Method and system for calculating composite flood risk based on HEV framework

Abstract

The invention discloses a method and system for calculating a composite flood risk based on an HEV framework, and the method comprises the following steps: obtaining rainfall monitoring data, water level monitoring data and terrain monitoring data, and obtaining a basic feature matrix group through a feature matrix decomposition algorithm; a risk factor recognition algorithm is adopted to obtain a standardized risk factor matrix set; adopting a data dynamic fusion algorithm to obtain a fusion data set; a multi-dimensional risk coupling analysis method is adopted to obtain a risk coupling field; adopting a risk cumulative effect quantification method to obtain a cumulative effect field; and according to the risk coupling field and the cumulative effect field, obtaining a comprehensive risk assessment report by adopting a dynamic comprehensive assessment method. Through a data processing algorithm, the method achieves the dynamic assessment of the composite flood disaster risk, and improves the accuracy and timeliness of risk assessment.

Description

Technical Field

[0001] The invention relates to flood disaster simulation technology, and in particular to a method and system for calculating composite flood risk based on a HEV framework. Background Art

[0002] The study of calculating compound flood risk based on the HEV framework is of great significance for urban disaster prevention and mitigation. Compound flood disasters have the characteristics of multi-source, coupling and chain effect, and are often caused by multiple factors such as heavy rain, failure of flood control projects, and poor drainage systems. This complexity makes it difficult for traditional risk assessment methods for single disasters to cope with it. The risk assessment method based on the HEV framework can comprehensively characterize the risk characteristics of compound flood disasters by systematically considering the three dimensions of hazard, exposure and vulnerability, and provide a scientific basis for urban flood control project planning, emergency plan formulation and risk management.

[0003] At present, the research on compound flood risk assessment mainly focuses on the following aspects: first, the spatial analysis method based on GIS is used to evaluate the spatial distribution of risks through overlay analysis; second, the weight of each assessment indicator is determined by using the hierarchical analysis method; third, the loss rate curve is established using historical disaster data; fourth, the hydrological and hydrodynamic model is used to simulate the evolution of floods. These methods have achieved quantitative assessment of compound flood risks to a certain extent, but most of them use static assessment methods, which are difficult to reflect the dynamic evolution characteristics of risks. At the same time, existing research focuses more on direct economic losses, and insufficient consideration is given to indirect losses and social impacts.

[0004] However, the existing research still has the following technical problems: First, in terms of data processing, there is a lack of dynamic fusion mechanism for multi-source heterogeneous data, especially the effective integration of real-time monitoring data such as remote sensing and the Internet of Things with historical statistical data; second, in terms of feature extraction, existing methods are difficult to accurately identify and quantify the nonlinear coupling relationship between risk factors, resulting in insufficient in-depth analysis of the risk chain transmission mechanism; third, in terms of vulnerability assessment, the adaptive and recovery capabilities of the disaster-prone body are insufficiently assessed, especially the lack of characterization of the co-evolutionary characteristics of the social-economic-engineering system; fourth, in terms of dynamic risk assessment, existing methods are difficult to effectively deal with the problem of spatiotemporal scale conversion, resulting in low accuracy and reliability of risk prediction; fifth, in terms of risk classification, there is a lack of dynamic classification methods that consider the cumulative effects of multi-level risks, making it difficult to accurately reflect the progressive evolution characteristics of risks; sixth, in terms of early warning, existing methods are not accurate enough in identifying the key nodes of risk evolution, making it difficult to achieve precision and differentiation of risk early warning. These technical problems seriously restrict the accuracy and practicality of compound flood risk assessment. Summary of the invention

[0005] The purpose of the invention is to provide a method and system for calculating composite flood risk based on the HEV framework, in order to solve the above-mentioned problems existing in the prior art.

[0006] Technical solution, a method for calculating composite flood risk based on the HEV framework, including the following steps:

[0007] Step S1, obtaining rainfall monitoring data, water level monitoring data and terrain monitoring data, using a feature matrix decomposition algorithm to obtain a basic feature matrix group; according to the basic feature matrix group, using a feature importance calculation method to obtain a feature importance vector; according to the feature importance vector and the basic feature matrix group, using an adaptive feature reconstruction algorithm to obtain a reconstructed feature field; according to the reconstructed feature field, using a multi-scale consistency verification method to obtain a dynamic feature verification report;

[0008] Step S2, obtaining the reconstructed characteristic field, historical disaster data and geographic information data, adopting a multidimensional risk factor analysis method to obtain a standardized risk factor matrix set; according to the standardized risk factor matrix set, adopting a dynamic association network generation algorithm to obtain a dynamic association network sequence; according to the dynamic association network sequence, adopting a key path extraction algorithm to obtain a key path set and stability assessment data; according to the key path set and the stability assessment data, adopting a risk cascade propagation calculation method to obtain cascade effect assessment data; according to the cascade effect assessment data, the key path set and the standardized risk factor matrix set, adopting a multi-level risk map construction method to obtain a comprehensive risk map;

[0009] Step S3, obtaining the reconstructed characteristic field, the comprehensive risk map and the pre-stored socio-economic data, adopting the adaptive hazard-bearing body classification method to obtain the classified hazard-bearing body data set; adopting the multimodal response analysis method based on the classified hazard-bearing body data set and the historical disaster response data to obtain the vulnerability response surface; adopting the dynamic adaptive capacity assessment method based on the vulnerability response surface and the socio-economic capacity data to obtain the adaptive capacity assessment matrix; adopting the evolutionary pattern recognition method based on the vulnerability response surface and the adaptive capacity assessment matrix to obtain the evolutionary pattern library; adopting the multidimensional comprehensive assessment method based on the evolutionary pattern library, the adaptive capacity assessment matrix and the classified hazard-bearing body data set to obtain the comprehensive vulnerability assessment field;

[0010] Step S4, obtain the reconstructed characteristic field, the comprehensive risk map and the comprehensive vulnerability assessment field, adopt the heterogeneous data fusion method to obtain the fused data set; based on the fused data set, adopt the multidimensional risk coupling analysis method to obtain the risk coupling field; based on the risk coupling field and the historical risk evolution data, adopt the risk transformation mechanism identification method to obtain the risk transformation pattern set; based on the risk coupling field and the risk transformation pattern set, adopt the risk cumulative effect quantification method to obtain the cumulative effect field; based on the risk coupling field, the cumulative effect field and the risk transformation pattern set, adopt the dynamic comprehensive assessment method to obtain the comprehensive risk assessment report.

[0011] According to another aspect of the present application, a system for calculating composite flood risk based on a HEV framework is also provided, comprising:

[0012] at least one processor; and,

[0013] a memory communicatively connected to at least one of the processors; wherein,

[0014] The memory stores instructions that can be executed by the processor, and the instructions are used to be executed by the processor to implement the method for calculating composite flood risk based on the HEV framework described in any of the above technical solutions.

[0015] Beneficial effect: Through data processing algorithms, dynamic assessment of compound flood disaster risks is achieved, which improves the accuracy and timeliness of risk assessment. The technical effect will be explained in conjunction with specific implementation cases in the future. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a flow chart of the present invention.

[0017] Figure 2 It is a flow chart of step S1 of the present invention.

[0018] Figure 3 It is a flow chart of step S2 of the present invention.

[0019] Figure 4 It is a flow chart of step S3 of the present invention.

[0020] Figure 5 It is a flow chart of step S4 of the present invention. DETAILED DESCRIPTION

[0021] like Figure 1 As shown, a method for calculating composite flood risk based on the HEV framework includes the following steps:

[0022] Step S1, obtain rainfall monitoring data, water level monitoring data and terrain monitoring data, use feature matrix decomposition algorithm to obtain basic feature matrix group; according to the basic feature matrix group, use feature importance calculation method to obtain feature importance vector; according to feature importance vector and basic feature matrix group, use adaptive feature reconstruction algorithm to obtain reconstructed feature field; according to the reconstructed feature field, use multi-scale consistency verification method to obtain dynamic feature verification report.

[0023] Step S2, obtain the reconstructed characteristic field, historical disaster data and geographic information data output by step S1, adopt a multidimensional risk factor analysis method to obtain a standardized risk factor matrix set; based on the standardized risk factor matrix set, adopt a dynamic association network generation algorithm to obtain a dynamic association network sequence; based on the dynamic association network sequence, adopt a critical path extraction algorithm to obtain a critical path set and stability assessment data; based on the critical path set and stability assessment data, adopt a risk cascade propagation calculation method to obtain cascade effect assessment data; based on the cascade effect assessment data, the critical path set and the standardized risk factor matrix set, adopt a multi-level risk map construction method to obtain a comprehensive risk map.

[0024] Step S3, obtaining the reconstructed characteristic field output by step S1, the comprehensive risk map output by step S2 and the pre-stored socio-economic data, adopting an adaptive hazard-prone body classification method to obtain a classified hazard-prone body data set; adopting a multimodal response analysis method based on the classified hazard-prone body data set and historical disaster response data to obtain a vulnerability response surface; adopting a dynamic adaptive capacity assessment method based on the vulnerability response surface and socio-economic capacity data to obtain an adaptive capacity assessment matrix; adopting an evolutionary pattern recognition method based on the vulnerability response surface and the adaptive capacity assessment matrix to obtain an evolutionary pattern library; adopting a multidimensional comprehensive assessment method based on the evolutionary pattern library, the adaptive capacity assessment matrix and the classified hazard-prone body data set to obtain a comprehensive vulnerability assessment field.

[0025] Step S4, obtain the reconstructed feature field output by step S1, the comprehensive risk map output by step S2 and the comprehensive vulnerability assessment field output by step S3, adopt the heterogeneous data fusion method to obtain the fused data set; based on the fused data set, adopt the multidimensional risk coupling analysis method to obtain the risk coupling field; based on the risk coupling field and historical risk evolution data, adopt the risk transformation mechanism identification method to obtain the risk transformation pattern set; based on the risk coupling field and the risk transformation pattern set, adopt the risk cumulative effect quantification method to obtain the cumulative effect field; based on the risk coupling field, the cumulative effect field and the risk transformation pattern set, adopt the dynamic comprehensive assessment method to obtain the comprehensive risk assessment report.

[0026] This embodiment constructs a complete composite flood risk assessment technology system by integrating four key links: feature extraction, risk analysis, vulnerability assessment and comprehensive assessment. First, through the feature matrix decomposition and reconstruction algorithm, efficient processing of multi-source monitoring data is achieved; second, through dynamic correlation network and key path analysis, the law of risk propagation is accurately grasped; third, through multimodal response analysis and evolutionary pattern recognition, accurate assessment of the vulnerability of the disaster-bearing body is achieved; finally, through risk coupling analysis and cumulative effect quantification, the dynamic comprehensive assessment of composite risks is completed. This implementation not only overcomes the technical difficulties in data processing, risk propagation, vulnerability assessment and comprehensive assessment in traditional methods, but also realizes the full process intelligence of composite flood risk assessment through the organic connection of various links. Especially in the characterization of risk dynamic characteristics, this embodiment can accurately grasp the law of risk evolution, provide scientific decision-making support for urban flood control and disaster reduction, and improve the accuracy, timeliness and reliability of composite flood risk assessment.

[0027] like Figure 2 As shown, according to one aspect of the present application, step S1 is specifically:

[0028] Step S11, obtain the rainfall monitoring data, water level monitoring data and terrain monitoring data of the monitoring site, and construct the rainfall spatial distribution matrix, water level spatial distribution matrix and terrain feature matrix according to the spatial location information; according to the rainfall spatial distribution matrix, water level spatial distribution matrix and terrain feature matrix, obtain a set of standardized spatial distribution matrices through linear transformation method.

[0029] Step S12, obtain the standardized spatial distribution matrix set in step S11, calculate the correlation coefficient between spatial positions, and obtain the spatial correlation matrix; calculate the change characteristics of the time series according to the time series data in the standardized spatial distribution matrix set, and obtain the time evolution matrix; calculate the degree of coupling between variables according to the different monitoring variable data in the standardized spatial distribution matrix set, and obtain the variable coupling matrix; according to the spatial correlation matrix, the time evolution matrix and the variable coupling matrix, use multi-level decomposition operation to obtain the basic feature matrix group.

[0030] Step S13, obtain the basic feature matrix group in step S12, calculate the change amplitude of each feature in the time dimension, and obtain the time-varying weight vector; calculate the spatial influence coefficient according to the spatial distribution range of the basic feature matrix group, and obtain the spatial influence matrix; according to the time-varying weight vector and the spatial influence matrix, use weighted fusion operation to obtain the feature importance vector.

[0031] Step S14, obtain the basic feature matrix group in step S12 and the feature importance vector in step S13, construct a nonlinear weight function, calculate the dynamic weight coefficient of each feature, and obtain a dynamic weight matrix; according to the dynamic weight matrix, perform weighted combination on the basic feature matrix group to obtain an initial reconstructed feature field; obtain continuity constraints in time and space, optimize and adjust the initial reconstructed feature field, and obtain a reconstructed feature field.

[0032] Step S15, obtain the reconstructed feature field in step S14 and the standardized spatial distribution matrix set in step S11, calculate the feature consistency index of the local area, and obtain the local consistency matrix; calculate the feature consistency index of the global range according to the reconstructed feature field, and obtain the global consistency vector; construct a dynamic threshold judgment criterion according to the local consistency matrix and the global consistency vector, and generate a dynamic feature verification report.

[0033] This embodiment realizes efficient integration and feature extraction of rainfall, water level and terrain monitoring data through dynamic fusion of multi-source data and feature matrix decomposition algorithm. Specifically, the feature matrix decomposition algorithm is used to process multi-source monitoring data, which can effectively reduce the data dimension and extract key feature information; the feature importance calculation method can quantitatively evaluate the contribution of different features and realize feature optimization; the adaptive feature reconstruction algorithm can automatically adjust the reconstruction weight according to the feature importance to improve the accuracy of the feature field; finally, the multi-scale consistency test is used to ensure the reliability of the reconstructed feature field. This embodiment not only solves the problem that multi-source heterogeneous data is difficult to effectively fuse in traditional methods, but also ensures the accuracy and reliability of feature extraction through feature reconstruction and verification mechanism. In practical applications, this embodiment can effectively extract key information such as rainfall intensity, water level changes, terrain characteristics, etc., provide a reliable data basis for subsequent risk assessment, and improve the accuracy of composite flood risk assessment.

[0034] like Figure 3 As shown, according to one aspect of the present application, step S2 is specifically:

[0035] Step S21, obtain the reconstructed characteristic field output by step S1, the pre-stored historical disaster data and geographic information data, extract the hydrometeorological parameters, and obtain the hydrometeorological factor matrix; extract the topographic and geomorphic characteristics based on the geographic information data, and obtain the topographic and geomorphic factor matrix; obtain the land use and flood control project data, extract the human activity characteristics, and obtain the human activity factor matrix; according to the hydrometeorological factor matrix, the topographic and geomorphic factor matrix and the human activity factor matrix, adopt the standardized processing method to obtain the standardized risk factor matrix set.

[0036] Step S22, obtain the standardized risk factor matrix set output by step S21, calculate the correlation strength between the risk factors at different times, and obtain a time-varying correlation strength matrix; based on the time-varying correlation strength matrix, construct a dynamic threshold discriminant function, screen significant correlation relationships, and obtain a significant correlation matrix; based on the significant correlation matrix, construct a time-varying network structure to obtain a dynamic correlation network sequence.

[0037] Step S23, obtain the dynamic association network sequence output by step S22, calculate the importance of each network node, and obtain the node importance vector; identify the main path of risk propagation based on the node importance vector and the dynamic association network sequence, and obtain the critical path set; based on the critical path set, evaluate the time stability of the path to obtain stability evaluation data.

[0038] Step S24, obtain the critical path set and stability assessment data output from step S23, calculate the risk propagation intensity on the critical path, and obtain the propagation intensity matrix; calculate the delayed effect of risk propagation based on the propagation intensity matrix to obtain the time lag effect matrix; calculate the cumulative impact of the risk based on the propagation intensity matrix and the time lag effect matrix to obtain the cascade effect assessment data.

[0039] Step S25, obtain the cascade effect assessment data output by step S24, the critical path set output by step S23 and the standardized risk factor matrix set output by step S21, calculate the basic risk distribution, and obtain the basic risk layer; calculate the associated risk distribution according to the basic risk layer and the critical path set, and obtain the associated risk layer; calculate the cascade risk distribution according to the associated risk layer and the cascade effect assessment data, and obtain the cascade risk layer; according to the basic risk layer, the associated risk layer and the cascade risk layer, adopt the weighted superposition method to obtain the comprehensive risk map.

[0040] This embodiment achieves a comprehensive quantitative assessment of compound flood risks through multidimensional risk factor analysis and dynamic association network generation algorithm. First, based on the reconstruction of characteristic fields, historical disaster data and geographic information data, a multidimensional risk factor analysis method is used to extract standardized risk factors; then, the dynamic association network algorithm is used to capture the time-varying correlation between risk factors; the key path extraction algorithm is used to identify the main path of risk propagation; finally, based on the cascade effect evaluation and multi-level risk map construction method, the visual expression of risks is achieved. This embodiment overcomes the defect of traditional risk assessment methods that it is difficult to depict the dynamic correlation between risk factors, and accurately grasps the propagation law of compound risks through dynamic association network and key path analysis. In particular, in terms of the quantification of risk cascade effects, it can accurately evaluate the cumulative effect and propagation characteristics of risks, providing a scientific basis for risk prevention and control. This embodiment can effectively identify risk hotspots and key propagation paths, providing important support for disaster prevention and mitigation decisions.

[0041] like Figure 4 As shown, according to one aspect of the present application, step S3 is specifically:

[0042] Step S31, obtain the reconstructed characteristic field output by step S1, the comprehensive risk map output by step S2 and the pre-stored socio-economic data, extract the spatial distribution characteristics of the hazard-prone body, and obtain the hazard-prone body characteristic matrix; construct dynamic classification rules based on time series data to obtain dynamic classification criteria; classify the hazard-prone body according to the hazard-prone body characteristic matrix and the dynamic classification criteria to obtain a classified hazard-prone body data set.

[0043] Step S32, obtaining the classified disaster-prone body data set output by step S31 and the pre-stored historical disaster response data, extracting the response characteristics of different types of disaster-prone bodies, and obtaining a response characteristic matrix; calculating the sensitivity of the disaster-prone body to disasters based on the response characteristic matrix, and obtaining a time-varying sensitivity matrix; evaluating the recovery capacity of the disaster-prone body based on the response characteristic matrix and the time-varying sensitivity matrix, and obtaining a recovery capacity matrix; constructing a response surface model based on the response characteristic matrix, the time-varying sensitivity matrix, and the recovery capacity matrix, and obtaining a vulnerability response surface.

[0044] Step S33, obtain the vulnerability response surface output by step S32 and the pre-stored socio-economic capacity data, construct an adaptability assessment index system, and obtain an adaptability index matrix; calculate the resource allocation efficiency based on the socio-economic capacity data to obtain a resource efficiency matrix; evaluate the system resilience based on the adaptability index matrix and the resource efficiency matrix to obtain a system resilience matrix; adopt a comprehensive assessment method based on the adaptability index matrix, the resource efficiency matrix and the system resilience matrix to obtain an adaptability assessment matrix.

[0045] Step S34, obtain the vulnerability response surface output by step S32 and the adaptability assessment matrix output by step S33, extract the spatiotemporal variation characteristics of vulnerability, and obtain an evolutionary feature vector set; based on the evolutionary feature vector set, identify typical evolutionary patterns and obtain key evolutionary patterns; based on the key evolutionary patterns, predict the vulnerability change trend and obtain evolutionary trend data; based on the key evolutionary patterns and evolutionary trend data, construct an evolutionary pattern database and obtain an evolutionary pattern library.

[0046] Step S35, obtain the evolution pattern library output by step S34, the adaptability assessment matrix output by step S33 and the classified hazard-prone body data set output by step S31, calculate the static vulnerability distribution, and obtain the static vulnerability layer; calculate the dynamic vulnerability distribution according to the evolution pattern library, and obtain the dynamic vulnerability layer; calculate the adaptive vulnerability distribution according to the adaptability assessment matrix, and obtain the adaptive vulnerability layer; according to the static vulnerability layer, the dynamic vulnerability layer and the adaptive vulnerability layer, adopt a multidimensional integration method to obtain a comprehensive vulnerability assessment field.

[0047] This embodiment achieves accurate assessment of the vulnerability of the disaster-prone body through adaptive disaster-prone body classification and multimodal response analysis methods. First, the disaster-prone body is classified by an adaptive classification method, and then a vulnerability response surface is established based on historical disaster response data, and the recovery capacity of the disaster-prone body is quantified by a dynamic adaptive capacity assessment method. Finally, a comprehensive vulnerability assessment field is obtained through evolutionary pattern recognition and multidimensional comprehensive assessment methods. This embodiment breaks through the limitation that it is difficult to characterize the dynamic response characteristics of the disaster-prone body in traditional vulnerability assessment methods. Through multimodal response analysis and dynamic adaptive capacity assessment, the response law of the disaster-prone body in different disaster scenarios is accurately grasped. In particular, in terms of evolutionary pattern recognition, it can predict the evolution trend of the vulnerability of the disaster-prone body and provide forward-looking guidance for risk prevention and control. This embodiment can accurately assess the vulnerability characteristics of different types of disaster-prone bodies and provide a scientific basis for the formulation of differentiated protection measures.

[0048] like Figure 5 As shown, according to one aspect of the present application, step S4 is specifically:

[0049] Step S41, obtain the reconstructed feature field output by step S1, the comprehensive risk map output by step S2 and the comprehensive vulnerability assessment field output by step S3, calculate the correlation characteristics between the data, and obtain the data correlation matrix; according to the spatiotemporal distribution characteristics, construct the consistency test standard to obtain the spatiotemporal consistency test matrix; according to the data correlation matrix and the spatiotemporal consistency test matrix, fuse the multi-source data to obtain the fused data set.

[0050] Step S42, obtain the fused data set output by step S41, calculate the degree of coupling between different risk types, and obtain the risk coupling intensity matrix; calculate the risk amplification effect based on the risk coupling intensity matrix to obtain the risk amplification coefficient matrix; calculate the risk suppression effect based on the risk coupling intensity matrix to obtain the risk suppression coefficient matrix; based on the risk coupling intensity matrix, the risk amplification coefficient matrix and the risk suppression coefficient matrix, adopt the coupling effect integration method to obtain the risk coupling field.

[0051] Step S43, obtain the risk coupling field output by step S42 and the pre-stored historical risk evolution data, extract the risk conversion characteristics, and obtain the conversion characteristic matrix; identify the key conversion nodes according to the conversion characteristic matrix, and obtain the key node data; calculate the conversion probability according to the conversion characteristic matrix and the key node data, and obtain the conversion probability matrix; construct a conversion pattern database according to the conversion characteristic matrix, the key node data and the conversion probability matrix, and obtain the risk conversion pattern set.

[0052] Step S44, obtain the risk coupling field output by step S42 and the risk conversion pattern set output by step S43, calculate the risk cumulative intensity, and obtain the cumulative intensity matrix; evaluate the time lag impact according to the cumulative intensity matrix to obtain the time lag impact matrix; calculate the long-term effect according to the cumulative intensity matrix and the time lag impact matrix to obtain the long-term effect matrix; and obtain the cumulative effect field by using the effect superposition method according to the cumulative intensity matrix, the time lag impact matrix and the long-term effect matrix.

[0053] Step S45, obtain the risk coupling field output by step S42, the cumulative effect field output by step S44 and the risk transformation pattern set output by step S43, calculate the immediate risk distribution, and obtain the immediate risk layer; calculate the cumulative risk distribution based on the cumulative effect field to obtain the cumulative risk layer; calculate the evolutionary risk distribution based on the risk transformation pattern set to obtain the evolutionary risk layer; construct a comprehensive risk index based on the immediate risk layer, the cumulative risk layer and the evolutionary risk layer to obtain a comprehensive risk assessment report.

[0054] This embodiment realizes a dynamic comprehensive assessment of compound flood risks through heterogeneous data fusion and multi-dimensional risk coupling analysis methods. First, the heterogeneous data fusion method is used to integrate and reconstruct the characteristic field, the comprehensive risk map and the comprehensive vulnerability assessment field. Then, the multi-dimensional risk coupling analysis method is used to identify the interaction relationship between risk factors, and based on the risk transformation mechanism identification and cumulative effect quantification method, a dynamic comprehensive assessment report is finally obtained. This embodiment overcomes the problem that it is difficult to deal with the multi-dimensional risk coupling effect in traditional risk assessment methods. Through risk coupling analysis and cumulative effect quantification, the comprehensive impact of compound risks is accurately assessed. In particular, in terms of risk transformation mechanism identification, it can reveal the inherent laws of risk evolution and provide a scientific basis for risk warning. This embodiment can comprehensively assess the dynamic characteristics of compound flood risks and provide reliable support for urban flood control and disaster reduction decisions.

[0055] According to one aspect of the present application, step S11 is specifically:

[0056] Step S111, obtain the original rainfall monitoring data, original water level monitoring data and original terrain monitoring data of the monitoring site, adopt a multi-source data quality evaluation method (mainly including basic methods such as data integrity test, consistency test and outlier detection), calculate the data reliability index, and obtain a data quality evaluation matrix; according to the data quality evaluation matrix, screen valid data points to obtain a valid monitoring data set.

[0057] Step S112, obtain the valid monitoring data set outputted in step S111, use the spatiotemporal outlier intelligent recognition algorithm to identify abnormal data points, and obtain an outlier labeling matrix; according to the outlier labeling matrix, correct the abnormal data to obtain a corrected monitoring data set; according to the corrected monitoring data set, use the data gap filling intelligent algorithm to fill in the missing data to obtain a complete monitoring data set. Specifically, the spatiotemporal outlier intelligent recognition algorithm is as follows:

[0058] Construct the spatiotemporal correlation matrix ST(i,j) to represent the spatiotemporal correlation between monitoring points i and j; calculate the local anomaly index: LAI(x) = Σ[w(i)*|x(i) - μ(i)| / σ(i)], where w(i) is the spatiotemporal weight; design the dynamic threshold function: θ(t) = βexp(-αt) + γ, where α, β, and γ are adaptive parameters; when LAI(x) > θ(t), it is judged as an outlier; where x is the data point to be detected; μ(i) is the local mean of the i-th monitoring point; σ(i) is the local standard deviation of the i-th monitoring point; θ(t) is the dynamic threshold function, and t is the time variable.

[0059] The data filling intelligent algorithm has the following process: construct a multi-layer spatiotemporal attention network MSTA; the time attention layer At(i,j) = softmax(Qt*Kt^T); the spatial attention layer As(i,j) = softmax(Qs*Ks^T); output the missing value y = MSTA(xt-k:t-1, xs), where xt is the time series, xs is the spatial feature, Qs and Ks are the spatial query matrix and the spatial key value matrix respectively; T is the transpose symbol, xt-k:t-1 represents the historical time series from tk to t-1, and k is the backtracking time step.

[0060] Step S113, obtain the complete monitoring data set output by step S112, adopt the flood numerical simulation method to simulate the levee breach process, and obtain the breach process data; simulate the breach water flow evolution process according to the breach process data, and obtain the water flow evolution data; calculate the waterlogging process according to the water flow evolution data, and obtain the water accumulation process data; adopt the simulation parameter adaptive optimization method according to the breach process data, the water flow evolution data and the water accumulation process data, and obtain the optimized simulation result.

[0061] Step S114, obtain the optimization simulation results output by step S113, extract the maximum flooding water depth data, and obtain the water depth distribution matrix; according to the optimization simulation results, extract the maximum flow velocity data to obtain the flow velocity distribution matrix; according to the optimization simulation results, extract the flooding duration data to obtain the flooding duration matrix; according to the water depth distribution matrix, the flow velocity distribution matrix and the flooding duration matrix, construct a spatial distribution data set to obtain a spatial distribution matrix set.

[0062] Step S115, obtaining the spatial distribution matrix set output by step S114, using a spatial interpolation optimization algorithm (including classical spatial interpolation methods such as Kriging interpolation and inverse distance weighted method) to perform spatial processing on discrete data to obtain a continuously distributed data set; based on the continuously distributed data set, using a scale conversion algorithm (upsampling, downsampling) to unify the spatial resolution to obtain a unified scale data set; based on the unified scale data set, using a standardized processing method to obtain a standardized spatial distribution matrix set.

[0063] This embodiment achieves high-quality preprocessing and spatialization of monitoring data through intelligent data processing and simulation analysis methods. Specifically, first, a multi-source data quality evaluation method is used to evaluate the reliability of the original monitoring data and screen out valid data points; then, the spatiotemporal outlier intelligent identification algorithm is used to correct the abnormal data, and the data gap filling intelligent algorithm is used to fill the missing values; then, the flood numerical simulation method is used to simulate the levee breach and water flow evolution process to obtain accurate hydrodynamic parameters; finally, the standardized spatial distribution of data is achieved through spatial interpolation optimization and scale conversion algorithms. This embodiment overcomes the problems of inaccurate outlier identification, poor missing value filling effect, and low simulation accuracy of hydrodynamic processes in traditional data processing methods. In particular, in terms of data spatialization processing, the reliability of data and the accuracy of spatial expression are improved through multi-level data quality control and optimization processing. This embodiment can provide high-quality basic data support for subsequent risk assessment, effectively improving the reliability of the entire assessment process.

[0064] According to one aspect of the present application, step S12 is specifically:

[0065] Step S121, obtain the standardized spatial distribution matrix set output by step S11, use a dynamic feature decomposition algorithm to extract time series change features, and obtain a time series feature matrix; based on the standardized spatial distribution matrix set, use a spatial feature recognition algorithm to extract spatial distribution features and obtain a spatial feature matrix; based on the standardized spatial distribution matrix set, use a multivariate feature extraction algorithm to identify variable features and obtain a variable feature matrix.

[0066] Among them, the dynamic feature decomposition algorithm F(t,s,v) = DFD(X); where F is the decomposed feature tensor; t, s, v are time, space, and variable dimensions; X is the original data matrix; DFD is the dynamic feature decomposition function.

[0067] Step S122, obtain the time series feature matrix, spatial feature matrix and variable feature matrix output by step S121, use the adaptive correlation analysis method to calculate the time dimension correlation coefficient to obtain the time correlation matrix; calculate the spatial position correlation coefficient according to the spatial feature matrix to obtain the spatial correlation matrix; calculate the correlation coefficient between variables according to the variable feature matrix to obtain the variable correlation matrix.

[0068] Step S123, obtain the time correlation matrix, space correlation matrix and variable correlation matrix output by step S122, use the dynamic weight calculation method to determine the time dimension weight, and obtain the time weight vector; calculate the spatial position weight according to the spatial correlation matrix to obtain the spatial weight vector; calculate the variable weight according to the variable correlation matrix to obtain the variable weight vector.

[0069] Step S124, obtain the time weight vector, space weight vector and variable weight vector outputted in step S123, and use a multi-dimensional coupling degree calculation method to calculate the time-space coupling strength to obtain a time-space coupling matrix; calculate the time-varying coupling strength according to the time weight vector and the variable weight vector to obtain a time-varying coupling matrix; calculate the space-varying coupling strength according to the space weight vector and the variable weight vector to obtain a space-varying coupling matrix.

[0070] The calculation method of multidimensional coupling degree is: C(i,j) = [λtCt(i,j) + λsCs(i,j) + λv*Cv(i,j)] / Σλ; C(i,j) is the comprehensive coupling degree; Ct, Cs, Cv are the coupling degrees of time, space and variable dimensions; λt, λs, λv are the weight coefficients of each dimension; i, j are the characteristic indicator numbers.

[0071] Step S125, obtain the spatiotemporal coupling matrix, time-varying coupling matrix and space-varying coupling matrix output by step S124, use a multi-level feature decomposition algorithm to extract the dominant feature pattern and obtain the dominant feature matrix; based on the dominant feature matrix, identify the secondary feature pattern and obtain the secondary feature matrix; based on the dominant feature matrix and the secondary feature matrix, construct a feature hierarchy to obtain a basic feature matrix group.

[0072] This embodiment realizes the accurate extraction of basic features and the intelligent construction of feature matrix through multi-dimensional feature analysis and dynamic weight optimization method. First, the dynamic feature decomposition algorithm is used to extract features from three dimensions: time series, space and variable; then the correlation coefficient of each dimension is calculated through the adaptive correlation analysis method; then the dynamic weight calculation method is used to determine the weight coefficient of each dimension; then the coupling relationship between features is evaluated through the multi-dimensional coupling degree calculation method; finally, a hierarchical feature matrix system is constructed using a multi-level feature decomposition algorithm. This embodiment solves the problem that it is difficult to handle the dynamic association of multi-dimensional features in traditional feature extraction methods. Through adaptive weight optimization and multi-level decomposition, the hierarchical structure and dynamic characteristics of features are accurately characterized. Especially in the quantification of feature coupling relationships, it can accurately identify dominant features and secondary features, providing reliable feature support for risk assessment. This embodiment improves the accuracy of feature extraction and the rationality of feature expression.

[0073] According to one aspect of the present application, step S15 is specifically:

[0074] Step S151, obtain the reconstructed feature field output by step S14 and the set of standardized spatial distribution matrices output by step S11, use a multi-scale decomposition algorithm to scale decompose the reconstructed feature field to obtain a multi-scale feature sequence; calculate the feature distribution at each scale based on the multi-scale feature sequence to obtain a scale distribution matrix; calculate the correspondence between each scale based on the multi-scale feature sequence and the set of standardized spatial distribution matrices to obtain a scale correspondence matrix.

[0075] Step S152, obtain the scale distribution matrix and scale correspondence matrix output by step S151, use the local feature consistency evaluation method to calculate the feature difference of the local area, and obtain the local difference matrix; according to the local difference matrix, identify the feature mutation area and obtain the mutation area identifier; according to the local difference matrix and the mutation area identifier, evaluate the local consistency level and obtain the local consistency index.

[0076] Local feature consistency evaluation method LCI(x) = Σ[w(i)*exp(-d(x,i) / σ)*S(x,i)]; where LCI is the local consistency index; w(i) is the spatial weight coefficient; d(x,i) is the spatial distance function; σ is the spatial scale parameter, and S(x,i) is the feature similarity function.

[0077] Step S153, obtain the local consistency index output by step S152 and the multi-scale feature sequence output by step S151, use the index forward algorithm to transform the inverse index to obtain a forward index set; according to the forward index set, construct the index polarity discrimination criterion to obtain the polarity discrimination matrix; according to the forward index set and the polarity discrimination matrix, optimize the index direction to obtain the optimized index set.

[0078] Step S154, obtain the optimization indicator set output by step S153, use an adaptive normalization algorithm to calculate the indicator variation range, and obtain an indicator range matrix; determine the normalization parameter according to the indicator range matrix to obtain a normalized parameter set; perform indicator normalization operation according to the optimization indicator set and the normalized parameter set to obtain a normalized indicator matrix.

[0079] Step S155, obtain the normalized indicator matrix output by step S154 and the local consistency index output by step S152, use the dynamic threshold construction algorithm to calculate the verification thresholds of each level, and obtain a verification threshold sequence; according to the normalized indicator matrix and the verification threshold sequence, perform multi-level verification to obtain a verification result matrix; according to the verification result matrix and the local consistency index, generate a feature verification report to obtain a dynamic feature verification report.

[0080] Dynamic threshold construction algorithm θ(t,s) = α(t)*β(s) + γ; where θ is the dynamic threshold function; t,s are time and space variables; α(t) is the time adjustment function; β(s) is the space adjustment function; γ is the benchmark threshold.

[0081] This embodiment realizes the reliability assessment of feature reconstruction results through multi-scale feature verification and dynamic threshold construction methods. First, a multi-scale decomposition algorithm is used to decompose the scale of the reconstructed feature field; then, a local feature consistency evaluation method is used to identify the feature mutation area; then, an indicator positive and adaptive normalization algorithm is used to optimize the evaluation indicator system; finally, a dynamic threshold construction algorithm is used to realize intelligent feature verification. This embodiment overcomes the problem that traditional feature verification methods are difficult to handle multi-scale features and dynamic threshold determination, and improves the accuracy of feature verification through local consistency evaluation and dynamic threshold optimization. Especially in the identification of feature mutation areas, it can accurately capture abnormal changes in features and provide important support for risk warning. This embodiment effectively ensures the reliability of feature reconstruction results and provides quality assurance for subsequent risk assessment.

[0082] According to one aspect of the present application, step S21 is specifically:

[0083] Step S211, obtain the reconstructed characteristic field output by step S1, the pre-stored historical disaster data and geographic information data, use a multi-dimensional risk factor identification algorithm to extract the hazard index set, and obtain the hazard index matrix; extract the exposure index set based on the historical disaster data and geographic information data to obtain the exposure index matrix; extract the vulnerability index set based on the historical disaster data and geographic information data to obtain the vulnerability index matrix.

[0084] Multidimensional risk factor identification algorithm H = f(R, P, V); where H is the comprehensive risk index; R is the hazard index; P is the exposure index; V is the vulnerability index; and f is the comprehensive risk assessment function.

[0085] Step S212, obtain the hazard index matrix, exposure index matrix and vulnerability index matrix output by step S211, use the index correlation analysis method to calculate the correlation between the indicators, and obtain the index correlation matrix; according to the index correlation matrix, identify the redundant indicators and obtain the redundant indicator set; according to the hazard index matrix, exposure index matrix and vulnerability index matrix, eliminate the indicators in the redundant indicator set and obtain the preliminary index system.

[0086] Step S213, obtain the preliminary indicator system output by step S212, use the indicator sensitivity analysis method to calculate the influence of each indicator on the result, and obtain the sensitivity matrix; according to the sensitivity matrix, determine the key indicators and obtain the key indicator set; according to the key indicator set, construct a hierarchical analysis structure and obtain the indicator hierarchy system.

[0087] Indicator sensitivity analysis method SI(i) = dY / dXi * Xi / Y + Σ(d²Y / dXidXj * Xi*Xj / Y); where SI is the sensitivity index; Y is the risk assessment result; Xi,Xj are risk indicators; d is the partial derivative operator.

[0088] Step S214, obtain the indicator hierarchy system output by step S213, use the expert scoring method to collect expert scoring data, and obtain the expert scoring matrix; calculate the consistency ratio based on the expert scoring matrix to obtain the consistency test result; use the improved hierarchical analysis method based on the expert scoring matrix and the consistency test result to obtain the initial weight vector.

[0089] Analytic Hierarchy Process (AHP), W = AHP(A); where W is the weight vector; A is the judgment matrix.

[0090] Step S215, obtain the initial weight vector output by step S214 and the sensitivity matrix output by step S213, adopt an adaptive weight optimization algorithm, fuse the sensitivity analysis results, and obtain the optimized weight vector; construct a HEV risk indicator system based on the optimized weight vector and the indicator hierarchy system to obtain a risk indicator system; determine the indicator attributes and thresholds based on the risk indicator system and the optimized weight vector to obtain a standardized risk factor matrix set.

[0091] Adaptive weight optimization algorithm W*(t) = W(t) + ηgrad L(W(t)); where W*(t) is the optimized weight; W(t) is the initial weight; t is the number of iterations; η is the learning rate; L is the loss function; grad is the gradient operator.

[0092] This embodiment realizes the scientific quantification of risk factors through the construction of a multi-level indicator system and a dynamic weight optimization method. First, a multi-dimensional risk factor identification algorithm is used to extract the hazard, exposure and vulnerability indicators respectively; then, the indicator correlation analysis method is used to identify and eliminate redundant indicators; then, the indicator sensitivity analysis method is used to determine the key indicators; then, the indicator weights are preliminarily determined through expert scoring and hierarchical analysis methods; finally, an adaptive weight optimization algorithm is used to construct a scientific risk indicator system. This embodiment solves the problems of arbitrary indicator selection and subjective weight determination in traditional risk indicator construction methods, and improves the scientific nature of risk quantification through multi-level indicator screening and weight optimization. Especially in the identification of key indicators, it can accurately grasp the dominant factors affecting the formation of risks and provide targeted guidance for risk prevention and control. This embodiment improves the accuracy and reliability of risk factor quantification.

[0093] According to one aspect of the present application, step S25 is specifically as follows:

[0094] Step S251, obtain the cascade effect assessment data output by step S24, the critical path set output by step S23 and the standardized risk factor matrix set output by step S21, adopt the risk factor dynamic decomposition method to extract the time characteristics of the risk and obtain the time characteristic sequence; according to the spatial distribution of the cascade effect, extract the spatial characteristics of the risk and obtain the spatial characteristic sequence; according to the correlation between the risk factors, extract the element characteristics of the risk and obtain the element characteristic sequence.

[0095] The dynamic decomposition method of risk factors is RD(t,s) = α(t)*F(s) + β(t)*G(s) + ε(t,s), where RD is the risk decomposition result, t, s are time and space variables, F(s) is the spatial main mode function, G(s) is the spatial secondary mode function, α(t), β(t) are time coefficient functions, and ε is the residual term.

[0096] Step S252, obtain the time feature sequence, space feature sequence and element feature sequence output by step S251, use a multidimensional risk quantification algorithm to calculate the basic risk intensity, and obtain basic risk data; based on the basic risk data, construct a risk level classification criterion to obtain a risk level standard; based on the basic risk data and the risk level standard, classify the risks to obtain a basic risk layer.

[0097] Step S253, obtain the basic risk layer output by step S252 and the critical path set output by step S23, use the associated risk propagation model to calculate the risk propagation intensity, and obtain the propagation intensity matrix; evaluate the associated risk level based on the propagation intensity matrix and the critical path set to obtain the associated risk data; construct the associated risk distribution based on the associated risk data and the risk level standard to obtain the associated risk layer.

[0098] Step S254, obtain the cascade effect evaluation data output from step S24 and the associated risk layer output from step S253, use the cascade effect superposition algorithm to calculate the cascade effect strength, and obtain the cascade effect strength data; according to the cascade effect strength data, evaluate the cascade risk level to obtain the cascade risk data; according to the cascade risk data and the risk level standard, construct the cascade risk distribution to obtain the cascade risk layer. Cascade effect superposition algorithm: CE(t) = Σ[Ki*Ri(t)exp(-λiτi)], where CE is the cascade effect strength, Ki is the propagation coefficient, Ri is the risk intensity, λi is the attenuation coefficient, τi is the lag parameter, and t is the time variable.

[0099] Step S255, obtain the basic risk layer output by step S252, the associated risk layer output by step S253 and the cascade risk layer output by step S254, use the multi-layer risk fusion algorithm to calculate the risk coupling degree between layers, and obtain the risk coupling matrix; according to the risk coupling matrix, construct the risk map superposition rule to obtain the map superposition criterion; according to the basic risk layer, the associated risk layer, the cascade risk layer and the map superposition criterion, generate a comprehensive risk distribution map to obtain a comprehensive risk map. Multi-layer risk fusion algorithm R = Σ(wi*Ri); where R is the comprehensive risk value; wi is the risk weight of each layer; Ri is the risk value of each layer.

[0100] This implementation realizes the intelligent construction of a comprehensive risk map through multi-level risk analysis and dynamic fusion methods. First, the dynamic decomposition method of risk factors is used to extract the time, space and factor feature sequences; then, the basic risk intensity is calculated and graded through a multi-dimensional risk quantification algorithm; then, the associated risk propagation model is used to evaluate the risk propagation characteristics; then, the cascade effect superposition algorithm is used to quantify the cascade effect intensity; finally, a multi-layer risk fusion algorithm is used to construct a comprehensive risk map. This embodiment overcomes the problem that it is difficult to deal with multi-level risk coupling and dynamic evolution in traditional risk map construction methods. Through risk factor decomposition and multi-layer fusion, the spatiotemporal distribution characteristics of risks are accurately characterized. Especially in the quantification of risk cascade effects, it can effectively evaluate the cumulative impact of risks and provide a global perspective for risk prevention and control. This embodiment improves the expressive power and practical value of risk maps, and provides an intuitive spatial reference for disaster prevention and mitigation decisions.

[0101] According to one aspect of the present application, step S31 is specifically:

[0102] Step S311, obtain pre-stored land use data, satellite remote sensing images and historical land use records, use a multi-temporal land use change analysis method to extract the characteristics of land use type changes, and obtain a land type change matrix; according to the land type change matrix, identify the main change areas and obtain key area identifiers; according to the land type change matrix and key area identifiers, construct a land use classification system to obtain land use classification data.

[0103] Land use change analysis method: LUCC = (At - At-1) / At-1 * 100%; where At is the land use area at time t; At-1 is the land use area at time t-1.

[0104] Step S312: obtain pre-stored statistical yearbook data, population census data and economic survey data, use socio-economic data mining algorithm to extract population distribution characteristics, and obtain a population distribution matrix; extract economic development indicators based on the statistical yearbook data to obtain an economic indicator matrix; construct a socio-economic characteristic database based on the population distribution matrix and the economic indicator matrix to obtain socio-economic characteristic data.

[0105] Step S313, obtain the land use classification data output by step S311 and the socio-economic characteristic data output by step S312, use the multi-source data association analysis method to establish the corresponding relationship between land use and economic indicators, and obtain the land use economic association matrix; calculate the economic value per unit area based on the land use economic association matrix to obtain the unit value matrix; calculate the regional asset distribution based on the land use classification data and the unit value matrix to obtain the asset distribution data.

[0106] Step S314, obtain the asset distribution data output from step S313 and the socio-economic characteristic data output from step S312, use the spatial interpolation optimization algorithm (Kriging interpolation) to calculate the spatial distribution of the economic data, and obtain the economic spatial distribution matrix; construct a population density distribution model based on the socio-economic characteristic data to obtain the population density matrix; generate socio-economic spatial distribution data based on the economic spatial distribution matrix and the population density matrix to obtain spatialized socio-economic data.

[0107] Step S315, obtaining the spatialized socio-economic data output from step S314, the reconstructed characteristic field output from step S1, and the comprehensive risk map output from step S2, using a hazard-prone body classification and recognition algorithm to identify the hazard-prone body type, and obtain a hazard-prone body type matrix; based on the hazard-prone body type matrix and the spatialized socio-economic data, evaluating the hazard-prone body distribution characteristics, and obtaining the hazard-prone body distribution characteristics; constructing a hazard-prone body classification system based on the hazard-prone body type matrix, the hazard-prone body distribution characteristics, and preset classification criteria, and obtaining a classified hazard-prone body data set.

[0108] Multi-source data association analysis method RAI(i,j) = w1S(i,j) + w2T(i,j) + w3*E(i,j); where RAI is the association intensity index, S is the spatial association, T is the temporal association, E is the economic association, and w1, w2, w3 are weight coefficients.

[0109] The classification and identification algorithm of disaster-prone objects is V(x) = Σ[αi*Fi(x)]exp(-βd(x)); where V is the vulnerability index; Fi is the characteristic function; αi is the characteristic weight; β is the spatial attenuation coefficient; and d is the distance function.

[0110] This embodiment achieves accurate classification and spatial distribution expression of disaster-prone bodies through multi-source data mining and spatial processing methods. First, a multi-temporal land use change analysis method is used to extract land use change characteristics; then, a socio-economic data mining algorithm is used to obtain population and economic distribution characteristics; then, a multi-source data association analysis method is used to establish a corresponding relationship between land use and economic indicators; then, a spatial interpolation optimization algorithm is used to spatialize socio-economic data; finally, a disaster-prone body classification and identification algorithm is used to construct a classification system. This embodiment solves the problem that traditional disaster-prone body classification methods are difficult to handle multi-source heterogeneous data and spatial distribution characteristics, and improves the accuracy of disaster-prone body classification through multi-dimensional data mining and spatial processing. In particular, in terms of the spatial expression of socio-economic characteristics, it can accurately depict the spatial distribution law of disaster-prone bodies and provide a reliable basis for vulnerability assessment. This embodiment improves the scientific nature of disaster-prone body classification and the accuracy of spatial expression.

[0111] According to one aspect of the present application, step S35 is specifically:

[0112] Step S351, obtain the evolution pattern library output by step S34, the adaptability assessment matrix output by step S33 and the classified disaster-prone body data set output by step S31, use the direct loss assessment method to calculate the loss rate under different flooding depths, and obtain the basic loss rate matrix; according to the flooding duration and flow velocity data, correct the loss rate to obtain the corrected loss rate matrix; according to the classified disaster-prone body data set and the corrected loss rate matrix, calculate the direct economic loss to obtain the direct loss data.

[0113] Step S352, obtain the direct loss data and classified disaster-prone body data set output by step S351, use the static vulnerability assessment algorithm to calculate the basic vulnerability of the disaster-prone body, and obtain the basic vulnerability matrix; according to the basic vulnerability matrix, extract the spatial distribution characteristics to obtain the vulnerability space characteristics; according to the basic vulnerability matrix and the vulnerability space characteristics, construct the static vulnerability distribution to obtain the static vulnerability layer.

[0114] Step S353, obtain the evolution pattern library output by step S34 and the static vulnerability layer output by step S352, use the dynamic vulnerability evolution model to calculate the vulnerability time evolution characteristics, and obtain the time-varying feature matrix; according to the time-varying feature matrix, identify the key evolution nodes and obtain the evolution node data; according to the time-varying feature matrix and the evolution node data, construct the dynamic vulnerability distribution and obtain the dynamic vulnerability layer.

[0115] Step S354, obtain the adaptability assessment matrix output by step S33 and the dynamic vulnerability layer output by step S353, adopt the adaptability impact assessment method to calculate the regulatory effect of adaptability on vulnerability, and obtain the adjustment coefficient matrix; according to the adjustment coefficient matrix, evaluate the adaptability improvement effect and obtain the improvement effect data; according to the adjustment coefficient matrix and the improvement effect data, construct the adaptive vulnerability distribution and obtain the adaptive vulnerability layer.

[0116] Step S355, obtain the static vulnerability layer output by step S352, the dynamic vulnerability layer output by step S353 and the adaptive vulnerability layer output by step S354, use a multi-dimensional vulnerability integration algorithm to calculate the vulnerability weight of each layer, and obtain a vulnerability weight vector; according to the vulnerability weight vector, construct a vulnerability comprehensive assessment criterion to obtain a comprehensive assessment criterion; according to the static vulnerability layer, the dynamic vulnerability layer, the adaptive vulnerability layer and the comprehensive assessment criterion, generate a comprehensive vulnerability assessment result to obtain a comprehensive vulnerability assessment field.

[0117] Direct loss assessment method L = Σ(ViDiRi), where L is the total loss, Vi is the value, Di is the loss rate, and Ri is the degree of disaster.

[0118] Dynamic vulnerability evolution model DV(t) = V0 + ∫[f(V,t) + g(A,t)]dt, where DV is dynamic vulnerability, V0 is initial vulnerability, f is internal evolution function, g is external action function, A is adaptive capacity, and t is time variable.

[0119] Multidimensional vulnerability integration algorithm IV = Σ[wiViexp(-μit) ][1 - AC(t)] , where IV is the comprehensive vulnerability, wi is the weight coefficient, Vi is the vulnerability of each dimension, μi is the time attenuation coefficient, and AC is the adaptability function.

[0120] This embodiment achieves a comprehensive assessment of the vulnerability of the disaster-prone body through multi-dimensional vulnerability assessment and dynamic integration methods. First, the direct loss assessment method is used to calculate the loss rate under different disaster scenarios; then the basic vulnerability distribution is constructed through the static vulnerability assessment algorithm; then the dynamic vulnerability evolution model is used to evaluate the time evolution characteristics; then the adaptive capacity impact assessment method is used to quantify the regulatory effect of adaptive capacity; finally, a multi-dimensional vulnerability integration algorithm is used to generate a comprehensive assessment result. This embodiment overcomes the problem that traditional vulnerability assessment methods are difficult to handle multi-dimensional vulnerability characteristics and dynamic evolution processes. Through multi-level assessment and dynamic integration, the spatiotemporal evolution characteristics of vulnerability are accurately characterized. Especially in terms of adaptive capacity assessment, it can effectively quantify the recovery capacity of the disaster-prone body and provide targeted suggestions for risk prevention and control. This embodiment improves the comprehensiveness and accuracy of vulnerability assessment.

[0121] According to one aspect of the present application, step S42 is specifically:

[0122] Step S421, obtain the fused data set output by step S41, use a multidimensional risk factor identification algorithm to extract the time variation characteristics of the risk factors, and obtain a time feature matrix; based on the fused data set, extract the spatial distribution characteristics of the risk factors to obtain a spatial feature matrix; based on the time feature matrix and the spatial feature matrix, construct a risk factor feature space to obtain risk feature space data.

[0123] Step S422, obtain the risk characteristic spatial data output in step S421, use the coupling degree dynamic calculation method to calculate the time coupling strength between risk factors, and obtain the time coupling matrix; calculate the spatial coupling strength between risk factors based on the risk characteristic spatial data to obtain the spatial coupling matrix; construct a comprehensive coupling strength distribution based on the time coupling matrix and the spatial coupling matrix to obtain the risk coupling strength matrix.

[0124] Step S423, obtain the risk coupling intensity matrix output by step S422, use the risk amplification effect identification algorithm to calculate the positive incentive effect between risk factors, and obtain the positive incentive matrix; identify the risk chain amplification node according to the risk coupling intensity matrix, and obtain the amplification node data; evaluate the risk amplification effect according to the positive incentive matrix and the amplification node data, and obtain the risk amplification coefficient matrix.

[0125] Step S424, obtain the risk coupling intensity matrix output by step S422 and the risk amplification coefficient matrix output by step S423, adopt the risk suppression mechanism analysis method to identify the risk suppression factors and obtain the suppression factor set; calculate the suppression intensity based on the suppression factor set to obtain the suppression intensity data; construct the risk suppression effect model based on the suppression factor set and the suppression intensity data to obtain the risk suppression coefficient matrix.

[0126] Step S425, obtain the risk coupling intensity matrix output by step S422, the risk amplification coefficient matrix output by step S423 and the risk suppression coefficient matrix output by step S424, use the risk effect comprehensive assessment algorithm to calculate the net effect of the risk action, and obtain the risk net effect matrix; according to the risk net effect matrix, construct a risk field distribution model to obtain risk field distribution data; according to the risk net effect matrix and the risk field distribution data, generate a risk coupling field distribution map to obtain the risk coupling field.

[0127] The coupling degree calculation method is C = sqrt(u*v) / (u+v); where u and v are coupling elements.

[0128] Risk amplification effect identification algorithm AE(t) = R0Π[1 + αiCi(t)], where AE is the amplification effect, R0 is the basic risk, αi is the amplification coefficient, and Ci is the coupling intensity.

[0129] Risk suppression mechanism analysis method SI(t) = Σ[βi*Ii(t)exp(-λit)], where SI is the suppression intensity, βi is the suppression coefficient, Ii is the intervention intensity, and λi is the time attenuation coefficient.

[0130] This embodiment realizes the accurate construction of the risk coupling field through a multi-dimensional risk coupling and comprehensive effect evaluation method. First, a multi-dimensional risk factor identification algorithm is used to extract the spatiotemporal characteristics of risk factors; then, the coupling relationship between risk factors is evaluated through a coupling degree dynamic calculation method; then, a risk amplification effect identification algorithm is used to quantify the positive incentive effect; then, a risk suppression mechanism analysis method is used to evaluate the suppression effect; finally, a risk effect comprehensive evaluation algorithm is used to generate a risk coupling field. This embodiment solves the problem that traditional risk coupling analysis methods are difficult to handle multi-dimensional risk interactions and dynamic evolution characteristics. Through multi-level coupling analysis and effect evaluation, the compound action mechanism of risks is accurately portrayed. Especially in terms of the quantification of risk amplification and suppression effects, the net effect of risks can be effectively evaluated, providing a scientific basis for risk prevention and control. This embodiment improves the accuracy and reliability of risk coupling analysis.

[0131] According to one aspect of the present application, step S45 is specifically:

[0132] Step S451, obtain the risk coupling field output by step S42, the cumulative effect field output by step S44 and the risk conversion pattern set output by step S43, use the instant risk assessment algorithm to calculate the risk intensity distribution at the current moment, and obtain the instant intensity matrix; according to the instant intensity matrix, identify the high-risk areas and obtain the key area identification; according to the instant intensity matrix and the key area identification, construct the instant risk distribution map and obtain the instant risk layer.

[0133] Step S452, obtain the cumulative effect field output by step S44 and the immediate risk layer output by step S451, adopt the risk cumulative effect analysis method to calculate the risk cumulative intensity, and obtain the cumulative intensity matrix; according to the cumulative intensity matrix, evaluate the impact range of the cumulative effect and obtain the impact range data; according to the cumulative intensity matrix and the impact range data, construct a cumulative risk distribution map to obtain the cumulative risk layer.

[0134] Step S453, obtain the risk conversion pattern set output by step S43 and the cumulative risk layer output by step S452, use the risk evolution trend analysis algorithm to predict the risk development trend, and obtain the trend prediction matrix; according to the trend prediction matrix, identify the key evolution nodes and obtain the evolution node data; according to the trend prediction matrix and the evolution node data, construct an evolution risk distribution map to obtain the evolution risk layer.

[0135] Step S454, obtain the immediate risk layer output by step S451, the cumulative risk layer output by step S452 and the evolutionary risk layer output by step S453, adopt a multi-dimensional risk comprehensive assessment method to calculate the comprehensive risk index, and obtain a comprehensive risk index matrix; according to the comprehensive risk index matrix, determine the risk level classification standard and obtain the level classification criteria; according to the comprehensive risk index matrix and the level classification criteria, perform risk level classification and obtain a risk classification result.

[0136] Step S455, obtain the risk grading results and comprehensive risk index matrix output by step S454, use the intelligent early warning analysis algorithm to identify the early warning trigger conditions, and obtain the early warning trigger rules; determine the early warning level according to the early warning trigger rules, and obtain the early warning level data; generate a comprehensive risk assessment report according to the risk grading results, early warning level data and the preset report template, and obtain a comprehensive risk assessment report, which includes a risk level distribution map, a risk evolution trend map, key risk area identification results and risk early warning recommendations.

[0137] Risk cumulative effect analysis, CR = Σ(Ri*ti); where CR is the cumulative risk; Ri is the risk intensity; ti is the action time;

[0138] Risk evolution trend analysis algorithm, RT(t) = αR(t) + β∫R(τ)dτ + γ*dR / dt; where RT is the risk trend; R is the risk intensity; α, β, γ are weight coefficients; τ is the integral variable;

[0139] Intelligent early warning analysis algorithm WL(t) = f[R(t), ΔR(t), ∫R(t)dt], where WL is the warning level, R is the risk value, ΔR is the risk change rate, and f is the warning discriminant function.

[0140] This embodiment realizes the dynamic assessment and early warning of comprehensive risks through multi-dimensional risk assessment and intelligent early warning analysis methods. First, an immediate risk assessment algorithm is used to calculate the current risk intensity distribution; then, the cumulative risk characteristics are evaluated through the risk cumulative effect analysis method; then, the risk evolution trend analysis algorithm is used to predict the risk development trend; then, a multi-dimensional risk comprehensive assessment method is used to calculate the comprehensive risk index; finally, an intelligent early warning analysis algorithm is used to generate a risk assessment report. This embodiment overcomes the problem that traditional risk assessment methods are difficult to handle multi-time scale risk evolution and early warning triggering, and improves the timeliness of risk assessment through multi-dimensional risk analysis and intelligent early warning. Especially in terms of risk early warning, it can accurately identify early warning triggering conditions and provide timely support for disaster prevention and mitigation decisions. This embodiment improves the predictability of risk assessment and the accuracy of early warning, and provides important decision-making support for urban flood control and disaster reduction. It should be noted that due to the large number of methods, no detailed description is given for the existing methods.

[0141] According to another aspect of the present application, a system for calculating composite flood risk based on a HEV framework is also provided, comprising:

[0142] at least one processor; and,

[0143] a memory communicatively connected to at least one of the processors; wherein,

[0144] The memory stores instructions that can be executed by the processor, and the instructions are used to be executed by the processor to implement the method for calculating composite flood risk based on the HEV framework described in any of the above embodiments.

[0145] It should be noted that the various specific technical features described in the above specific embodiments can be combined in any suitable manner without contradiction. In order to avoid unnecessary repetition, the present invention will not further describe various possible combinations.

Claims

1. A method for calculating composite flood risk based on the HEV framework, characterized in that: The steps include: Step S1, obtaining rainfall monitoring data, water level monitoring data and terrain monitoring data, using a feature matrix decomposition algorithm to obtain a basic feature matrix group; according to the basic feature matrix group, using a feature importance calculation method to obtain a feature importance vector; According to the feature importance vector and the basic feature matrix group, an adaptive feature reconstruction algorithm is used to obtain a reconstructed feature field; according to the reconstructed feature field, a multi-scale consistency test method is used to obtain a dynamic feature verification report; Step S2, obtaining the reconstructed characteristic field, historical disaster data and geographic information data, using a multidimensional risk factor analysis method to obtain a standardized risk factor matrix set; based on the standardized risk factor matrix set, using a dynamic association network generation algorithm to obtain a dynamic association network sequence; According to the dynamic association network sequence, the key path extraction algorithm is used to obtain the key path set and stability assessment data; according to the key path set and stability assessment data, the risk cascade propagation calculation method is used to obtain the cascade effect assessment data; according to the cascade effect assessment data, the key path set and the standardized risk factor matrix set, the multi-level risk map construction method is used to obtain the comprehensive risk map; Step S3, obtaining the reconstructed characteristic field, the comprehensive risk map and the pre-stored socio-economic data, adopting the adaptive hazard-bearing body classification method to obtain the classified hazard-bearing body data set; adopting the multimodal response analysis method based on the classified hazard-bearing body data set and the historical disaster response data to obtain the vulnerability response surface; adopting the dynamic adaptive capacity assessment method based on the vulnerability response surface and the socio-economic capacity data to obtain the adaptive capacity assessment matrix; adopting the evolutionary pattern recognition method based on the vulnerability response surface and the adaptive capacity assessment matrix to obtain the evolutionary pattern library; adopting the multidimensional comprehensive assessment method based on the evolutionary pattern library, the adaptive capacity assessment matrix and the classified hazard-bearing body data set to obtain the comprehensive vulnerability assessment field; Step S4, obtaining the reconstructed feature field, the comprehensive risk map and the comprehensive vulnerability assessment field, and using a heterogeneous data fusion method to obtain a fused data set; Based on the fused data set, a multi-dimensional risk coupling analysis method is used to obtain the risk coupling field; According to the risk coupling field and historical risk evolution data, the risk transformation mechanism identification method is adopted to obtain the risk transformation pattern set; Based on the risk coupling field and risk transformation pattern set, the risk cumulative effect quantification method is used to obtain the cumulative effect field; based on the risk coupling field, cumulative effect field and risk transformation pattern set, the dynamic comprehensive assessment method is used to obtain a comprehensive risk assessment report.

2. A method for calculating composite flood risk based on HEV framework according to claim 1, characterized in that: The step S1 is specifically as follows: Step S11, obtaining rainfall monitoring data, water level monitoring data and terrain monitoring data of the monitoring site, constructing a rainfall spatial distribution matrix, a water level spatial distribution matrix and a terrain feature matrix according to the spatial location information; obtaining a standardized spatial distribution matrix set through a linear transformation method according to the rainfall spatial distribution matrix, the water level spatial distribution matrix and the terrain feature matrix; Step S12, obtaining a standardized spatial distribution matrix set, calculating the correlation coefficient between spatial positions, and obtaining a spatial correlation matrix; calculating the change characteristics of the time series according to the time series data in the standardized spatial distribution matrix set, and obtaining a time evolution matrix; calculating the coupling degree between variables according to different monitoring variable data in the standardized spatial distribution matrix set, and obtaining a variable coupling matrix; using a multi-level decomposition operation according to the spatial correlation matrix, the time evolution matrix and the variable coupling matrix, obtaining a basic feature matrix group; Step S13, obtain the basic feature matrix group, calculate the variation range of each feature in the time dimension, and obtain the time-varying weight vector; calculate the spatial influence coefficient according to the spatial distribution range of the basic feature matrix group, and obtain the spatial influence matrix; according to the time-varying weight vector and the spatial influence matrix, use weighted fusion operation to obtain the feature importance vector; Step S14, obtaining a basic feature matrix group and a feature importance vector, constructing a nonlinear weight function, calculating a dynamic weight coefficient of each feature, and obtaining a dynamic weight matrix; performing weighted combination on the basic feature matrix group according to the dynamic weight matrix to obtain an initial reconstructed feature field; obtaining time and space continuity constraints, optimizing and adjusting the initial reconstructed feature field, and obtaining a reconstructed feature field; Step S15, obtain the reconstructed feature field and the standardized spatial distribution matrix set, calculate the feature consistency index of the local area, and obtain the local consistency matrix; based on the reconstructed feature field, calculate the feature consistency index of the global range to obtain the global consistency vector; based on the local consistency matrix and the global consistency vector, construct a dynamic threshold judgment criterion to generate a dynamic feature verification report.

3. A method for calculating composite flood risk based on HEV framework according to claim 2, characterized in that: The step S2 is specifically as follows: Step S21, obtaining the reconstructed characteristic field, pre-stored historical disaster data and geographic information data, extracting hydrological and meteorological parameters, and obtaining a hydrological and meteorological factor matrix; extracting topographic and geomorphic features according to the geographic information data, and obtaining a topographic and geomorphic factor matrix; Obtain land use and flood control project data, extract human activity characteristics, and obtain a human activity factor matrix; according to the hydro-meteorological factor matrix, topographic factor matrix, and human activity factor matrix, adopt a standardized processing method to obtain a standardized risk factor matrix set; Step S22, obtaining a standardized risk factor matrix set, calculating the correlation strength between risk factors at different times, and obtaining a time-varying correlation strength matrix; constructing a dynamic threshold discriminant function based on the time-varying correlation strength matrix, screening significant correlation relationships, and obtaining a significant correlation matrix; According to the significant correlation matrix, the time-varying network structure is constructed to obtain the dynamic correlation network sequence; Step S23, obtaining a dynamic association network sequence, calculating the importance of each network node, and obtaining a node importance vector; identifying the main path of risk propagation based on the node importance vector and the dynamic association network sequence, and obtaining a critical path set; based on the critical path set, evaluating the time stability of the path, and obtaining stability evaluation data; Step S24, obtaining a critical path set and stability assessment data, calculating the risk propagation intensity on the critical path, and obtaining a propagation intensity matrix; According to the transmission intensity matrix, the delayed effect of risk transmission is calculated to obtain the time lag effect matrix; According to the transmission intensity matrix and the time lag effect matrix, the cumulative impact of the risk is calculated to obtain the cascade effect assessment data; Step S25, obtaining cascading effect assessment data, a critical path set and a standardized risk factor matrix set, calculating the basic risk distribution, and obtaining a basic risk layer; According to the basic risk layer and the critical path set, the associated risk distribution is calculated to obtain the associated risk layer; According to the associated risk layers and cascade effect assessment data, the cascade risk distribution is calculated to obtain the cascade risk layers; According to the basic risk layer, associated risk layer and cascade risk layer, a weighted superposition method is adopted to obtain a comprehensive risk map.

4. A method for calculating composite flood risk based on HEV framework according to claim 3, characterized in that: The step S3 is specifically as follows: Step S31, obtaining the reconstructed characteristic field, the comprehensive risk map and the pre-stored socio-economic data, extracting the spatial distribution characteristics of the hazard-prone body, and obtaining the hazard-prone body characteristic matrix; constructing dynamic classification rules according to the time series data, and obtaining the dynamic classification criteria; classifying the hazard-prone body according to the hazard-prone body characteristic matrix and the dynamic classification criteria, and obtaining the classified hazard-prone body data set; Step S32, obtaining a classified disaster-prone body data set and pre-stored historical disaster response data, extracting response characteristics of different types of disaster-prone bodies, and obtaining a response characteristic matrix; According to the response characteristic matrix, the sensitivity of the disaster-prone body to the disaster is calculated to obtain the time-varying sensitivity matrix; according to the response characteristic matrix and the time-varying sensitivity matrix, the recovery capacity of the disaster-prone body is evaluated to obtain the recovery capacity matrix; according to the response characteristic matrix, the time-varying sensitivity matrix and the recovery capacity matrix, the response surface model is constructed to obtain the vulnerability response surface; Step S33, obtaining the vulnerability response surface and the pre-stored socio-economic capacity data, constructing an adaptability assessment index system, and obtaining an adaptability index matrix; calculating the resource allocation efficiency according to the socio-economic capacity data, and obtaining a resource efficiency matrix; evaluating the system resilience according to the adaptability index matrix and the resource efficiency matrix, and obtaining a system resilience matrix; using a comprehensive assessment method according to the adaptability index matrix, the resource efficiency matrix, and the system resilience matrix, obtaining an adaptability assessment matrix; Step S34, obtaining a vulnerability response surface and an adaptability assessment matrix, extracting spatiotemporal variation characteristics of vulnerability, and obtaining an evolution feature vector set; according to Evolution feature vector set, identify typical evolution patterns, and obtain key evolution patterns; predict vulnerability change trends based on key evolution patterns, and obtain evolution trend data; construct an evolution pattern database based on key evolution patterns and evolution trend data, and obtain an evolution pattern library; Step S35, obtain the evolution pattern library, the adaptability assessment matrix and the classified hazard-prone body data set, calculate the static vulnerability distribution, and obtain the static vulnerability layer; calculate the dynamic vulnerability distribution according to the evolution pattern library to obtain the dynamic vulnerability layer; calculate the adaptive vulnerability distribution according to the adaptability assessment matrix to obtain the adaptive vulnerability layer; according to the static vulnerability layer, the dynamic vulnerability layer and the adaptive vulnerability layer, adopt a multidimensional integration method to obtain a comprehensive vulnerability assessment field.

5. A method for calculating composite flood risk based on HEV framework according to claim 4, characterized in that: The step S4 is specifically as follows: Step S41, obtaining the reconstructed feature field, the comprehensive risk map and the comprehensive vulnerability assessment field, calculating the correlation characteristics between the data, and obtaining the data correlation matrix; constructing the consistency test standard according to the spatiotemporal distribution characteristics, and obtaining the spatiotemporal consistency test matrix; According to the data association matrix and the spatiotemporal consistency test matrix, the multi-source data are fused to obtain a fused data set; Step S42: Obtain a fused data set, calculate the coupling degree between different risk types, and obtain a risk coupling intensity matrix; According to the risk coupling intensity matrix, the risk amplification effect is calculated to obtain the risk amplification coefficient matrix; According to the risk coupling intensity matrix, the risk suppression effect is calculated to obtain the risk suppression coefficient matrix; According to the risk coupling intensity matrix, risk amplification coefficient matrix and risk suppression coefficient matrix, the risk coupling field is obtained by using the coupling effect integration method. Step S43: Obtain the risk coupling field and pre-stored historical risk evolution data, extract risk conversion features, and obtain a conversion feature matrix; Transformation feature matrix, identify key transformation nodes, and obtain key node data; calculate transformation probability based on transformation feature matrix and key node data, and obtain transformation probability matrix; construct transformation pattern database based on transformation feature matrix, key node data and transformation probability matrix, and obtain risk transformation pattern set; Step S44, obtaining the risk coupling field and the risk conversion pattern set, calculating the risk cumulative intensity, and obtaining a cumulative intensity matrix; According to the cumulative intensity matrix, the time lag impact is evaluated to obtain the time lag impact matrix; According to the cumulative intensity matrix and the time-lag influence matrix, the long-term effect is calculated to obtain the long-term effect matrix; according to the cumulative intensity matrix, the time-lag influence matrix and the long-term effect matrix, the effect superposition method is used to obtain the cumulative effect field; Step S45, obtaining the risk coupling field, cumulative effect field and risk conversion pattern set, calculating the immediate risk distribution, and obtaining the immediate risk layer; According to the cumulative effect field, the cumulative risk distribution is calculated to obtain the cumulative risk layer; According to the risk conversion pattern set, the evolutionary risk distribution is calculated to obtain the evolutionary risk layer; Based on the immediate risk layer, cumulative risk layer and evolutionary risk layer, a comprehensive risk index is constructed to obtain a comprehensive risk assessment report.

6. A method for calculating composite flood risk based on HEV framework according to claim 5, characterized in that: The step S11 is specifically as follows: Step S111, obtaining the original rainfall monitoring data, original water level monitoring data and original terrain monitoring data of the monitoring site, using a multi-source data quality evaluation method, calculating the data reliability index, and obtaining a data quality evaluation matrix; according to the data quality evaluation matrix, screening valid data points to obtain a valid monitoring data set; Step S112: Acquire a valid monitoring data set, use a spatiotemporal outlier intelligent recognition algorithm to identify abnormal data points, and obtain an outlier labeling matrix; According to the outlier labeling matrix, the abnormal data is corrected to obtain the corrected monitoring data set; according to the corrected monitoring data set, the data gap filling intelligent algorithm is used to supplement the missing data to obtain the complete monitoring data set; Step S113, obtaining a complete monitoring data set, using a flood numerical simulation method to simulate the levee breach process, and obtaining breach process data; According to the breach process data, the evolution process of breach water flow is simulated to obtain the water flow evolution data; according to the water flow evolution data, the waterlogging process is calculated to obtain the waterlogging process data; According to the breach process data, water flow evolution data and water accumulation process data, the simulation parameter adaptive optimization method is adopted to obtain the optimized simulation results; Step S114, obtaining the optimization simulation results, extracting the maximum flooding water depth data, and obtaining the water depth distribution matrix; according to the optimization simulation results, extracting the maximum flow velocity data, and obtaining the flow velocity distribution matrix; according to the optimization simulation results, extracting the flooding duration data, and obtaining the flooding duration matrix; according to the water depth distribution matrix, the flow velocity distribution matrix, and the flooding duration matrix, constructing a spatial distribution data set, and obtaining a spatial distribution matrix set; Step S115, obtain a set of spatial distribution matrices, use a spatial interpolation optimization algorithm to perform spatial processing on discrete data, and obtain a continuously distributed data set; based on the continuously distributed data set, use a scale conversion algorithm to unify the spatial resolution and obtain a unified scale data set; based on the unified scale data set, use a standardized processing method to obtain a standardized set of spatial distribution matrices.

7. The method for calculating composite flood risk based on the HEV framework according to claim 5, characterized in that: The step S12 is specifically as follows: Step S121, obtaining a standardized spatial distribution matrix set, using a dynamic feature decomposition algorithm to extract time series change features, and obtaining a time series feature matrix; according to the standardized spatial distribution matrix set, using a spatial feature recognition algorithm to extract spatial distribution features, and obtaining a spatial feature matrix; according to the standardized spatial distribution matrix set, using a multivariate feature extraction algorithm to identify variable features, and obtaining a variable feature matrix; Step S122, obtaining a time series feature matrix, a space feature matrix and a variable feature matrix, using an adaptive correlation analysis method to calculate the time dimension correlation coefficient to obtain a time correlation matrix; According to the spatial feature matrix, the spatial position correlation coefficient is calculated to obtain the spatial correlation matrix; According to the variable characteristic matrix, the correlation coefficients between variables are calculated to obtain the variable correlation matrix; Step S123, obtaining a time correlation matrix, a space correlation matrix and a variable correlation matrix, using a dynamic weight calculation method to determine the time dimension weight, and obtaining a time weight vector; according to the space correlation matrix, calculating the space position weight, and obtaining a space weight vector; According to the variable correlation matrix, the variable weights are calculated to obtain the variable weight vector; Step S124, obtaining a time weight vector, a space weight vector and a variable weight vector, using a multi-dimensional coupling degree calculation method to calculate the time-space coupling strength, and obtain a time-space coupling matrix; calculating the time-varying coupling strength according to the time weight vector and the variable weight vector, and obtaining a time-varying coupling matrix; calculating the space-varying coupling strength according to the space weight vector and the variable weight vector, and obtaining a space-varying coupling matrix; Step S125, obtaining a spatiotemporal coupling matrix, a time-varying coupling matrix and a space-varying coupling matrix, and using a multi-level eigendecomposition algorithm to extract dominant eigenmodes to obtain a dominant eigenmatrix; According to the dominant feature matrix, the secondary feature pattern is identified to obtain the secondary feature matrix; According to the dominant feature matrix and the secondary feature matrix, a feature hierarchy is constructed to obtain a basic feature matrix group.

8. The method for calculating composite flood risk based on HEV framework according to claim 5, characterized in that: The step S15 is specifically as follows: Step S151, obtaining a reconstructed feature field and a standardized spatial distribution matrix set, using a multi-scale decomposition algorithm to perform scale decomposition on the reconstructed feature field to obtain a multi-scale feature sequence; according to the multi-scale feature sequence, calculating the feature distribution at each scale to obtain a scale distribution matrix; according to the multi-scale feature sequence and the standardized spatial distribution matrix set, calculating the correspondence between each scale to obtain a scale correspondence matrix; Step S152, obtaining a scale distribution matrix and a scale correspondence matrix, using a local feature consistency evaluation method to calculate the feature difference of the local region to obtain a local difference matrix; identifying the feature mutation region according to the local difference matrix to obtain a mutation region identifier; evaluating the local consistency level according to the local difference matrix and the mutation region identifier to obtain a local consistency index; Step S153, obtaining the local consistency index and the multi-scale feature sequence, using the index forward algorithm to convert the reverse index to obtain a forward index set; constructing the index polarity discrimination criterion according to the forward index set to obtain a polarity discrimination matrix; optimizing the index direction according to the forward index set and the polarity discrimination matrix to obtain an optimized index set; Step S154, obtaining an optimization index set, using an adaptive normalization algorithm to calculate the index variation range, and obtaining an index range matrix; determining a normalization parameter according to the index range matrix, and obtaining a normalized parameter set; performing an index normalization operation according to the optimization index set and the normalized parameter set, and obtaining a normalized index matrix; Step S155, obtain the normalized indicator matrix and the local consistency indicator, adopt the dynamic threshold construction algorithm, calculate the verification threshold of each level, and obtain the verification threshold sequence; according to the normalized indicator matrix and the verification threshold sequence, perform multi-level verification to obtain the verification result matrix; according to the verification result matrix and the local consistency indicator, generate a feature verification report to obtain a dynamic feature verification report.

9. The method for calculating composite flood risk based on HEV framework according to claim 5, characterized in that: The step S21 is specifically as follows: Step S211, obtaining the reconstructed characteristic field, the pre-stored historical disaster data and geographic information data, using a multi-dimensional risk factor identification algorithm, extracting a hazard index set, and obtaining a hazard index matrix; extracting an exposure index set based on the historical disaster data and geographic information data, and obtaining an exposure index matrix; extracting a vulnerability index set based on the historical disaster data and geographic information data, and obtaining a vulnerability index matrix; Step S212: Obtain a hazard index matrix, an exposure index matrix, and a vulnerability index matrix, and use an index correlation analysis method to calculate the correlation between the indicators to obtain an index correlation matrix; identify redundant indicators based on the index correlation matrix to obtain a redundant index set; remove indicators from the redundant index set based on the hazard index matrix, the exposure index matrix, and the vulnerability index matrix to obtain a preliminary index system; Step S213, obtaining a preliminary index system, using an index sensitivity analysis method to calculate the degree of influence of each index on the result, and obtaining a sensitivity matrix; According to the sensitivity matrix, determine the key indicators and obtain the key indicator set; According to the key indicator set, a hierarchical analysis structure is constructed to obtain an indicator hierarchy system; Step S214, obtaining an indicator hierarchy system, using an expert scoring method, collecting expert scoring data, and obtaining an expert scoring matrix; calculating the consistency ratio according to the expert scoring matrix, and obtaining a consistency test result; using an improved hierarchical analysis method according to the expert scoring matrix and the consistency test result, obtaining an initial weight vector; Step S215, obtaining an initial weight vector and a sensitivity matrix, using an adaptive weight optimization algorithm, integrating sensitivity analysis results, and obtaining an optimized weight vector; According to the optimized weight vector and the indicator hierarchy system, the HEV risk indicator system is constructed to obtain the risk indicator system; according to the risk indicator system and the optimized weight vector, the indicator attributes and thresholds are determined to obtain the standardized risk factor matrix set.

10. A system for calculating composite flood risk based on the HEV framework, characterized in that: include: at least one processor; as well as, a memory communicatively connected to at least one of the processors; wherein, The memory stores instructions executable by the processor, and the instructions are used to be executed by the processor to implement the method for calculating composite flood risk based on the HEV framework as described in any one of claims 1 to 9.

Images