indicates extreme instability, while a low Q-value indicates stable direction. Now, let's assume a continuous region segment containing... The following are the angular difference data of the continuous offset direction of the disaster at each measuring point: ; First, calculate the average. : ; Then calculate (We set) ): ; ; ; ; ; Assuming the local standard deviation of each measuring point For (due to the requirements of the question) This is the standard deviation of the disaster offset angle difference in the area where the k-th measuring point is located. Here, it is assumed to be the standard deviation within a local sliding window. To simplify the calculation, we assume... The value is a fixed small value, such as 0.1, or it can be understood as the standard deviation of the tiny fluctuations of each measuring point (this needs to be obtained through actual data analysis). ; Now calculate the numerator. : ; ; ; ; ; Total of numerators ; Next, calculate the denominator. : ; ; ; ; ; Sum of denominators ; Finally, calculate Q: ; The results indicate that the displacement direction of the disaster area exhibits moderate fluctuations and dispersion within the analyzed segments; that is, the directional change is not entirely linear, but neither is there drastic chaos. Segments with stable displacement directions were extracted, and a preset threshold for angular change was set. This threshold is used to define the stability of the disaster displacement direction. For example, if... A value below 0.5 indicates a stable direction; if... A value between 0.5 and 2.0 indicates moderate directional fluctuation; if... A value higher than 2.0 is considered to indicate high directional volatility; the preset threshold is set here. This indicates low angle variation, meaning stable direction. A sliding window calculation is performed on the continuous offset direction data of the disaster situation, calculating within each window... Values, filter out those For consecutive time periods when the value is lower than the preset threshold of 0.8, stable offset segment information is obtained. This information includes the time periods during which the disaster area maintained a relatively stable direction of movement, as well as the start and end times of the segment.

[0031] S413: Call up the stable offset segment information, combine the offset direction angle in each segment, the intensity of disaster distribution between regions and the time series continuity index, identify the trajectory segments with strong continuity and concentrated disaster signals, and obtain the disaster dynamic tracking path segments; The system retrieves stable offset segment information. For each stable offset segment, it extracts the average offset direction angle of the disaster area within that segment. For example, the average offset direction of a segment might be due east (90 degrees). Simultaneously, it acquires the intensity information of the disaster distribution within that segment, such as the average inundated area of ​​floods and the average intensity of earthquakes. It then calculates the time series continuity index for that segment. This index quantifies the smoothness and predictability of the disaster evolution process. For example, it is evaluated by comparing the rate of change of disaster indicators in adjacent time frames. An index greater than 0.8 indicates high continuity. The system identifies trajectory segments with strong continuity and concentrated disaster signals. By comprehensively analyzing the above three parameters, it identifies trajectory segments that not only have stable movement directions but also significant disaster signals and high time series continuity. For example, if the average offset direction angle deviation of a segment is less than 5 degrees, the average inundated area remains above 5000 square meters, and the time series continuity index is greater than 0.8, then that segment is identified as a trajectory with strong continuity and concentrated disaster signals, thus obtaining the disaster dynamic tracking path segment. This segment is the core path of disaster evolution.

[0032] Please see Figure 6 The specific steps for obtaining the disaster behavior linkage response control instruction set are as follows: S511: Based on the changing trends of influencing factors and the level of disaster in the dynamic tracking path segment of disaster, extract the daily change value of influencing factors and the daily average value of disaster in the segment, analyze the magnitude of changes in influencing factors and the degree of fluctuation of disaster, and generate a group of synchronous influencing factor change characteristic values. Based on the changing trends of influencing factors and the level of disaster in the dynamic tracking path segments of the disaster, for each segment, data on various influencing factors (such as rainfall and river flow) related to that segment are collected daily, and the changes in influencing factors over 24 hours are calculated. For example, if the rainfall in a segment yesterday was 30 mm and today is 80 mm, the change is 50 mm. At the same time, the daily disaster index data for that segment are statistically analyzed, and its daily average value is calculated. For example, if the daily average inundation depth in a segment is 1.5 meters, the magnitude of changes in influencing factors and the degree of disaster fluctuation are analyzed. Cross-analysis is performed on the extracted daily changes in influencing factors and the daily average value of disaster to assess the correlation between the magnitude of changes in influencing factors and the degree of disaster fluctuation. For example, by calculating the correlation coefficient, the positive relationship between the magnitude of changes in rainfall and the daily average value of inundation depth is analyzed, and a correlation threshold is set. For example, if the correlation coefficient exceeds 0.7, it is considered that there is a significant correlation, and a set of synchronous influencing factor change characteristic values ​​is generated.

[0033] S512: Call the synchronous impact factor change feature value group, and based on the disaster recovery efficiency layer grid efficiency sequence, combined with the impact factor change magnitude, disaster fluctuation degree and recovery efficiency, extract the turning point of the continuous period, filter the segment number of the turning point feature and summarize it to obtain the sudden change synchronous segment index list. The system invokes a set of synchronous impact factor change feature values. A disaster recovery efficiency layer is pre-constructed, and the region is divided into grid cells. Each grid cell is associated with a historical recovery efficiency sequence (values ​​between 0 and 1). Combining the magnitude of impact factor changes, the degree of disaster fluctuation, and recovery efficiency, for each feature value pair in the feature value set, it is matched with the corresponding grid cell in the disaster recovery efficiency layer. This comprehensively considers the current magnitude of impact factor changes (e.g., rainfall increasing from 30mm to 80mm), the corresponding degree of disaster fluctuation (e.g., flood depth reaching 1.5m), and the historical recovery efficiency sequence of that grid cell. For example, the historical recovery efficiency of a certain grid cell is... Under similar rainfall intensities, the recovery efficiency reached 0.6 within 3 days after the disaster. Inflection points were extracted over continuous periods to identify "inflection points" that showed significant changes in the recovery efficiency sequence. An inflection point refers to the time point when the recovery efficiency changes from stable or slow change to rapid increase or decrease. For example, in the 36th hour after the disaster, the recovery efficiency value of the grid rapidly increased from 0.2 to 0.6. By calculating the first and second derivatives of the recovery efficiency, points with prominent slope change rates were identified. The segment numbers with inflection characteristics were filtered and summarized. The numbers of all the disaster dynamic tracking path segments where the identified inflection points are located were filtered and summarized to obtain a list of mutation synchronization segment indexes.

[0034] S513: Based on the list of synchronous mutation zones, match the coordinate units of the disaster distribution layer corresponding to the zone number, filter and mark overlapping grids with the same index, and output the disaster behavior linkage response control instruction set. Based on the mutation synchronization segment index list, for each segment number in the index list, all geographic coordinate units corresponding to that number are located in the disaster distribution layer. These coordinate units constitute the physical extent of the mutation synchronization segment in geographic space. Overlapping grids with consistent indices are filtered and marked. After locating the coordinate units, it is checked whether they overlap with grid units in the disaster recovery efficiency layer. Grid units that completely or partially overlap and whose recovery efficiency sequences match the mutation synchronization characteristics are selected. Overlapping grid units are marked, for example, using specific colors or symbols on the geographic information system interface. Highlighting and outputting a disaster behavior linkage response control instruction set, based on the marked overlapping grid cells and their associated disaster situation, influencing factors, and recovery efficiency information, generates a series of specific and executable disaster behavior linkage response control instructions. For example, if an overlapping grid cell shows a sudden synchronous feature of "a sharp increase in rainfall leading to a rapid rise in water depth and low historical recovery efficiency", the generated instructions include: "Activate the emergency drainage pumping station in the area, notify residents to evacuate, and dispatch rescue teams to designated shelters". The instruction set contains detailed information such as instruction type, target area, execution action, required resources, and estimated execution time.

[0035] The resilient integration system for multi-source heterogeneous disaster investigation and assessment models is used to execute the aforementioned resilient integration method for multi-source heterogeneous disaster investigation and assessment models. The system includes: The disaster baseline construction module collects data from differentiated sources in the disaster assessment model, extracts disaster operation data, and formats the disaster data and influencing factors by combining time and spatial identifiers to generate a disaster operation parameter set. The disaster anomaly detection module is based on the disaster operation parameter set. According to the disaster distribution status and influencing factor data, it filters disaster units that exceed the preset threshold, removes interference data associated with anomalies, integrates the remaining disaster distribution status and operation information, marks abnormal areas, and generates a disaster anomaly marking map. The impact behavior extraction module extracts the disaster distribution status, impact factor change values, and disaster intensity based on the disaster anomaly marker map, identifies disaster units that conform to the disaster pattern, matches disaster impact behavior templates, overlays location codes and time labels, performs spatial positioning, and generates disaster impact behavior distribution results. The disaster path tracking module extracts time tags and disaster operation information based on the distribution results of disaster impact behavior, identifies the spatial distance between disaster behavior areas in adjacent time periods, determines the direction of offset and continuous trajectory, integrates path segments with consistent offset direction, and obtains disaster dynamic tracking path segments. The behavior linkage analysis module is based on the dynamic tracking path segment of the disaster. According to the changing trend of the influencing factors, the level of disaster and the recovery efficiency of the path segment, it performs time series comparison, identifies the synchronous turning point, locates the overlapping position of the turning area and the disaster behavior, and outputs the disaster behavior linkage response control instruction set.

[0036] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A flexible integration method for multi-source heterogeneous disaster investigation and assessment models, characterized in that, Includes the following steps: S1: Based on the initial input of multi-source heterogeneous data, the data from different sources in the disaster investigation and assessment model are restructured in a structured manner, and combined with time labels and spatial identifiers to generate a disaster assessment baseline layer; S2: Based on the disaster assessment benchmark layer, analyze the correlation between disaster impact factors and disaster distribution characteristics, identify abnormal disaster fluctuation ranges and impact factor deviation ranges, screen disaster units that exceed preset thresholds, remove related interference data, and generate a disaster anomaly diagnosis layer. S3: Based on the disaster anomaly diagnosis layer, extract the disaster distribution status and the change value of influencing factors, identify key disaster nodes and combinations of influencing factors, screen disaster units that conform to known disaster patterns, and overlay location codes to mark the corresponding areas to generate a disaster impact behavior distribution map; S4: Based on the continuous change information of multiple time periods in the disaster impact behavior distribution map, identify the disaster offset of adjacent areas in time periods, extract the offset direction and propagation trajectory, filter out the area segments with the same continuous offset direction, and obtain the disaster dynamic tracking path segment.

2. The flexible integration method for multi-source heterogeneous disaster investigation and assessment models according to claim 1, characterized in that, The disaster assessment baseline layer includes a disaster distribution feature value set, an influencing factor attribute raster, and a spatial operation parameter surface. The disaster anomaly diagnosis layer includes disaster fluctuation identification results, influencing factor deviation markers, and interference data removal masks. The disaster impact behavior distribution map includes a target disaster candidate set, key node response templates, and regional location codes. The disaster dynamic tracking path segment includes a continuous offset trend trajectory, a path direction vector set, and regional offset nodes.

3. The flexible integration method for multi-source heterogeneous disaster investigation and assessment models according to claim 1, characterized in that, The specific steps for obtaining the disaster assessment baseline layer are as follows: S111: Based on the initial input of multi-source heterogeneous data, collect data from different sources in the disaster assessment model, combine time labels to perform frame sequence matching, remove disaster units in the disaster distribution sequence whose difference between adjacent frames exceeds the fluctuation critical threshold, and generate a multi-dimensional disaster matrix; S112: Based on the multidimensional disaster matrix, extract the intensity of influencing factors and disaster distribution data under the same time and space identifiers, filter out records with missing items, and generate a valid disaster operation factor data group; S113: Call the effective disaster operation factor data group, identify the standard deviation of the influence factor intensity in the disaster unit group with the same disaster operation mode and normalize it, calculate the normalized disaster difference intensity value, reclassify the disaster units in the region, and establish a disaster assessment benchmark layer.

4. The flexible integration method for multi-source heterogeneous disaster investigation and assessment models according to claim 3, characterized in that, The specific steps for obtaining the disaster anomaly diagnosis layer are as follows: S211: Based on the disaster assessment benchmark layer, extract the disaster distribution status and influencing factor data in the layer, match and compare the fluctuation value of the disaster unit with the interval of the influencing factor, analyze the corresponding relationship, and obtain the disaster fluctuation interval; S212: Call the disaster fluctuation range, combine it with the distribution status of disaster units in the assessment layer, determine whether it falls into the specified range, identify the difference of disaster units that are not in the range, determine the objects to be removed based on the difference, and obtain a list of disaster unit numbers to be removed. S213: Call the list of disaster unit numbers to be removed, delete the interference data associated with the corresponding numbers in the evaluation layer, reorganize the spatial distribution of the remaining disaster units, and generate a disaster anomaly diagnosis layer.

5. The flexible integration method for multi-source heterogeneous disaster investigation and assessment models according to claim 4, characterized in that, The specific steps for obtaining the disaster impact behavior distribution map are as follows: S311: Based on the disaster anomaly diagnosis layer, extract the distribution status and influencing factor change values ​​of the disaster units in the layer, combine the differentiated disaster nodes with the corresponding influencing factor data, identify typical disaster pattern areas, and obtain disaster pattern characterization combinations; S312: Call the disaster model characterization combination, detect the fluctuation differences of disaster units and the degree of coordinated change of influencing factors in the combination area, aggregate and compare the response values ​​of disaster units at key nodes, identify the areas with potential disaster impact behavior, and obtain the disaster impact sensitive area index set; S313: Call the disaster-affected sensitive area index set, match the disaster situation units and disaster pattern templates in the sensitive areas, filter the disaster situation units that meet the feature requirements, and mark the location coding information of the corresponding areas to generate a disaster impact behavior distribution map.

6. The flexible integration method for multi-source heterogeneous disaster investigation and assessment models according to claim 5, characterized in that, The specific steps for obtaining the disaster dynamic tracking path segment are as follows: S411: Based on the continuous change information of multiple time periods in the disaster impact behavior distribution map, identify the geometric center offset of adjacent areas in the time series, extract the offset direction vector and displacement length, analyze the angle difference of the consistency of the offset direction between adjacent areas, and obtain continuous offset direction data of the disaster situation. S412: Call the disaster continuous offset direction data, calculate the angle fluctuation dispersion value according to the changing trend of the angle difference, extract the area segment with stable offset direction, filter the continuous segment with angle change below the preset threshold, and obtain stable offset segment information. S413: Call the stable offset segment information, combine the offset direction angle in each segment, the intensity of disaster distribution between regions and the time series continuity index, identify the trajectory segments with strong continuity and concentrated disaster signals, and obtain the disaster dynamic tracking path segments.

7. The flexible integration method for multi-source heterogeneous disaster investigation and assessment models according to claim 1, characterized in that, The method also includes step S5: S5: Call the trend of changes in influencing factors and disaster level in the disaster dynamic tracking path segment, overlay the disaster recovery efficiency layer, compare the time synchronous change characteristics between influencing factors, identify the overlapping positions of segment coordinates and disaster distribution where turning points occur simultaneously, and output the disaster behavior linkage response control instruction set. The disaster behavior linkage response control instruction set includes the linkage change zone of influencing factors, the intersection point of disaster response, and the abnormal behavior indicator group.

8. The flexible integration method for multi-source heterogeneous disaster investigation and assessment models according to claim 7, characterized in that, The specific steps for obtaining the disaster behavior linkage response control instruction set are as follows: S511: Based on the changing trends of influencing factors and the level of disaster in the disaster dynamic tracking path segment, extract the daily change value of influencing factors and the daily average value of disaster in the segment, analyze the magnitude of changes in influencing factors and the degree of disaster fluctuation, and generate a group of synchronous influencing factor change characteristic values. S512: Call the synchronous impact factor change feature value group, and based on the disaster recovery efficiency layer grid efficiency sequence, combined with the impact factor change amplitude, disaster fluctuation degree and recovery efficiency, extract the turning point of the continuous period, filter the segment number of the turning point feature and summarize it to obtain the sudden change synchronous segment index list. S513: Based on the mutation synchronization section index list, match the disaster distribution layer coordinate units corresponding to the section number, filter overlapping grids with the same index and mark them, and output the disaster behavior linkage response control instruction set.

9. A flexible integrated system for multi-source heterogeneous disaster investigation and assessment models, characterized in that, The system is used to implement the flexible integration method for multi-source heterogeneous disaster investigation and assessment models according to any one of claims 1-8, and the system includes: The disaster baseline construction module collects data from differentiated sources in the disaster assessment model, extracts disaster operation data, and formats the disaster data and influencing factors by combining time and spatial identifiers to generate a disaster operation parameter set. The disaster anomaly detection module, based on the disaster operation parameter set, filters disaster units that exceed a preset threshold according to the disaster distribution status and influencing factor data, removes interfering data associated with anomalies, integrates the remaining disaster distribution status and operation information, marks abnormal areas, and generates a disaster anomaly marking map. Based on the disaster anomaly marker map, the impact behavior extraction module extracts the disaster distribution status, impact factor change values, and disaster intensity, identifies disaster units that conform to the disaster pattern, matches disaster impact behavior templates, overlays location codes and time tags, performs spatial positioning, and generates disaster impact behavior distribution results. Based on the disaster impact behavior distribution results, the disaster path tracking module extracts time tags and disaster operation information, identifies the spatial distance between disaster behavior areas in adjacent time periods, determines the offset direction and continuous trajectory, integrates path segments with consistent offset directions, and obtains disaster dynamic tracking path segments. The behavior linkage analysis module, based on the disaster dynamic tracking path segment, compares the time series according to the changing trend of the influencing factors of the path segment, the disaster level and the recovery efficiency, identifies the synchronous turning point, locates the overlapping position of the turning area and the disaster behavior, and outputs the disaster behavior linkage response control instruction set.

Images