A comprehensive decision-making method for high temperature disaster prevention measures

Abstract

The present invention discloses a comprehensive decision-making method for high temperature disaster prevention measures, which includes determining high temperature risk factors; establishing a mapping relationship between each high temperature disaster prevention measure and the high temperature risk factor; determining a target research area; collecting historical data of high temperature assessment indicators, establishing and training a deep learning model; the deep learning model uses the high temperature assessment index as input and the high temperature risk factor as output; based on the deep learning model, predicting the high temperature risk factor of the target research area within a specified time; establishing a decision matrix including different high temperature disaster prevention measures based on the prediction results; converting the decision matrix into a fuzzy evaluation matrix based on a trapezoidal fuzzy number algorithm; defuzzifying the fuzzy evaluation matrix to obtain the expected value of the normalized attribute value; sorting each high temperature disaster prevention measure, and obtaining the optimal decision based on the mapping relationship between each high temperature disaster prevention measure and the high temperature risk factor. The present invention can solve the problems that the prior art cannot make decisions on disaster prevention measures in advance, and the decision redundancy, low decision efficiency, and limited decision accuracy can achieve the purpose of dynamically and accurately making short-term measures for the upcoming high temperature extreme climate.

Description

Technical Field

[0001] The present invention relates to the field of high temperature disaster prevention, and in particular to a comprehensive decision-making method for high temperature disaster prevention measures. Background Art

[0002] Since different cities are affected by factors such as economy, population, topography, and location, dividing high temperature levels based solely on temperature cannot fully reflect high temperature characteristics and differences in decision-making.

[0003] Decisions on high temperature disaster prevention usually rely on historical experience and are made based on expert judgment and the similarity of historical high temperature cases. However, this traditional decision-making method relies too much on individual decision makers and lacks consideration of objectivity, regional differences, and quantitative decision-making effects. With the development of cities, relying solely on inherent experience may bring more problems, which are specifically reflected in the following aspects: (1) Extreme high temperature events show differences in time and space. It is difficult to fully evaluate the spatial distribution of high temperatures and effectively determine the key areas for high temperature decisions by relying solely on a single meteorological, hydrological or economic indicator. (2) The decision-making process is too subjective and lacks sufficient quantitative support. (3) There are problems such as decision redundancy, low decision efficiency, and inability to evaluate decision benefits in decision-making methods.

[0004] In the prior art, the applicant in this case previously applied for "A method, system, device and medium for generating high temperature disaster prevention measures" (authorization announcement number CN116703036B), which generates high temperature disaster prevention measures by classifying several high temperature assessment factors and calculating high temperature risk indicators. However, this technology can only generate disaster prevention measures after high temperature weather has occurred, and it is impossible to make disaster prevention preparations in advance. In addition, this technology still has shortcomings such as decision-making redundancy, low decision-making efficiency, and limited decision-making accuracy. Summary of the invention

[0005] The present invention provides a comprehensive decision-making method for high-temperature disaster prevention measures to solve the problems of the existing technology being unable to make decisions on disaster prevention measures in advance, and having redundancy in decision-making, low efficiency in decision-making, and limited accuracy in decision-making, so as to achieve the purpose of being able to dynamically and accurately decide on short-term measures for the upcoming high-temperature extreme climate.

[0006] The present invention is achieved through the following technical solutions:

[0007] A comprehensive decision-making method for high temperature disaster prevention measures, comprising:

[0008] Determine high temperature risk factors based on a number of high temperature assessment indicators; collect a number of high temperature disaster prevention measures and establish a mapping relationship between each high temperature disaster prevention measure and the high temperature risk factor;

[0009] Identify target study areas;

[0010] Collect historical data of high temperature assessment indicators, establish and train a deep learning model; the deep learning model takes the high temperature assessment indicators as input and the high temperature risk factors as output;

[0011] Based on the deep learning model, predict the high temperature risk factors in the target study area within a specified time;

[0012] Based on the prediction results, a decision matrix including different high temperature disaster prevention measures is established;

[0013] Based on the trapezoidal fuzzy number algorithm, the decision matrix is ​​converted into a fuzzy evaluation matrix;

[0014] Defuzzify the fuzzy evaluation matrix to obtain the expected value of the normalized attribute value;

[0015] The various high temperature disaster prevention measures are ranked, and the optimal decision is obtained based on the mapping relationship between each high temperature disaster prevention measure and high temperature risk factors.

[0016] In view of the problems in the prior art that disaster prevention measures cannot be decided in advance, and that decisions are redundant, inefficient, and accurate, the present invention proposes a comprehensive decision-making method for high-temperature disaster prevention measures. The method first collects high-temperature assessment indicators and determines the corresponding high-temperature risk factors; then collects different high-temperature disaster prevention measures, and establishes a mapping relationship between high-temperature disaster prevention measures and high-temperature risk factors to provide a basis for quantifying the effects of subsequent measures; the method also collects historical data of high-temperature assessment indicators, and uses the historical data as samples to establish and train a deep learning model for predicting various high-temperature risk factors; at this point, the preparation work of the method is completed. Afterwards, based on the established deep learning model and with the determined target study area as the spatial scope, the value of the high temperature risk factor within the specified time is predicted, and the high temperature risk factor of the target study area within the specified future time is obtained. Based on different high temperature disaster prevention measures, a decision matrix is ​​established under different high temperature disaster prevention measures; then, based on the trapezoidal fuzzy number algorithm, the decision matrix is ​​converted into a fuzzy evaluation matrix, and then the fuzzy evaluation matrix is ​​defuzzified to obtain the expected value of the normalized attribute value, and the expected value is sorted to obtain the sorting results of different high temperature disaster prevention measures, and then the optimal decision is obtained based on the mapping relationship between the high temperature disaster prevention measures prepared in advance and the high temperature risk factors.

[0017] In the present application, the decision matrix is ​​converted into a fuzzy evaluation matrix through a trapezoidal fuzzy number algorithm, that is, trapezoidal fuzzy operations are performed on all data in the decision matrix to obtain corresponding trapezoidal fuzzy numerical values; the present application can use the existing trapezoidal fuzzy number algorithm to perform relevant operations.

[0018] Compared with the prior art, this application uses deep learning technology to quickly and accurately predict the upcoming high temperature extreme climate and output high temperature risk factors. Compared with the prior art that can only generate disaster prevention measures after the high temperature weather has occurred, it solves the problem of decision-making lag and achieves the purpose of making short-term disaster prevention decisions in advance to guide disaster prevention response and reduce the interference of extreme high temperature weather on social and economic activities. In addition, this application proposes a decision-making method different from the prior art, which can effectively achieve the optimal quantitative output of various response measures for high temperature extreme events and significantly improve the accuracy of decision-making; the introduction of trapezoidal fuzzy number algorithm, calculation of expected value after defuzzification and other steps can effectively overcome the temporal and spatial differences of extreme high temperature events, and significantly reduce the subjective interference in the decision-making process, and achieve more sufficient quantitative support for decision-making results.

[0019] Preferably, the high temperature risk factors include hazard factors, vulnerability factors and exposure factors;

[0020] High temperature assessment indicators include meteorological data indicators and social and economic data indicators; among them:

[0021] Meteorological data indicators include any one or more of the following: maximum surface temperature, maximum air temperature, average surface temperature, average air temperature, air pressure, specific humidity, wind speed, and precipitation;

[0022] Socioeconomic data indicators include any one or more of the following: population density, land use type, building type, GDP density.

[0023] Furthermore, the deep learning model includes:

[0024] The ConvLSTM module takes the meteorological data index as input, performs a convolution and then outputs it;

[0025] The VGGNet model takes the socioeconomic data indicators and the output of the ConvLSTM module as input and the high temperature risk factor as output.

[0026] The VGGNet model is an existing deep learning model. The inventors of this case found during the research process that if the existing VGGNet model is directly used to predict high temperature risk factors in this application, there is a defect that the prediction accuracy is significantly weakened. The reason for this is that the traditional VGGNet model cannot reflect the temporal continuity of the meteorological data indicators in the high temperature evaluation index. In order to overcome this problem, this solution improves the existing VGGNet model by adding a ConvLSTM module to the VGGNet model. The meteorological data indicators in the high temperature evaluation index are preprocessed by the ConvLSTM module. The preprocessing method is to perform a convolution through the ConvLSTM module. Then, the meteorological data indicators preprocessed by the ConvLSTM module and the socio-economic data indicators that have not been preprocessed are input into the existing VGGNet model together.

[0027] This solution specifically improves and optimizes the VGGNet model for comprehensive decision-making on high temperature disaster prevention measures, enabling the deep learning model to capture more continuity information from the input meteorological data, significantly improving the prediction accuracy of high temperature risk factors.

[0028] Furthermore, the method of establishing a decision matrix including different high temperature disaster prevention measures includes:

[0029] The predicted high temperature risk factors and the measure cost index are used as matrix factors, and each matrix factor is graded respectively; the measure cost index includes personnel input and workload;

[0030] Based on the expert scoring method, the corresponding level of the predicted high temperature risk factor and the corresponding level of the measure cost index under different high temperature disaster prevention measures are determined;

[0031] The levels of all matrix factors corresponding to each high temperature disaster prevention measure are used as the decision matrix.

[0032] This scheme first determines the matrix factors of the decision matrix. In addition to the high temperature risk factors predicted by the deep learning model, the matrix factors used also include cost indicators such as personnel input and workload; each matrix factor is divided into several levels to facilitate the subsequent quantification process. Then, based on the results of expert scoring, the levels of each matrix factor under different high temperature disaster prevention measures are divided, and then a decision matrix including different high temperature disaster prevention measures is obtained.

[0033] Furthermore, the method of defuzzifying the fuzzy evaluation matrix and obtaining the expected value of the normalized attribute value includes:

[0034] Extract the trapezoidal fuzzy number of each matrix factor under different high temperature disaster prevention measures, and establish a trapezoidal fuzzy number decision matrix;

[0035] Defuzzification is performed on the trapezoidal fuzzy number decision matrix, and the trapezoidal fuzzy number is converted into the expected value;

[0036] Calculate the expected vector for each matrix factor;

[0037] Calculate the normalized attribute values ​​to obtain a matrix of normalized attribute values;

[0038] Get the expected value of the normalized attribute value;

[0039] Based on the prediction results of the deep learning model, the expected value of the normalized attribute value is corrected.

[0040] This solution innovatively establishes an expected value quantification method based on the comprehensive effects of high temperature response measures, which can effectively achieve the optimal quantitative output of various high temperature disaster prevention measures for extreme high temperature events.

[0041] Preferably, in the trapezoidal fuzzy number decision matrix, each high temperature disaster prevention measure and each matrix factor corresponds to a data set, and the data set is expressed as: in, They represent the four trapezoidal fuzzy numbers corresponding to the mth matrix factor under the i-th high temperature disaster prevention measure;

[0042] The expected vector of each matrix factor is expressed as: (γ 1m ,γ 2m ,…γ im ,…γ Am );

[0043] in, i represents the i-th high temperature disaster prevention measure; m represents the m-th matrix factor; A is the total number of high temperature disaster prevention measures;

[0044] The normalized attribute value is calculated by the following formula: Where: represents the normalized attribute value; β i The matrix representing the trapezoidal fuzzy number; max γim Represents the maximum expected value of the mth matrix factor under the i-th high temperature disaster prevention measure.

[0045] In the trapezoidal fuzzy number decision matrix, there are a total of A types of high temperature disaster prevention measures and several matrix factors. Each high temperature disaster prevention measure and each matrix factor corresponds to a data set. The four trapezoidal fuzzy numbers in each data set are used to describe the minimum value, rising part, falling part and maximum value of the fuzziness. This scheme converts the trapezoidal fuzzy number into an expected value through defuzzification. Then, for each matrix factor, an expected vector for different high temperature disaster prevention measures can be extracted; based on the expected vector, the maximum expected vector max can be extracted.γim , and then combined with the defuzzification results, the normalized attribute value corresponding to the matrix factor can be obtained.

[0046] The defuzzification process can be implemented by using an existing defuzzification algorithm.

[0047] In addition, the inventor team of this case also found in the research process that after defuzzification and obtaining the normalized attribute values, the attribute values ​​corresponding to certain measure effects can be used directly, but the attribute values ​​corresponding to certain measure effects deviate from the actual ones, so it is necessary to correct the expected values ​​of the normalized attribute values. Furthermore, the correction method includes:

[0048] Based on the deep learning model, predict the prediction interval of each high temperature risk factor under different high temperature disaster prevention measures;

[0049] For any high temperature risk factor, perform the following judgment:

[0050] If the expected value of the normalized attribute value is within the prediction interval of the high temperature risk factor, no correction is made;

[0051] If the expected value of the normalized attribute value is not within the prediction interval of the high temperature risk factor, the expected value of the normalized attribute value corresponding to the high temperature risk factor is corrected to the threshold of the prediction interval.

[0052] The correction method of this scheme can correct the expected value of the normalized attribute value, which is more conducive to the dynamic selection of preferential, targeted and short-term high temperature disaster prevention measures.

[0053] Furthermore, the method for ranking various high temperature disaster prevention measures includes:

[0054] Establish a sorting matrix with each high temperature disaster prevention measure as a row and the matrix factors as a column;

[0055] Calculate the consistency index and inconsistency index of each high temperature disaster prevention measure;

[0056] Calculate the credibility index of each high temperature disaster prevention measure;

[0057] The credibility index is converted into a net credibility index, and a ranking result is obtained according to the credibility index.

[0058] The traditional sorting method in the prior art will only get a unique sorting result when the high temperature disaster prevention measures remain unchanged; however, in this solution, the expected value of the normalized attribute value is corrected, so the sorting result should also be dynamic. The sorting method proposed in this solution is more suitable for this dynamic judgment process, and can use the prediction results of the deep learning model to dynamically correct the expected value of the normalized attribute value, thereby further improving the final sorting result and the accuracy of the optimal decision.

[0059] Preferably, the formula for calculating the consistency index and the inconsistency index of each high temperature disaster prevention measure is:

[0060]

[0061] Where: C j (i, k) is the consistency index, which means that on the jth matrix factor, a ij Better than a kj degree; D j (i, k) is the inconsistency index, which means that on the jth matrix factor, a ij Worse than a kj degree; ij is the expected value of the normalized attribute value corresponding to the i-th high temperature disaster prevention measure and the j-th matrix factor; a kj is the expected value of the normalized attribute value corresponding to the kth high temperature disaster prevention measure and the jth matrix factor; q j is the indifference threshold; p j is better than the threshold; v j is inferior to the threshold.

[0062] Preferably, the formula for calculating the credibility index of each high temperature disaster prevention measure is:

[0063]

[0064] Where: S(i, k) is a ij Compared to a kj The credibility index; C(i, k) is the overall consistency index;

[0065] The formula for converting the credibility index to the net credibility index is:

[0066]

[0067] Where: Z i is the net credibility index, which represents the degree to which the ith high temperature disaster prevention measure is superior to other high temperature disaster prevention measures; A is the total number of high temperature disaster prevention measures; ∑ i∈A S(i, k) is the sum of all S(i, k); ∑ i∈AS(k,i) is the sum of all S(k,i); S(k,i) is a kj Compared to a ij credibility index.

[0068] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0069] 1. The present invention provides a comprehensive decision-making method for high-temperature disaster prevention measures, which creatively integrates high-temperature risk assessment, deep learning technology and multi-attribute decision-making into the decision-making response of high-temperature disaster prevention measures. Through a complete decision-making process, it solves the problems of subjectivity, decision redundancy and inability to evaluate decision benefits in existing high-temperature decisions.

[0070] 2. The present invention uses deep learning technology to quickly and accurately predict the upcoming high temperature extreme climate in the future and output high temperature risk factors. Compared with the existing technology that can only generate corresponding disaster prevention measures after the high temperature weather has occurred, it solves the problem of decision-making lag and achieves the purpose of making short-term disaster prevention decisions in advance to guide disaster prevention response and reduce the interference of extreme high temperature weather on social and economic activities.

[0071] 3. The present invention can effectively achieve the optimal quantitative output of various response measures for extreme high temperature events, significantly improving the accuracy of decision-making; the introduction of trapezoidal fuzzy number algorithm, calculation of expected value after defuzzification and other steps can effectively overcome the temporal and spatial differences of extreme high temperature events, and significantly reduce the subjective interference in the decision-making process, thereby achieving more sufficient quantitative support for decision-making results.

[0072] 4. The present invention specifically improves and optimizes the VGGNet model for comprehensive decision-making on high temperature disaster prevention measures, so that the deep learning model can capture more continuity information from the input meteorological data, significantly improving the prediction accuracy of high temperature risk factors.

[0073] 5. The present invention can correct the expected value of the attribute value after normalization, which is more conducive to the dynamic selection of preferential, targeted and short-term high-temperature disaster prevention measures; and the proposed sorting method is more suitable for this dynamic judgment process, and can use the prediction results of the deep learning model to dynamically correct the expected value of the attribute value after normalization, thereby further improving the final sorting result and improving the accuracy of the optimal decision. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] The drawings described herein are used to provide a further understanding of the embodiments of the present invention, constitute a part of this application, and do not constitute a limitation of the embodiments of the present invention. In the drawings:

[0075] Figure 1 It is a schematic diagram of a flow chart of a specific embodiment of the present invention;

[0076] Figure 2 It is a schematic diagram of the final measure sorting result in a specific embodiment of the present invention. DETAILED DESCRIPTION

[0077] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with embodiments and drawings. The exemplary embodiments of the present invention and their description are only used to explain the present invention and are not intended to limit the present invention.

[0078] Embodiment 1:

[0079] like Figure 1 A comprehensive decision-making method for high temperature disaster prevention measures shown in the figure includes the following steps:

[0080] S1. Determine high temperature risk factors based on a number of high temperature assessment indicators; collect a number of high temperature disaster prevention measures, and establish a mapping relationship between each high temperature disaster prevention measure and the high temperature risk factor;

[0081] Among them, high temperature risk factors include hazard factors, vulnerability factors and exposure factors;

[0082] High temperature assessment indicators include meteorological data indicators and social and economic data indicators; specifically:

[0083] The meteorological data indicators include any one or more of the following: maximum surface temperature, maximum air temperature, average surface temperature, average air temperature, air pressure, specific humidity, wind speed, and precipitation;

[0084] The socio-economic data indicators include any one or more of the following: population density, land use type, building type, and GDP density.

[0085] In this embodiment, for the hazard factor, the maximum surface temperature, maximum air temperature, average surface temperature, air pressure, wind speed and precipitation data are normalized and then weighted averaged to obtain it; for the vulnerability factor, the vegetation index, water body index, land use type, building type, GDP density and other data are normalized and then weighted averaged to obtain it; for the exposure factor, the population density, population aged 14 and below, population aged 65 and above data are normalized and then weighted averaged to obtain it.

[0086] In addition, the mapping relationship between each high temperature disaster prevention measure and the high temperature risk factor can be achieved through existing technology, and it is not difficult for technical personnel in this field to implement. For example, correlation analysis can be completed through Spearman correlation analysis method, Pearson correlation analysis method or Gamma correlation analysis method. The specific mapping method is determined according to actual needs, and this embodiment does not make specific limitations here.

[0087] S2. Determine the target research area;

[0088] Preferably, the high temperature risk factor is calculated by the analytic hierarchy process, and the spatial distribution of the high temperature level in the study area is obtained according to the natural breakpoint method. The high-risk area is the key area for decision-making on extreme high temperature events. Subsequent decision optimization can determine whether the target study area is only for the key area or for the entire study area according to decision needs.

[0089] S3. Collect historical data of high temperature assessment indicators, establish and train a deep learning model; the deep learning model takes the high temperature assessment indicators as input and the high temperature risk factors as output.

[0090] The deep learning model in this embodiment adds a ConvLSTM module on the basis of the traditional VGGNet model, where:

[0091] The ConvLSTM module takes the meteorological data index as input, performs a convolution and then outputs it;

[0092] The VGGNet model takes the socioeconomic data indicators and the output of the ConvLSTM module as input and the high temperature risk factor as output.

[0093] S4. Predicting high temperature risk factors in the target study area within a specified time based on the deep learning model;

[0094] S5. Based on the prediction results, establish a decision matrix including different high temperature disaster prevention measures;

[0095] S6. Based on a trapezoidal fuzzy number algorithm, converting the decision matrix into a fuzzy evaluation matrix;

[0096] S7, defuzzifying the fuzzy evaluation matrix to obtain the expected value of the normalized attribute value;

[0097] S8. Sort the various high temperature disaster prevention measures and obtain the optimal decision based on the mapping relationship between the various high temperature disaster prevention measures and the high temperature risk factors.

[0098] Embodiment 2:

[0099] A comprehensive decision-making method for high temperature disaster prevention measures, based on Example 1:

[0100] Step S5 specifically includes:

[0101] S501. Using the hazard factor, vulnerability factor, exposure factor, personnel input and workload as matrix factors, each matrix factor is graded respectively; the measure cost indicators include personnel input and workload;

[0102] S502. Based on the expert scoring method, determine the level corresponding to the predicted high temperature risk factor and the level corresponding to the measure cost indicator under different high temperature disaster prevention measures;

[0103] S503: Using the levels of all matrix factors corresponding to each high temperature disaster prevention measure as the decision matrix.

[0104] Step S7 specifically includes:

[0105] S701, extracting the trapezoidal fuzzy number of each matrix factor under different high temperature disaster prevention measures, and establishing a trapezoidal fuzzy number decision matrix;

[0106] In the trapezoidal fuzzy number decision matrix, each high temperature disaster prevention measure and each matrix factor corresponds to a data set, which is expressed as: in, They represent the four trapezoidal fuzzy numbers corresponding to the mth matrix factor under the i-th high temperature disaster prevention measure;

[0107] S702, defuzzifying the trapezoidal fuzzy number decision matrix, and converting the trapezoidal fuzzy number into an expected value;

[0108] S703, calculate the expected vector of each matrix factor; expressed as: (γ 1m ,γ 2m ,…γ im ,…γ Am );

[0109] in, i represents the i-th high temperature disaster prevention measure; m represents the m-th matrix factor; A is the total number of high temperature disaster prevention measures;

[0110] S704, calculating the normalized attribute values ​​to obtain a matrix of normalized attribute values;

[0111] The normalized attribute value is calculated using the following formula: Where: represents the normalized attribute value; β i The matrix representing the trapezoidal fuzzy number; max γim Represents the maximum expected value of the mth matrix factor under the i-th high temperature disaster prevention measure.

[0112] S705, obtaining the expected value of the normalized attribute value;

[0113] S706: Based on the prediction result of the deep learning model, correct the expected value of the normalized attribute value; the specific process is as follows:

[0114] Based on the deep learning model, predict the prediction interval of each high temperature risk factor under different high temperature disaster prevention measures;

[0115] For any high temperature risk factor, perform the following judgment:

[0116] If the expected value of the normalized attribute value is within the prediction interval of the high temperature risk factor, no correction is made;

[0117] If the expected value of the normalized attribute value is not within the prediction interval of the high temperature risk factor, the expected value of the normalized attribute value corresponding to the high temperature risk factor is corrected to the threshold of the prediction interval.

[0118] In this embodiment, the method for ranking various high temperature disaster prevention measures includes:

[0119] S801, establishing a sorting matrix with each high temperature disaster prevention measure as a row and the matrix factors as a column;

[0120] S802. Calculate the consistency index and inconsistency index of each high temperature disaster prevention measure: The specific calculation formula is:

[0121]

[0122] Where: C j (i, k) is the consistency index, which means that on the jth matrix factor, a ij Better than a kj degree; D j (i, k) is the inconsistency index, which means that on the jth matrix factor, a ij Worse than a kj degree; ij is the expected value of the normalized attribute value corresponding to the i-th high temperature disaster prevention measure and the j-th matrix factor; a kj is the expected value of the normalized attribute value corresponding to the kth high temperature disaster prevention measure and the jth matrix factor; q j is the indifference threshold; p j is better than the threshold; v j is inferior to the threshold.

[0123] Among them, the indifference threshold, better than the threshold and worse than the threshold are all technical terms in this field, which can be specifically understood as:

[0124] Indifference value q j : The absolute value of measures 1 and 2 is less than or equal to q j , it indicates that for measure i, there is no significant difference between measure 1 and measure 2.

[0125] Better than the threshold p j :Measure 1 and measure 2 have absolute values ​​greater than or equal to p j, it means that for measure i, measure 1 is better than measure 2; and the absolute value of measure 1 and measure 2 is greater than or equal to p j Less than or equal to q j , it means that for measure i, measure 1 is slightly better than measure 2.

[0126] Worse than threshold v j :Measure 1 and measure 2 have absolute values ​​greater than or equal to v j , it means that for measure i, measure 1 is inferior to measure 2.

[0127] S803. Calculate the credibility index of each high temperature disaster prevention measure; the specific calculation formula is:

[0128]

[0129] Where: S(i, k) is a ij Compared to a kj The credibility index; C(i, k) is the overall consistency index;

[0130] S804, converting the credibility index into a net credibility index:

[0131] Z i =Σ i∈A S(i, k)-∑ i∈A S(k,i)

[0132] Where: Z i is the net credibility index, which represents the degree to which the ith high temperature disaster prevention measure is superior to other high temperature disaster prevention measures; A is the total number of high temperature disaster prevention measures; ∑ i∈A S(i, k) is the sum of all S(i, k); ∑ i∈A S(k, i) is the sum of all S(k, i); S(k, i) is a kj Compared to a ij Credibility index;

[0133] S805, obtaining a ranking result according to the credibility index; Z i The larger it is, the better the effect of the corresponding i-th high temperature disaster prevention measure is, and the higher its ranking is.

[0134] Embodiment 3:

[0135] A comprehensive decision-making method for high temperature disaster prevention measures is provided. Based on any of the above embodiments, this embodiment takes a certain area in the middle and lower reaches of the Yangtze River as the research area and a certain day in 2022 as the designated research time for illustration.

[0136] In this embodiment, 10 different high temperature disaster prevention measures were considered, and 11 experts were invited to conduct a detailed evaluation of each measure indicator. The established decision matrix is ​​shown in Table 1, where S1 to S4 represent different levels:

[0137] Table 1 Decision matrix

[0138]

[0139]

[0140] This embodiment uses a triangular fuzzy language term set, and the trapezoidal fuzzy number decision matrix calculated is shown in Table 2:

[0141] Table 2 Trapezoidal fuzzy number decision matrix

[0142]

[0143] Then, each trapezoidal fuzzy number is converted into an expected value according to the trapezoidal fuzzy number normalization method. Taking "personnel input" as an example, the trapezoidal fuzzy number decision matrix of 10 measures is calculated through the triangular fuzzy language term set:

[0144]

[0145] After converting the above matrix into expected values, the expected vector is: (0.6675, 0.3325, 0.4175, 0.255, 0.4175, 0.6675, 0.4175, 0.375, 0.4175, 0.375), and the maximum expected value is 0.6675.

[0146] After that, the matrix of normalized attribute values ​​can be obtained as shown in Table 3:

[0147] Table 3 Matrix of normalized attribute values

[0148] <![CDATA[a 1 ]]> <![CDATA[a 2 ]]> <![CDATA[a 3 ]]> <![CDATA[a 4 ]]> 0.749 1.0 1.0 1.24 0.25 0.49 0.49 0.74 0.25 0.49 0.74 1.0 0 0 0.52 1.0 0.25 0.49 0.74 1.0 0.49 0.96 1.0 1.5 0.25 0.49 0.75 1.0 0.25 0.30 0.69 1.0 0.25 0.50 0.75 1.0 0.25 0.30 0.69 1.0

[0149] The expected values ​​of the normalized attribute values ​​are shown in Table 4:

[0150] Table 4 Expected values ​​of attribute values ​​after normalization

[0151]

[0152] In this embodiment, based on the prediction results of the deep learning model, the prediction intervals of each high temperature risk factor are obtained, and a threshold conversion table as shown in Table 5 is established:

[0153] Table 5 Threshold conversion table

[0154] Exposure Index Threshold Danger Index Threshold Vulnerability Index Threshold <![CDATA[0.12<I pe ≤0.29]]> 0.25 <![CDATA[0.12<I E ≤0.34]]> 0.37 <![CDATA[0.12<I C ≤0.29]]> 0.25 <![CDATA[0.29<I pe ≤0.47]]> 0.38 <![CDATA[0.34<I E ≤0.50]]> 0.5 <![CDATA[0.29<I C ≤0.64]]> 0.56 <![CDATA[0.47<I pe ≤0.64]]> 0.56 <![CDATA[0.59<I E ≤0.78]]> 0.63 <![CDATA[0.64<I C ≤0.78]]> 0.63 <![CDATA[0.64<I pe ≤0.78]]> 0.63 <![CDATA[0.78<I E ≤1]]> 1 <![CDATA[0.78<I C ≤1]]> 1 <![CDATA[0.78<I pe ≤1]]> 1

[0155] Among them, I pe Represents the actual exposure index; I E Represents the actual risk index; I C Represents the actual vulnerability index.

[0156] Based on Table 5, Table 4 is modified to obtain the expected value of the modified and normalized attribute value as shown in Table 6:

[0157] Table 6 Expected values ​​of corrected and normalized attribute values

[0158]

[0159] The final measure ranking result obtained in this embodiment is as follows: Figure 2 The results show that the high temperature weather on that day had the greatest impact on industry, followed by population and finally agriculture. The top four measures were two industrial measures, population measures and agricultural measures, which is consistent with the characteristics that the industry was the most severely affected by this high temperature event.

[0160] This embodiment screens out the top four measures, and uses statistical analysis methods and numerical simulation methods to quantify the relationship between the four measures and the cooling effect.

[0161] The optimization results based on these four measures show that the maximum temperature drop in the long-term target study area is 0.42°C. This shows that under certain conditions, the adoption of a series of measures, including stopping outdoor work, opening temporary summer resorts, and strengthening the issuance of high temperature subsidies, can lead to a maximum temperature drop of 0.42°C in the objective function. It can be shown that the implementation of these measures has significantly reduced the risk of high temperatures, especially in provincial capitals and the Yangtze River Delta region. Although the risk still exists, its severity has decreased significantly.

[0162] The optimization results also show that the effects of implementing these high temperature disaster prevention measures vary in different regions; especially for the original high-risk areas, the impact of these measures is more significant. Therefore, the method of this application can also be used to make quantitative decisions and estimate the effects of measures for high-risk areas alone.

[0163] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

[0164] It should be noted that, in this article, terms such as "comprises", "includes" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or apparatus that includes a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or apparatus.

Claims

1. A comprehensive decision-making method for high temperature disaster prevention measures, characterized in that: include: Based on several high temperature assessment indicators, high temperature risk factors are determined, and several high temperature disaster prevention measures are collected; among them, high temperature risk factors include hazard factors, vulnerability factors and exposure factors; high temperature assessment indicators include meteorological data indicators and socio-economic data indicators, and the meteorological data indicators include: maximum surface temperature, maximum air temperature, average surface temperature, average air temperature, air pressure, specific humidity, wind speed, and precipitation; the socio-economic data indicators include: population density, land use type, building type, and GDP density; for hazard factors, the maximum surface temperature, maximum air temperature, average surface temperature, air pressure, wind speed, and precipitation are normalized and weighted averaged; for vulnerability factors, the vegetation index, water body index, land use type, building type, and GDP density are normalized and weighted averaged; for exposure factors, the population density, the population aged 14 and below, and the population aged 65 and above are normalized and weighted averaged; Identify target study areas; Collect historical data of high temperature assessment indicators, establish and train a deep learning model; the deep learning model takes the high temperature assessment indicators as input and the high temperature risk factors as output; Based on the deep learning model, predict the high temperature risk factor of the target study area within a specified time, where the specified time is a future time; Based on the expert scoring method, a decision matrix including different high temperature disaster prevention measures was established; The method for establishing a decision matrix including different high temperature disaster prevention measures based on the expert scoring method includes: The exposure measure effect, the hazard measure effect, the vulnerability measure effect, and the measure cost index are used as matrix factors. The measure cost index includes personnel input and workload. Each matrix factor is divided into several levels. Based on the expert scoring method, the corresponding levels of the exposure measure effect, the hazard measure effect, the vulnerability measure effect, and the measure cost index under different high temperature disaster prevention measures are determined; The levels of all matrix factors corresponding to each high temperature disaster prevention measure are used as the decision matrix; Based on the trapezoidal fuzzy number algorithm, the decision matrix is ​​converted into a fuzzy evaluation matrix. In the trapezoidal fuzzy number type fuzzy evaluation matrix, there are multiple high temperature disaster prevention measures and several matrix factors. Each high temperature disaster prevention measure and each matrix factor corresponds to a data set, and each data set includes four trapezoidal fuzzy numbers. Defuzzify the fuzzy evaluation matrix to obtain the expected value of the normalized attribute value; The method of performing defuzzification processing on the fuzzy evaluation matrix to obtain the expected value of the normalized attribute value includes: Extract the trapezoidal fuzzy number of each matrix factor in all high temperature disaster prevention measures, and establish a trapezoidal fuzzy number decision matrix; Defuzzification is performed on the trapezoidal fuzzy number decision matrix, and the trapezoidal fuzzy number is converted into the expected value; Calculate the expected vector for each matrix factor; Calculate the normalized attribute values ​​to obtain a matrix of normalized attribute values; Get the expected value of the normalized attribute value; In the trapezoidal fuzzy number decision matrix, each high temperature disaster prevention measure and each matrix factor corresponds to a data set, and the data set is expressed as: in, They represent the four trapezoidal fuzzy numbers corresponding to the mth matrix factor under the i-th high temperature disaster prevention measure; The expected vector of each matrix factor is expressed as: (γ 1m ,γ 2m ,…γ im ,…γ Am ); in, i represents the ith high temperature disaster prevention measure; m represents the mth matrix factor; The normalized attribute value is calculated by the following formula: Where: represents the normalized attribute value; β i represents the trapezoidal fuzzy number matrix; max γim represents the maximum expected value of the mth matrix factor under the ith high temperature disaster prevention measure; Based on the prediction results of the deep learning model, the expected value of the normalized attribute value is corrected; The method for correcting the expected value of the normalized attribute value comprises: Based on the deep learning model, predict the prediction interval of each high temperature risk factor; For any high temperature risk factor, perform the following judgment: If the expected value of the normalized attribute value is within the prediction interval of the high temperature risk factor, no correction is made; If the expected value of the normalized attribute value is not within the prediction interval of the high temperature risk factor, the expected value of the normalized attribute value corresponding to the high temperature risk factor is corrected to the threshold of the prediction interval; Sort various high temperature disaster prevention measures to get the best decision; The method for ranking various high temperature disaster prevention measures includes: A sorting matrix is ​​established with each high temperature disaster prevention measure as a row and the matrix factor as a column; the consistency index and inconsistency index of each high temperature disaster prevention measure are calculated, and the formula is: Where: C j (i,k) is the consistency index, which means that on the jth matrix factor, a ij Better than a kj degree; D j (i,k) is the inconsistency index, which means that on the jth matrix factor, a ij Worse than a kj degree; ij is the expected value of the normalized attribute value corresponding to the i-th high temperature disaster prevention measure and the j-th matrix factor; a kj is the expected value of the normalized attribute value corresponding to the kth high temperature disaster prevention measure and the jth matrix factor; q j is the indifference threshold; p j is better than the threshold; v j is inferior to the threshold; Calculate the credibility index S(i, k) of each high temperature disaster prevention measure; Convert the credibility index S(i, k) into a net credibility index Z i , according to the net credibility index Z i Get the sorting result, Z i The larger it is, the better the effect of the corresponding i-th high temperature disaster prevention measure is, and the higher its ranking is.

2. A comprehensive decision-making method for high temperature disaster prevention measures according to claim 1, characterized in that: The deep learning model includes: The ConvLSTM module takes the meteorological data index as input, performs a convolution and then outputs it; The VGGNet model takes the socioeconomic data indicators and the output of the ConvLSTM module as input and the high temperature risk factor as output.

Images