Flood risk dynamic zoning method and system based on multiple factor optimization

Abstract

The present invention discloses a method and system for dynamic zoning of flood risk based on multiple factor optimization, constructs an initial flood risk assessment index set, screens to obtain a flood risk assessment index set and calculates the frequency value of each index; constructs a comprehensive discrimination index value to further screen to obtain a key flood risk assessment index set; constructs a risk assessment model and a key flood risk assessment index weight optimization model, and uses a genetic-local search hybrid algorithm optimized by simulated annealing to calculate the optimal weight combination scheme; calculates the similarity between the scenario at each time point in the study area and the historical typical flood events to select the optimal weight combination scheme, and inputs the risk assessment model for calculation to obtain the dynamic zoning of flood risk in the study area. The present invention can not only provide a scientific and reasonable dynamic risk assessment, but also provide effective decision support for relevant decision makers, which is helpful to improve the flood control resilience and the ability to cope with extreme environments in the basin area.

Description

Technical Field

[0001] The invention relates to a flood risk dynamic zoning method, in particular to a flood risk dynamic zoning method based on multiple optimization of factors. Background Art

[0002] Flood risk assessment and management are important links in urban planning, environmental protection and disaster emergency response. With the intensification of global climate change and the acceleration of urbanization, the frequency and intensity of flood events are increasing, seriously threatening human life, property safety and the ecological environment. Therefore, how to effectively identify and assess flood risks and conduct reasonable dynamic zoning has become an urgent problem to be solved.

[0003] Traditional flood risk assessment is usually carried out by drawing a static risk zoning map by superimposing risk factors according to certain weights. The selection of each risk factor and the setting of weights are often based on subjective experience, and there is also a lack of verification of the accuracy of risk assessment. The present invention aims at the problems existing in the traditional method. When selecting risk assessment indicators, at least two meta-analysis methods are used to select flood risk assessment indicators forwardly and reversely until convergence, and a comprehensive discrimination indicator is proposed to identify key flood risk assessment indicators; based on the prior information of historical typical scenarios, the general set and special set of historical typical flood events are divided, and the objective function is that the risk assessment model is closest to the historical actual situation, a risk assessment indicator weight optimization model is constructed, and an intelligent optimization algorithm is proposed to solve the optimization model to obtain the corresponding risk assessment indicator weight optimization scheme; a temporal and spatial similarity threshold is introduced, and a flood risk dynamic zoning method is proposed to more flexibly respond to changing flood risks, adjust management strategies in a timely manner, and improve the comprehensiveness and accuracy of flood risk assessment. Summary of the invention

[0004] The invention aims to provide a method for dynamic zoning of flood risks based on multiple optimization of factors to solve the above problems existing in the prior art. On the other hand, a system for dynamic zoning of flood risks based on multiple optimization of factors is provided.

[0005] The technical solution is a method for dynamic flood risk zoning based on multiple optimization factors, including the following steps:

[0006] Step S1, obtaining and retrieving flood risk assessment literature, using at least two meta-analysis methods to fuse and identify flood risk assessment indicators, forming an initial flood risk assessment indicator set and clustering, and extracting classification keywords for each type of flood risk assessment indicator; based on the classification keywords, using at least two meta-analysis methods to identify flood risk assessment indicators again, screening to obtain a preliminary screening flood risk assessment indicator set and calculating the frequency value of each flood risk assessment indicator therein;

[0007] Step S2, reading the initial screening flood assessment index set, and calculating the contribution of each flood risk assessment index therein, calculating the comprehensive index value of each flood risk assessment index in turn, further calculating the mean of the comprehensive evaluation value of the flood risk assessment index, and then calculating the comprehensive discrimination index value, and screening the flood risk assessment index again based on the comprehensive discrimination index value of the flood risk assessment index to obtain the key flood risk assessment index set;

[0008] Step S3, extracting water level and flow data of the study area, dividing the study area into different levels, obtaining risk level raster data of the study area, rasterizing key flood risk assessment indicators on the study area, and using them as inputs of the pre-built key flood risk assessment indicator weight optimization model and risk assessment model, solving them using a genetic-local search hybrid algorithm optimized by simulated annealing, and obtaining the optimal weight combination scheme of the key flood risk assessment indicator set;

[0009] Step S4: During operation, the similarity between the scenario at each time point in the study area and the scenario of typical historical flood events is calculated, and the optimal weight combination scheme of the key flood risk assessment indicator set is selected based on the similarity, and the scheme is input into the risk assessment model for calculation to obtain the dynamic zoning of flood risk in the study area.

[0010] According to one aspect of the present application, step S1 further comprises:

[0011] Step S11, obtaining and retrieving flood risk assessment literature, using basic meta-analysis, network meta-analysis, multivariate meta-analysis and Bayesian meta-analysis in turn to identify four sets of flood risk assessment indicator sets, integrating the four sets of flood risk assessment indicator sets to obtain an initial flood risk assessment indicator set and counting the number of occurrences of each flood risk assessment indicator in the initial flood risk assessment indicator set;

[0012] Step S12: divide each flood risk assessment indicator in the initial flood risk assessment indicator set into a number of words, calculate the TF-IDF value of each word in turn, use the words with TF-IDF values ​​less than the threshold as the keywords of the current flood risk assessment indicator, and divide the initial flood risk assessment indicator set into M categories based on the keywords using a hierarchical clustering method, each category has its own classification keyword, and M is a positive integer greater than 2;

[0013] Step S13: re-identify flood risk assessment indicators based on the M-category classification keywords, screen out new flood risk assessment indicators and calculate the frequency values ​​of the new flood risk assessment indicators, construct a flood risk assessment indicator set and calculate the frequency value of each indicator in the initial screening flood risk assessment indicator set.

[0014] According to one aspect of the present application, step S13 is further:

[0015] Step S13a, using M-class classification keywords to sequentially use meta-analysis, network meta-analysis, multivariate meta-analysis and Bayesian meta-analysis to identify flood risk assessment indicators in flood risk assessment literature, screen out newly emerged flood risk assessment indicators and calculate the frequency values ​​of the newly emerged flood risk assessment indicators;

[0016] Step S13b, calculating the frequency value of each flood risk assessment indicator and sorting them by size, and selecting the flood risk assessment indicator with the smallest frequency value;

[0017] Step S13c, comparing the frequency values ​​of the newly emerged flood risk assessment indicators with the flood risk assessment indicators with the minimum frequency value, and classifying the newly emerged flood risk assessment indicators that are not less than the minimum frequency value of the flood risk assessment indicators in the initial flood risk assessment indicator set into the initial flood risk assessment indicator set;

[0018] Step S13d, repeating steps S13a to S13c until no new indicators can be classified into the initial flood risk assessment indicator set, stopping iteration, and obtaining the flood risk assessment indicator set and the frequency value of each indicator in the flood risk assessment indicator set.

[0019] According to one aspect of the present application, step S2 further comprises:

[0020] Step S21, reading the initial screening flood assessment index set, and calculating the contribution of each index in the initial screening flood risk assessment index set in sequence;

[0021] Step S22, calculate the comprehensive index value of each flood risk assessment indicator in turn, calculate the product of the frequency value and the contribution of each indicator in the flood risk assessment indicator set in turn as the comprehensive index value of the flood risk assessment indicator, and calculate the mean of the comprehensive index values ​​of the flood risk assessment indicators, calculate the comprehensive discrimination index value of each flood risk assessment indicator based on the comprehensive index value of the flood risk assessment indicator and the mean of the comprehensive index value of the flood risk assessment indicator, and according to a preset threshold, remove the flood risk assessment indicators whose comprehensive discrimination index values ​​are less than the threshold to obtain a key flood risk assessment indicator set.

[0022] According to one aspect of the present application, step S21 is further:

[0023] Step S21a, reading the initial screening flood assessment index set and the typical flood event data of the study area in history, obtaining the measured data corresponding to each indicator in the initial screening flood risk assessment index set from the typical flood event data, and calculating the historical mean of each flood risk assessment indicator in turn as the first assessment result of the flood risk assessment indicator;

[0024] Step S21b, sequentially selecting a flood risk assessment indicator and keeping other flood risk assessment indicators unchanged, calculating the standard deviation of the selected flood risk assessment indicator, taking the sum of the standard deviation of the selected flood risk assessment indicator and the historical mean of the selected flood risk assessment indicator as a new indicator value and evaluating the risk, to obtain a second evaluation result of the selected flood risk assessment indicator;

[0025] Step S21c, sequentially calculating the contribution of each indicator in the initial screening flood assessment indicator set, wherein the contribution is the ratio of the difference between the first assessment result and the second assessment result of the flood risk assessment indicator to the historical mean.

[0026] According to one aspect of the present application, step S3 further comprises:

[0027] Step S31, extracting water level and flow data of the study area based on historical typical flood event data of the study area, dividing the study area into five levels, namely: high flood risk area, high flood risk area, medium flood risk area, low flood risk area, and low flood risk area, obtaining risk level raster data of the study area, and rasterizing flood risk assessment indicators on the study area;

[0028] Step S32, constructing a weight optimization model of the coupled risk assessment model, collecting N historical typical flood events and inputting them into the weight optimization model, calculating a sampling threshold, and dividing the N historical typical flood events into a general set and a special set according to the sampling threshold, wherein the events within the sampling threshold are the general set, and the events outside the sampling threshold are the special set, and N is a positive integer greater than 100;

[0029] Step S33, using a genetic-local search hybrid algorithm optimized by simulated annealing to cyclically solve the weight optimization model of the coupled risk assessment model to obtain an optimal weight combination scheme, including an optimal weight combination scheme for a set of key flood risk assessment indicators in a general set and an optimal weight combination scheme for a set of key flood risk assessment indicators for each historical typical flood event in a special set.

[0030] According to one aspect of the present application, step S32 is further:

[0031] Step S32a, constructing a weight optimization model of the coupled risk assessment model, wherein the objective function is: the output result of the risk assessment model is closest to the historical actual situation;

[0032] Step S32b, collecting N typical historical flood events, and sequentially sorting the flood risk assessment indicator grids of each typical historical flood event in a set order to obtain a spatial feature sequence of each key flood risk assessment indicator at each time point;

[0033] Step S32c, counting the sequences of all key flood risk assessment indicators of each historical typical flood event at each time point, obtaining the multidimensional space curve corresponding to each historical typical flood event at each time point, and taking the average value as the multidimensional space curve of each historical typical flood event;

[0034] Step S32d, calculate the Hausdorff distance between the multidimensional space curves of each historical typical flood event in turn, and calculate the mean of the Hausdorff distances between the multidimensional space curves of all historical typical flood events, the mean is the sampling threshold, and the historical typical flood events whose Hausdorff distances between the multidimensional space curves of other historical typical flood events are greater than the sampling threshold are regarded as special historical typical flood events and classified into a special set, and other historical typical flood events are classified into a general set.

[0035] According to one aspect of the present application, step S33 is further:

[0036] Step S33a, generating an initial population, calculating the fitness value of each individual in the population in turn, selecting individuals with high fitness values ​​as parents to provide genes for the next generation;

[0037] Step S33b, using simulated annealing to determine the range of the parent neighborhood, and generating new individuals in the parent neighborhood, wherein the new individuals are new weight combination schemes of the key flood risk assessment indicator set, and random mutation is used to update the genes of the new individuals;

[0038] Step S33c, configure the risk assessment model using the new weight combination scheme of the key flood risk assessment indicator set, run the configured risk assessment model, obtain the risk zoning of the study area, that is, the results corresponding to the new individuals, compare the results of the new individuals with the risk level grid data of the study area, and eliminate individuals with low fitness;

[0039] Step S33d, repeat steps S33a-S33c until the temperature drops to the threshold and the iteration is stopped, and the optimal weight combination scheme of the key flood risk assessment indicator set for each historical typical flood event is obtained, including: the optimal weight combination scheme of the key flood risk assessment indicator set of the general set and the key flood risk assessment indicator set of each historical typical flood event in the special set.

[0040] According to one aspect of the present application, step S4 is further:

[0041] Step S41, when working, obtain the scenario at each time point in the current study area, calculate the Hausdorff distance of the multidimensional space curve between each typical flood event scenario in the general set in turn, and use the maximum distance as the dynamic threshold of spatiotemporal similarity of historical typical flood events;

[0042] Step S42: Input the scenario at each time point in the study area into the risk assessment model, determine the similarity between the scenario at each time point in the study area and the scenario of each historical typical flood event in the special set based on the dynamic threshold of the spatiotemporal similarity of historical typical flood events, select the optimal weight combination scheme of the key flood risk assessment indicator set, and calculate the dynamic zoning of flood risk in the study area.

[0043] According to one aspect of the present application, step S42 is further:

[0044] Step S42a, sequentially sorting the flood risk assessment index grids of each historical typical flood event in the general set and the special set according to a set order, to obtain a spatial feature sequence of each key flood risk assessment index at each time point;

[0045] Step S42b, counting the sequences of all key flood risk assessment indicators of each historical typical flood event at each time point, obtaining the multidimensional space curve corresponding to each historical typical flood event at each time point, and taking the average value as the multidimensional space curve of each historical typical flood event;

[0046] Step S42c, calculating the Hausdorff distance between the multidimensional space curves of each historical typical flood event in turn, and taking the largest distance as the spatiotemporal similarity;

[0047] Step S42d, sequentially calculating the Hausdorff distance of the multidimensional space curve between each typical flood event in the general set, and taking the maximum distance as the dynamic threshold of the spatiotemporal similarity of the historical typical flood events.

[0048] According to another aspect of the present application, a flood risk dynamic zoning system based on multiple factor optimization is provided, comprising:

[0049] at least one processor; and

[0050] a memory communicatively connected to at least one of the processors; wherein,

[0051] 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 dynamic flood risk zoning method based on multiple factor optimization described in any of the above technical solutions.

[0052] Beneficial effects: A dynamic flood risk zoning method based on multiple factor optimization is adopted, and multiple factors affecting flood risk are integrated through statistical and computational models, which overcomes the limitations of single factor analysis, improves the comprehensiveness and accuracy of flood risk assessment, realizes real-time monitoring and assessment of flood risk levels, provides timely warnings and response plans, and uses data mining and intelligent algorithms to provide technical support for complex factor integration and optimization, thereby improving the scientificity and operability of overall decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 It is a flow chart of the present invention.

[0054] Figure 2 It is a flow chart of step S1 of the present invention.

[0055] Figure 3 It is a flow chart of step S2 of the present invention.

[0056] Figure 4 It is a flow chart of step S3 of the present invention.

[0057] Figure 5 It is a flow chart of step S4 of the present invention. DETAILED DESCRIPTION

[0058] like Figure 1 As shown, the following technical solution is proposed. According to one aspect of the present application, a method for dynamic zoning of flood risks based on multiple factor optimization is provided, comprising the following steps:

[0059] Step S1, obtaining and retrieving flood risk assessment literature, using at least two meta-analysis methods to fuse and identify flood risk assessment indicators, forming an initial flood risk assessment indicator set and clustering, and extracting classification keywords for each type of flood risk assessment indicator; based on the classification keywords, using at least two meta-analysis methods to identify flood risk assessment indicators again, screening to obtain a preliminary screening flood risk assessment indicator set and calculating the frequency value of each flood risk assessment indicator therein;

[0060] Step S2, reading the initial screening flood assessment index set, and calculating the contribution of each flood risk assessment index therein, calculating the comprehensive index value of each flood risk assessment index in turn, further calculating the mean of the comprehensive evaluation value of the flood risk assessment index, and then calculating the comprehensive discrimination index value, and screening the flood risk assessment index again based on the comprehensive discrimination index value of the flood risk assessment index to obtain the key flood risk assessment index set;

[0061] Step S3, extracting water level and flow data of the study area, dividing the study area into different levels, obtaining risk level raster data of the study area, rasterizing key flood risk assessment indicators on the study area, and using them as inputs of the pre-built key flood risk assessment indicator weight optimization model and risk assessment model, solving them using a genetic-local search hybrid algorithm optimized by simulated annealing, and obtaining the optimal weight combination scheme of the key flood risk assessment indicator set;

[0062] Step S4: During operation, the similarity between the scenario at each time point in the study area and the scenario of typical historical flood events is calculated, and the optimal weight combination scheme of the key flood risk assessment indicator set is selected based on the similarity, and the scheme is input into the risk assessment model for calculation to obtain the dynamic zoning of flood risk in the study area.

[0063] like Figure 2 As shown, according to one aspect of the present application, the step S1 is further:

[0064] Step S11, obtaining and retrieving flood risk assessment literature, using basic meta-analysis, network meta-analysis, multivariate meta-analysis and Bayesian meta-analysis in turn to identify four sets of flood risk assessment indicator sets, integrating the four sets of flood risk assessment indicator sets to obtain an initial flood risk assessment indicator set and counting the number of occurrences of each flood risk assessment indicator in the initial flood risk assessment indicator set;

[0065] In this embodiment, a multi-method forward and reverse double preliminary selection of flood risk assessment indicators is adopted. First, four meta-analysis methods, namely meta-analysis, network meta-analysis, multivariate meta-analysis and Bayesian meta-analysis, are used to identify flood literature in turn, and flood indicator sets are obtained respectively. Among them, meta-analysis finds a total of a indicators, network meta-analysis finds a total of b indicators, multivariate meta-analysis finds a total of c indicators, and Bayesian meta-analysis finds a total of d indicators. For a certain indicator, it appears aa, bb, cc, and dd times respectively, then the number of times the indicator appears is (aa+bb+cc+dd) / (a+b+c+d).

[0066] In this embodiment, four different meta-analysis methods are used simultaneously to complement each other and optimize indicator screening according to different research designs and data characteristics. This comprehensiveness can reduce the bias caused by a single method and provide more comprehensive analysis results. At the same time, it facilitates the handling of heterogeneity problems. Through random effects, the results from different studies can be better integrated. Through meta-analysis, data on flood risk assessment indicators can be summarized from multiple studies on a global or regional scale, thereby improving statistical power and the scientificity and effectiveness of subsequent decision-making.

[0067] Step S12: divide each flood risk assessment indicator in the initial flood risk assessment indicator set into a number of words, calculate the TF-IDF value of each word in turn, use the words with TF-IDF values ​​less than the threshold as the keywords of the current flood risk assessment indicator, and divide the initial flood risk assessment indicator set into M categories based on the keywords using a hierarchical clustering method, each category has its own classification keyword, and M is a positive integer greater than 2;

[0068] TF-IDF is a weighted technology commonly used in text mining and information retrieval. It aims to evaluate the importance of a word in a document collection or corpus. It is composed of word frequency and inverse document frequency. In this embodiment, the TF-IDF value of the words of each indicator is calculated and the keywords of the flood risk assessment indicators are identified based on the TF-IDF value. Then, based on the keywords, the hierarchical clustering method is used to classify flood indicators with similar keywords into the same category, thereby obtaining a classification of flood indicators with keywords.

[0069] In a certain embodiment, specifically:

[0070] Given three flood risk assessment indicators, they are: Indicator 1: total, rainfall, amount; Indicator 2: area, average, rainfall, amount; Indicator 3: time period, rainfall, intensity;

[0071] Calculate the TF values ​​separately, where the quantity appears once in indicator 1, the total number of words is 3, TF=1 / 3≈0.33; the quantity appears once in indicator 2, the total number of words is 4, TF=1 / 4=0.25; the time period appears once in indicator 3, the total number of words is 3, TF=1 / 3≈0.33; calculate the TF values ​​of the 3 words in indicator 1, the TF values ​​of the 4 words in indicator 2, and the TF values ​​of the 3 words in indicator 3 in turn;

[0072] Calculate the IDF value. If the quantity appears in two indicators, IDF=log(3 / 2) ≈0.176; if the time period appears in one indicator, IDF=log(3 / 1) ≈0.477. Calculate the IDF values ​​of the three words in indicator 1, the four words in indicator 2, and the three words in indicator 3 in sequence.

[0073] Calculate the TF-IDF value. The TF-IDF value of the quantity in indicator 1 is 0.33×0.176=0.05808. The TF-IDF value of the quantity in indicator 2 is 0.25×0.176=0.044. Calculate the TF-IDF values ​​of all words in turn. Based on the TF-IDF value of each word, the keywords of indicator 1 are: rainfall, quantity; the keywords of indicator 2 are: rainfall, quantity; the keywords of indicator 3 are: rainfall.

[0074] Step S13: re-identify flood risk assessment indicators based on the M-category classification keywords, screen out new flood risk assessment indicators and calculate the frequency values ​​of the new flood risk assessment indicators, construct a flood risk assessment indicator set and calculate the frequency value of each indicator in the initial screening flood risk assessment indicator set.

[0075] According to one aspect of the present application, step S13 is further:

[0076] Step S13a, using M-class classification keywords to sequentially use basic meta-analysis, network meta-analysis, multivariate meta-analysis and Bayesian meta-analysis to identify flood risk assessment indicators in flood risk assessment literature, screen out newly emerged flood risk assessment indicators and calculate the frequency values ​​of newly emerged flood risk assessment indicators;

[0077] Step S13b, calculating the frequency value of each flood risk assessment indicator and sorting them by size, and selecting the flood risk assessment indicator with the smallest frequency value;

[0078] Step S13c, comparing the frequency values ​​of the newly emerged flood risk assessment indicators with the flood risk assessment indicators with the minimum frequency value, and classifying the newly emerged flood risk assessment indicators that are not less than the minimum frequency value of the flood risk assessment indicators in the initial flood risk assessment indicator set into the initial flood risk assessment indicator set;

[0079] Step S13d, repeating steps S13a to S13c until no new indicators can be classified into the initial flood risk assessment indicator set, stopping iteration, and obtaining the flood risk assessment indicator set and the frequency value of each indicator in the flood risk assessment indicator set.

[0080] After completing the double screening of forward and reverse screening and clustering, forward screening is performed again with the keywords of various indicators as search conditions to obtain new flood risk assessment indicators. By comparing with the initial flood risk assessment indicator set obtained for the first time, different flood risk assessment indicators are selected, and the frequency values ​​of flood risk assessment indicators are obtained by calculating the number of times the flood risk assessment indicators appear. The frequency value of the flood risk assessment indicator with the lowest frequency value in the initial flood risk assessment indicator set is compared with the frequency value of the newly emerging flood risk assessment indicator, and the newly emerging flood risk assessment indicators are screened. New indicators with frequency values ​​less than the minimum value are deleted, and new indicators greater than or equal to the minimum value are added to the initial flood risk assessment indicator set to obtain the flood risk assessment indicator set, and the frequency values ​​of all flood risk assessment indicators in the flood risk assessment indicator set are obtained.

[0081] like Figure 3 As shown, according to one aspect of the present application, the step S2 is further:

[0082] Step S21, reading the initial screening flood assessment index set, and calculating the contribution of each index in the initial screening flood risk assessment index set in sequence;

[0083] Step S22, calculate the comprehensive index value of each flood risk assessment indicator in turn, calculate the product of the frequency value and the contribution of each indicator in the flood risk assessment indicator set in turn as the comprehensive index value of the flood risk assessment indicator, and calculate the mean of the comprehensive index values ​​of the flood risk assessment indicators, calculate the comprehensive discrimination index value of each flood risk assessment indicator based on the comprehensive index value of the flood risk assessment indicator and the mean of the comprehensive index value of the flood risk assessment indicator, and according to a preset threshold, remove the flood risk assessment indicators whose comprehensive discrimination index values ​​are less than the threshold to obtain a key flood risk assessment indicator set.

[0084] In this embodiment, the comprehensive value of the indicator, that is, the importance of the indicator, is represented by the product of the frequency value and the contribution degree of the indicator. When the frequency value and the contribution degree are both large, the comprehensive value of the indicator is also large; when one of the frequency value and the contribution degree is large and the other is small, the comprehensive value of the indicator is moderate; when the frequency value and the contribution degree are both small, the comprehensive value of the indicator is small. Therefore, in this embodiment, whether the indicator is deleted is determined according to the size of the comprehensive value of each indicator. When the comprehensive value of a certain indicator is more than one order of magnitude smaller than the average of the comprehensive values ​​of all indicators, it is considered that the importance of the indicator is significantly lower than the average level of all indicators, and the indicator can be deleted. Therefore, in this embodiment, the comprehensive discrimination index is defined as P j , P j =lg(F j / F* j ), where F j is the comprehensive value of a certain indicator, F* j is the mean of the comprehensive values ​​of all indicators;

[0085] Comprehensive judgment index P j It reflects the difference in magnitude between the comprehensive value of the indicator and the mean of the comprehensive value of all indicators. When the comprehensive discrimination index is greater than 0, it indicates that the comprehensive value of the indicator is greater than the average level; when the comprehensive discrimination index is less than 0, it indicates that the comprehensive value of the indicator is less than the average level; when the comprehensive discrimination index is less than or equal to -1, it indicates that the comprehensive value of the indicator differs from the mean of the comprehensive value of all indicators by more than one order of magnitude, which is significantly lower than the average level. It can be deleted. After deleting all indicators whose comprehensive values ​​differ from the mean of the comprehensive values ​​of all indicators by more than one order of magnitude, the key flood assessment indicator set is obtained.

[0086] According to one aspect of the present application, step S21 is further:

[0087] Step S21a, reading the initial screening flood assessment index set and the typical flood event data of the study area in history, obtaining the measured data corresponding to each indicator in the initial screening flood risk assessment index set from the typical flood event data, and calculating the historical mean of each flood risk assessment indicator in turn as the first assessment result of the flood risk assessment indicator;

[0088] Step S21b, sequentially selecting a flood risk assessment indicator and keeping other flood risk assessment indicators unchanged, calculating the standard deviation of the selected flood risk assessment indicator, taking the sum of the standard deviation of the selected flood risk assessment indicator and the historical mean of the selected flood risk assessment indicator as a new indicator value and evaluating the risk, to obtain a second evaluation result of the selected flood risk assessment indicator;

[0089] Step S21c, sequentially calculating the contribution of each indicator in the initial screening flood assessment indicator set, wherein the contribution is the ratio of the difference between the first assessment result and the second assessment result of the flood risk assessment indicator to the historical mean.

[0090] When analyzing the contribution of flood risk assessment indicators, although some indicators are not so sensitive, their own variation range is particularly large, resulting in a greater impact on the results than sensitive indicators. Therefore, in this embodiment, the contribution degree is used to screen the sensitivity of the indicators.

[0091] like Figure 4 As shown, according to one aspect of the present application, step S3 is further:

[0092] Step S31, extracting water level and flow data of the study area based on historical typical flood event data of the study area, dividing the study area into five levels, namely: high flood risk area, high flood risk area, medium flood risk area, low flood risk area, and low flood risk area, obtaining risk level raster data of the study area, and rasterizing flood risk assessment indicators on the study area;

[0093] In this embodiment, since it is known whether the study area has had flood risks in the past, that is, whether it has occurred or not, the risk value cannot be obtained. Therefore, a grade classification method is adopted to subdivide the study area into different risk levels through water level and flow data, which is helpful to identify high-risk areas. At the same time, through rasterization processing, water level and flow data can be effectively integrated with other geographic information for subsequent spatial analysis. The obtained raster data can be visualized and the flood risk distribution can be intuitively presented. When the subsequent dynamic zoning of flood risks is carried out, as new flood events and environmental changes occur, the raster data can be dynamically updated to enhance the real-time and accuracy of flood risk assessment.

[0094] Step S32, constructing a weight optimization model of the coupled risk assessment model, collecting N historical typical flood events and inputting them into the weight optimization model, calculating a sampling threshold, and dividing the N historical typical flood events into a general set and a special set according to the sampling threshold, wherein the events within the sampling threshold are the general set, and the events outside the sampling threshold are the special set, and N is a positive integer greater than 100;

[0095] Typical historical flood events are relatively extreme, so some factors of a certain event may be extremely accidental, so it is impossible to take all of them into account. In a certain embodiment, 100 typical historical flood events are extracted, and the global 100 events are set to be optimal. By calculating the optimal parameter combination, a global similarity of 70% is finally obtained. When 90 of them are set to be optimal, a similarity of 90% is obtained for the 90 events. Therefore, in this embodiment, a weight optimization model is first constructed, and the sampling threshold is obtained by the weight optimization model. Then, the sampling threshold is used to screen the historical typical flood events for optimal parameter calculation, which can obtain a more reasonable optimization solution.

[0096] Therefore, in this embodiment, firstly, the multidimensional spatial curves of N collected historical typical flood events are obtained by calculation, and the Hausdorff distances between the multidimensional spatial curves of the historical typical flood events are calculated to judge the similarity between each other, and the mean of the Hausdorff distances is calculated as the sampling threshold, and by comparing the relationship between the Hausdorff distances of a certain historical typical flood event and other historical typical flood events and the threshold, all events with distances greater than the threshold are screened as special historical typical flood events, which are most advantageous for global optimization calculation after screening, and the special historical typical flood events that are screened out are classified into a special set, and the rest are a general set, which are subsequently used to calculate the optimal weight combination scheme of the key flood risk assessment indicator set for the general scenario, and each historical typical flood event in the special set needs to calculate the optimal weight combination scheme of the corresponding key flood risk assessment indicator set.

[0097] Step S33, using a genetic-local search hybrid algorithm optimized by simulated annealing to cyclically solve the weight optimization model of the coupled risk assessment model to obtain an optimal weight combination scheme, including an optimal weight combination scheme for a set of key flood risk assessment indicators in a general set and an optimal weight combination scheme for a set of key flood risk assessment indicators for each historical typical flood event in a special set.

[0098] According to one aspect of the present application, step S32 is further:

[0099] Step S32a, constructing a weight optimization model of the coupled risk assessment model, wherein the objective function is: the output result of the risk assessment model is closest to the historical actual situation;

[0100] In a certain embodiment, specifically:

[0101] A weight optimization model is constructed with the minimum cumulative error between the risk assessment model results and the historical actual situation as the objective function;

[0102] In another embodiment, specifically:

[0103] A weight optimization model is constructed with the objective function of minimizing the maximum error between the risk assessment model results and the historical actual situation.

[0104] Step S32b, collecting N typical historical flood events, and sequentially sorting the flood risk assessment indicator grids of each typical historical flood event in a set order to obtain a spatial feature sequence of each key flood risk assessment indicator at each time point;

[0105] Step S32c, counting the sequences of all key flood risk assessment indicators of each historical typical flood event at each time point, obtaining the multidimensional space curve corresponding to each historical typical flood event at each time point, and taking the average value as the multidimensional space curve of each historical typical flood event;

[0106] Step S32d, calculate the Hausdorff distance between the multidimensional space curves of each historical typical flood event in turn, and calculate the mean of the Hausdorff distances between the multidimensional space curves of all historical typical flood events, the mean is the sampling threshold, and the historical typical flood events whose Hausdorff distances between the multidimensional space curves of other historical typical flood events are greater than the sampling threshold are regarded as special historical typical flood events and classified into a special set, and other historical typical flood events are classified into a general set.

[0107] The Hausdorff distance has significant advantages in evaluating curve similarity, especially when measuring the similarity of irregular shapes and complex paths. The Hausdorff distance can take into account the distance between the farthest points of the two curves, thereby effectively capturing the overall difference of the curves. In this embodiment, the dynamic threshold of spatiotemporal similarity can be better calculated;

[0108] On the other hand, since the Hausdorff distance uses the farthest point criterion, it can resist the influence of small noises. Even if there are small disturbances and changes in the curve, the Hausdorff distance can remain stable.

[0109] Hausdorff distance provides a global perspective and good robustness in curve similarity assessment, and can effectively identify the similarity between two curves, providing a reliable basis for decision makers;

[0110] In a certain embodiment, specifically:

[0111] For a study area with 10,000 grids, the key flood risk assessment indicators are screened as cumulative rainfall, rainfall intensity, soil moisture, and population density. For two typical historical flood events, each time point corresponds to a 4-dimensional spatial curve, recorded as curves A and B. The curve is a point set consisting of 10,000 points, and each point is a grid in the study area;

[0112] For each point in curve A, calculate the minimum distance to all points in curve B and take the maximum value;

[0113] For each point in curve B, calculate the minimum distance to all points in curve A and take the maximum value;

[0114] Calculate the final Hausdorff distance and determine the similarity between curve A and curve B based on the Hausdorff distance.

[0115] According to one aspect of the present application, step S33 is further:

[0116] Step S33a, generating an initial population, calculating the fitness value of each individual in the population in turn, selecting individuals with high fitness values ​​as parents to provide genes for the next generation;

[0117] Step S33b, using simulated annealing to determine the range of the parent neighborhood, and generating new individuals in the parent neighborhood, wherein the new individuals are new weight combination schemes of the key flood risk assessment indicator set, and random mutation is used to update the genes of the new individuals;

[0118] Step S33c, configure the risk assessment model using the new weight combination scheme of the key flood risk assessment indicator set, run the configured risk assessment model, obtain the risk zoning of the study area, that is, the results corresponding to the new individuals, compare the results of the new individuals with the risk level grid data of the study area, and eliminate individuals with low fitness;

[0119] Step S33d, repeat steps S33a-S33c until the temperature drops to the threshold and the iteration is stopped, and the optimal weight combination scheme of the key flood risk assessment indicator set for each historical typical flood event is obtained, including: the optimal weight combination scheme of the key flood risk assessment indicator set of the general set and the key flood risk assessment indicator set of each historical typical flood event in the special set.

[0120] The genetic-local search hybrid algorithm is an optimization algorithm that combines a genetic algorithm with a local search, and utilizes the global search capability of the genetic algorithm and the refined search capability of the local search algorithm to improve the efficiency of solving the optimization problem and the quality of the solution; the genetic algorithm can explore multiple areas of the solution space, while the local search can refine and improve the quality of the solution. Compared with a single algorithm, the hybrid method can converge faster, and the local search can improve the quality of the final solution and achieve a better solution. However, the introduction of local search makes the algorithm design and implementation more complicated, and the local search increases the computational cost. Therefore, in this embodiment, simulated annealing is introduced to optimize the genetic-local search hybrid algorithm to enhance the flexibility and depth of the local search.

[0121] In a certain embodiment, specifically:

[0122] Five flood risk assessment indicators are set, namely: rainfall, terrain slope, soil permeability, watershed area and population density;

[0123] The goal is to ensure that the sum of the weights of all indicators is 1 and to maximize the weighted risk value;

[0124] Define a flood risk assessment model and output a weighted risk value;

[0125] Set weight constraints;

[0126] Randomly generate an initial weight combination, calculate the risk value based on the current weight combination, and calculate a new risk value by changing the weights through cross-mutation;

[0127] When the new risk value is greater than the initial risk value, the new combination is accepted. If not, the iteration continues and the temperature is gradually reduced with the iteration until the temperature drops to the threshold and the iteration stops.

[0128] In each iteration, a local search is applied to the current best solution to find the best local solution;

[0129] Get the optimal weight combination and corresponding risk value.

[0130] like Figure 5 As shown, according to one aspect of the present application, step S4 is further:

[0131] Step S41, when working, obtain the scenario at each time point in the current study area, calculate the Hausdorff distance of the multidimensional space curve between each typical flood event scenario in the general set in turn, and use the maximum distance as the dynamic threshold of spatiotemporal similarity of historical typical flood events;

[0132] Step S42: Input the scenario at each time point in the study area into the risk assessment model, determine the similarity between the scenario at each time point in the study area and the scenario of each historical typical flood event in the special set based on the dynamic threshold of the spatiotemporal similarity of historical typical flood events, select the optimal weight combination scheme of the key flood risk assessment indicator set, and calculate the dynamic zoning of flood risk in the study area.

[0133] In this embodiment, since the scenario of the study area will change over time, when calculating the dynamic zoning of flood risk in the study area, the Hausdorff distance between the multidimensional spatial curve of each historical typical flood event in the special set is first calculated according to the current scenario of the study area, and the relationship between the calculated Hausdorff distance and the dynamic threshold of spatiotemporal similarity of the general set is determined. If the calculated Hausdorff distance is less than the dynamic threshold of spatiotemporal similarity, it means that the current scenario of the study area is similar to the historical typical flood events in the special set. Therefore, the optimal weight combination scheme of the key flood risk assessment indicator set corresponding to the similar historical typical flood events in the special set is selected to calculate the current flood risk zoning of the study area. When all the calculated Hausdorff distances are greater than the dynamic threshold of spatiotemporal similarity, it means that the current scenario of the study area is not similar to all the historical typical flood events in the special set, then the current scenario of the study area belongs to the scenario of the general set, and the optimal weight combination scheme of the key flood risk assessment indicator set of the general set is selected to calculate the current flood risk zoning of the study area, and the dynamic zoning of flood risk in the study area is calculated according to the dynamic scenario of the study area.

[0134] According to another aspect of the present application, a flood risk dynamic zoning system based on multiple factor optimization is provided, comprising:

[0135] at least one processor; and

[0136] a memory communicatively connected to at least one of the processors; wherein,

[0137] The memory stores instructions that can be executed by the processor, and the instructions are used to be executed by the processor to implement any of the above-mentioned dynamic flood risk zoning methods based on multiple factor optimization.

[0138] The preferred embodiments of the present invention are described in detail above; however, the present invention is not limited to the specific details in the above embodiments. Within the technical concept of the present invention, various equivalent transformations can be made to the technical solutions of the present invention, and these equivalent transformations all belong to the protection scope of the present invention.

Claims

1. A dynamic flood risk zoning method based on multiple factor optimization, characterized by: The steps include: Step S1, obtaining and retrieving flood risk assessment literature, using at least two meta-analysis methods to fuse and identify flood risk assessment indicators, forming an initial flood risk assessment indicator set and clustering, and extracting classification keywords for each type of flood risk assessment indicator; based on the classification keywords, using at least two meta-analysis methods to identify flood risk assessment indicators again, screening to obtain a preliminary screening flood risk assessment indicator set and calculating the frequency value of each flood risk assessment indicator therein; Step S2, reading the initial screening flood assessment index set, and calculating the contribution of each flood risk assessment index therein, calculating the comprehensive index value of each flood risk assessment index in turn, further calculating the mean of the comprehensive evaluation value of the flood risk assessment index, and then calculating the comprehensive discrimination index value, and screening the flood risk assessment index again based on the comprehensive discrimination index value of the flood risk assessment index to obtain the key flood risk assessment index set; Step S3, extracting water level and flow data of the study area, dividing the study area into different levels, obtaining risk level raster data of the study area, rasterizing key flood risk assessment indicators on the study area, and using them as inputs of the pre-built key flood risk assessment indicator weight optimization model and risk assessment model, solving them using a genetic-local search hybrid algorithm optimized by simulated annealing, and obtaining the optimal weight combination scheme of the key flood risk assessment indicator set; Step S4: during operation, the similarity between the scenario at each time point in the study area and the scenario of a typical historical flood event is calculated, and based on the similarity, the optimal weight combination scheme of the key flood risk assessment indicator set is selected, and the scheme is input into the risk assessment model for calculation to obtain the dynamic zoning of flood risk in the study area; The step S1 is further as follows: Step S11, obtaining and retrieving flood risk assessment literature, using basic meta-analysis, network meta-analysis, multivariate meta-analysis and Bayesian meta-analysis in turn to identify four sets of flood risk assessment indicator sets, integrating the four sets of flood risk assessment indicator sets to obtain an initial flood risk assessment indicator set and counting the number of occurrences of each flood risk assessment indicator in the initial flood risk assessment indicator set; Step S12: divide each flood risk assessment indicator in the initial flood risk assessment indicator set into a number of words, calculate the TF-IDF value of each word in turn, use the words with TF-IDF values ​​less than the threshold as the keywords of the current flood risk assessment indicator, and divide the initial flood risk assessment indicator set into M categories based on the keywords using a hierarchical clustering method, each category has its own classification keyword, and M is a positive integer greater than 2; Step S13: re-identify flood risk assessment indicators based on the M-class classification keywords, screen out new flood risk assessment indicators and calculate the frequency values ​​of the new flood risk assessment indicators, construct a flood risk assessment indicator set and calculate the frequency value of each indicator in the initial screening flood risk assessment indicator set; The step S4 is further as follows: Step S41, when working, obtain the scenario at each time point in the current study area, calculate the Hausdorff distance of the multidimensional space curve between each typical flood event scenario in the general set in turn, and use the maximum distance as the dynamic threshold of spatiotemporal similarity of historical typical flood events; Step S42: Input the scenario at each time point in the study area into the risk assessment model, determine the similarity between the scenario at each time point in the study area and the scenario of each historical typical flood event in the special set based on the dynamic threshold of the spatiotemporal similarity of historical typical flood events, select the optimal weight combination scheme of the key flood risk assessment indicator set, and calculate the dynamic zoning of flood risk in the study area.

2. The method for dynamic flood risk zoning based on multiple factor optimization according to claim 1, characterized in that: The step S13 is further as follows: Step S13a, using M-class classification keywords to sequentially use basic meta-analysis, network meta-analysis, multivariate meta-analysis and Bayesian meta-analysis to identify flood risk assessment indicators in flood risk assessment literature, screen out newly emerged flood risk assessment indicators and calculate the frequency values ​​of newly emerged flood risk assessment indicators; Step S13b, calculating the frequency value of each flood risk assessment indicator and sorting them by size, and selecting the flood risk assessment indicator with the smallest frequency value; Step S13c, comparing the frequency values ​​of the newly emerged flood risk assessment indicators with the flood risk assessment indicators with the minimum frequency value, and classifying the newly emerged flood risk assessment indicators that are not less than the minimum frequency value of the flood risk assessment indicators in the initial flood risk assessment indicator set into the initial flood risk assessment indicator set; Step S13d, repeating steps S13a to S13c until no new indicators can be classified into the initial flood risk assessment indicator set, stopping iteration, and obtaining the flood risk assessment indicator set and the frequency value of each indicator in the flood risk assessment indicator set.

3. The method for dynamic flood risk zoning based on multiple factor optimization according to claim 1, characterized in that: The step S2 is further as follows: Step S21, reading the initial screening flood assessment index set, and calculating the contribution of each index in the initial screening flood risk assessment index set in sequence; Step S22, calculate the comprehensive index value of each flood risk assessment indicator in turn, calculate the product of the frequency value and the contribution of each indicator in the flood risk assessment indicator set in turn as the comprehensive index value of the flood risk assessment indicator, and calculate the mean of the comprehensive index values ​​of the flood risk assessment indicators, calculate the comprehensive discrimination index value of each flood risk assessment indicator based on the comprehensive index value of the flood risk assessment indicator and the mean of the comprehensive index value of the flood risk assessment indicator, and according to a preset threshold, remove the flood risk assessment indicators whose comprehensive discrimination index values ​​are less than the threshold to obtain a key flood risk assessment indicator set.

4. The method for dynamic flood risk zoning based on multiple factor optimization according to claim 3, characterized in that: The step S21 is further as follows: Step S21a, reading the initial screening flood assessment index set and the typical flood event data of the study area in history, obtaining the measured data corresponding to each indicator in the initial screening flood risk assessment index set from the typical flood event data, and calculating the historical mean of each flood risk assessment indicator in turn as the first assessment result of the flood risk assessment indicator; Step S21b, sequentially selecting a flood risk assessment indicator and keeping other flood risk assessment indicators unchanged, calculating the standard deviation of the selected flood risk assessment indicator, taking the sum of the standard deviation of the selected flood risk assessment indicator and the historical mean of the selected flood risk assessment indicator as a new indicator value and evaluating the risk, to obtain a second evaluation result of the selected flood risk assessment indicator; Step S21c, sequentially calculating the contribution of each indicator in the initial screening flood assessment indicator set, wherein the contribution is the ratio of the difference between the first assessment result and the second assessment result of the flood risk assessment indicator to the historical mean.

5. The method for dynamic flood risk zoning based on multiple factor optimization according to claim 1, characterized in that: The step S3 is further as follows: Step S31, extracting water level and flow data of the study area based on historical typical flood event data of the study area, dividing the study area into five levels, namely: high flood risk area, high flood risk area, medium flood risk area, low flood risk area, and low flood risk area, obtaining risk level raster data of the study area, and rasterizing flood risk assessment indicators on the study area; Step S32, constructing a weight optimization model of the coupled risk assessment model, collecting N historical typical flood events and inputting them into the weight optimization model, calculating a sampling threshold, and dividing the N historical typical flood events into a general set and a special set according to the sampling threshold, wherein the events within the sampling threshold are the general set, and the events outside the sampling threshold are the special set, and N is a positive integer greater than 100; Step S33, using a genetic-local search hybrid algorithm optimized by simulated annealing to cyclically solve the weight optimization model of the coupled risk assessment model to obtain an optimal weight combination scheme, including an optimal weight combination scheme for a set of key flood risk assessment indicators in a general set and an optimal weight combination scheme for a set of key flood risk assessment indicators for each historical typical flood event in a special set.

6. The method for dynamic flood risk zoning based on multiple factor optimization according to claim 5, characterized in that: The step S32 is further as follows: Step S32a, constructing a weight optimization model of the coupled risk assessment model, wherein the objective function is: the output result of the risk assessment model is closest to the historical actual situation; Step S32b, collecting N typical historical flood events, and sequentially sorting the flood risk assessment indicator grids of each typical historical flood event in a set order to obtain a spatial feature sequence of each key flood risk assessment indicator at each time point; Step S32c, counting the sequences of all key flood risk assessment indicators of each historical typical flood event at each time point, obtaining the multidimensional space curve corresponding to each historical typical flood event at each time point, and taking the average value as the multidimensional space curve of each historical typical flood event; Step S32d, calculate the Hausdorff distance between the multidimensional space curves of each historical typical flood event in turn, and calculate the mean of the Hausdorff distances between the multidimensional space curves of all historical typical flood events, the mean is the sampling threshold, and the historical typical flood events whose Hausdorff distances between the multidimensional space curves of other historical typical flood events are greater than the sampling threshold are regarded as special historical typical flood events and classified into a special set, and other historical typical flood events are classified into a general set.

7. The method for dynamic flood risk zoning based on multiple factor optimization according to claim 5, characterized in that: The step S33 is further as follows: Step S33a, generating an initial population, calculating the fitness value of each individual in the population in turn, selecting individuals with high fitness values ​​as parents to provide genes for the next generation; Step S33b, using simulated annealing to determine the range of the parent neighborhood, and generating new individuals in the parent neighborhood, wherein the new individuals are new weight combination schemes of the key flood risk assessment indicator set, and random mutation is used to update the genes of the new individuals; Step S33c, configure the risk assessment model using the new weight combination scheme of the key flood risk assessment indicator set, run the configured risk assessment model, obtain the risk zoning of the study area, that is, the results corresponding to the new individuals, compare the results of the new individuals with the risk level grid data of the study area, and eliminate individuals with low fitness; Step S33d, repeat step S33a-step S33c until the temperature drops to the threshold and the iteration is stopped, and the optimal weight combination scheme of the key flood risk assessment indicator set for each historical typical flood event is obtained, including: the optimal weight combination scheme of the key flood risk assessment indicator set of the general set and the key flood risk assessment indicator set of each historical typical flood event in the special set.

8. A dynamic flood risk zoning system based on multiple factor optimization, characterized by: 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 flood risk dynamic zoning method based on multiple factor optimization as described in any one of claims 1 to 7.

Images