CN114997514B - A method for evaluating and predicting the development degree of fissure disease in rammed earth ruins - Google Patents

Abstract

The invention proposes a method for evaluating and predicting the development degree of fissure disease in rammed earth ruins. The steps are: selecting factors directly related to the development of fissure disease to establish an evaluation index system; The first objective weight is calculated by the variable instability index method, the second objective weight is calculated by the improved entropy value method, and the comprehensive weight is obtained through equal-weight weighted average processing; the TOPSIS approach to the ideal solution and the comprehensive weight are used to evaluate the development level of crack disease; the BP neural network is constructed The prediction model uses the evaluation index data as input data and the evaluation results as output data to predict the future development trend of fissure diseases in rammed earth sites in arid areas of Northwest China. The invention evaluates and predicts the development of fissure disease in rammed earth ruins based on natural environment characteristics, proposes a method for predicting the development trend of fissure disease with high precision, and improves the effectiveness and controllability of fissure disease control.

Description

A method for evaluating and predicting the development degree of fissure disease in rammed earth ruins

technical field

The present invention relates to the technical field of disease development characteristics of rammed earth ruins, in particular to a method for evaluating and predicting the development degree of fissure disease in rammed earth ruins, which evaluates the development degree of fissure disease in rammed earth ruins based on natural environment characteristics and establishes the future development trend of fissure disease predictive model.

Background technique

The reasons for the development of diseases in rammed earth sites are relatively complex. Generally speaking, they mainly include internal factors and external factors. Meteorological environment, geological environment, etc. The safety and stability of rammed earth sites are seriously threatened by factors such as temperature, rainfall, wind, salinization and other factors such as natural erosion and human destruction. The process of quantitative change to qualitative change from a large number of typical diseases such as collapse and collapse to rapid extinction. It is a comprehensive applied science to effectively evaluate, predict, prevent, and slow down the damage of rammed earth ruins under the influence of natural environment characteristics, especially for earth ruins distributed in the arid regions of Northwest China, there is still a lack of disease characteristics and systematic evaluation at home and abroad There is a lack of targeted protection measures. Few studies have used effective evaluation methods to quantitatively distinguish the severity of disease development based on natural environment characteristics, and few studies have used effective prediction and early warning systems to further predict and analyze the future development trend of diseases, and then implement hierarchical scientific protection.

Contents of the invention

Aiming at the current technical problem of lack of scientific and quantitative evaluation and prediction and early warning system for the development of rammed earth site disease, the present invention proposes a method for evaluating and predicting the development degree of crack disease in rammed earth site, based on the characteristics of the natural environment. Evaluation and prediction, taking into account the evaluation indicators that affect the development of fissures in rammed earth sites; fully considering the subjective and objective combination weighting method, selecting the optimal ideal solution to score the development degree of fissures in the study area, and proposing high-precision prediction of fissures Develop trend methods to improve the effectiveness and controllability of fissure disease control.

In order to achieve the above object, the technical solution of the present invention is achieved in this way: a method for evaluating and predicting the development degree of crack disease in rammed earth ruins, the steps of which are as follows:

Step S1: Select factors directly related to the development of fissure disease to establish an evaluation index system. The evaluation indexes include: gap length, gap opening, fracture connectivity rate and natural environment characteristics;

Step S2: Use the fuzzy analytic hierarchy process to calculate the subjective weight according to the evaluation index system, use the multivariate instability index method to calculate the first objective weight according to the data of the evaluation index system in the study area, and use the improved entropy value method to calculate the second objective weight. Equal-weighted weighted average processing of subjective weight, first objective weight and second objective weight to obtain the comprehensive weight of the evaluation index;

Step S3: Use TOPSIS to approximate the ideal solution method to evaluate the development level of fissure disease: calculate the Euclidean distance between each evaluation plan and the positive and negative ideal solution according to the comprehensive weight determined in step S2, and use the Euclidean distance to evaluate the development of fissure disease in multiple rammed earth sites in the arid Northwest region Evaluate the degree to obtain the development grade of fissure disease;

Step S4: Use machine learning BP neural network to build a BP neural network prediction model, use all the evaluation indicators of the evaluation index system as input data, and the evaluation results of step S3 as output data to predict the future development trend of crack diseases in multiple rammed earth ruins in arid areas of Northwest China Make predictions and verify the prediction results.

Preferably, the gap length in the evaluation index includes the total gap length and the average gap length; the gap opening includes the maximum gap opening and the average gap opening; the fracture connectivity rate includes volume density and linear density; natural environment characteristics include annual average Temperature, average annual rainfall, average annual evaporation, drought index and average annual sunshine hours.

Preferably, the directionality of each evaluation index is judged to obtain that the directionality of each evaluation index is negative.

Preferably, the implementation method of the fuzzy analytic hierarchy process is:

(1) Construct the FAHP model: including target layer, criterion layer, and index layer. The target layer is the degree of fracture disease development A; the criterion layer includes gap length B1, gap opening degree B2, fracture connectivity rate B3, and environmental factors B4; the index layer includes Total gap length C1, average gap length C2, maximum gap opening C3, average gap opening C4, volume density C5, line density C6, annual average temperature C7, annual average rainfall C8, annual average evaporation C9, drought index C10 and the average annual sunshine hours C11;

(2) Construct fuzzy judgment matrix A=(a _ij ) _n×n through expert scoring, where element a _ii =0.5; a _ij +a _ji =1, a _ij ≥0; n represents the number of selected evaluation indicators;

(3) Calculate the subjective weight of each evaluation index according to the fuzzy judgment matrix A;

Among them, r _i represents the sum of the evaluation index scales of the i-th row of the fuzzy judgment matrix A, r _j represents the j column element corresponding to r _i ; r _ij represents the matrix element constructed to obtain the subjective weight W _i ;

(4) Consistency check:

The feature matrix W=(W _ij ) _n× n is constructed by the weight vector (W ₁ ,W ₂ ,W ₃ ..,W _i ,...W _n ) ^T , and the element W _ij of the feature matrix W is:

Check the consistency of fuzzy judgment matrix A and feature matrix W:

Construct Fuzzy Complementary Judgment Matrix

Take the threshold α=0.1; the smaller the threshold α, the higher the satisfaction of the fuzzy judgment matrix A and the higher the consistency requirement.

Preferably, the method for calculating the subjective weight is: constructing the fuzzy judgment matrix A1 of the gap length B1, the gap opening B2, the gap connectivity rate B3 and the natural environment feature B4 in the criterion layer through expert scoring, and calculating the gap length of the criterion layer The weights of B1, gap opening B2, fracture connectivity rate B3, and natural environment characteristics B4; the fuzzy judgment matrix A2 of the index layer of the gap length B1, the fuzzy judgment matrix A3 of the index layer of the gap opening B2, and the fissure The fuzzy judgment matrix A4 of the index layer of the connectivity rate B3 and the fuzzy judgment matrix A5 of the index layer of the natural environment feature B4 are calculated according to the fuzzy judgment matrix A2 of the total gap length C1 and the average gap length C2 weight of the evaluation index of the index layer, according to The fuzzy judgment matrix A3 calculates the weights of the maximum gap opening degree C3 and the average gap opening degree C4 of the evaluation index of the index layer, and calculates the weights of the volume density C5 and linear density C6 of the evaluation index of the index layer according to the fuzzy judgment matrix A4, and according to the fuzzy judgment Matrix A5 calculates the weights of the annual average temperature C7, annual average rainfall C8, annual average evaporation C9, drought index C10 and annual average sunshine hours C11 of the evaluation indicators of the index layer; the weights of the evaluation indicators of the index layer and the corresponding criteria The weights of the layers are multiplied to obtain the subjective weights of each evaluation index.

Preferably, the multivariate instability index method analyzes the coefficient of variation between the evaluation indicators in the research area in a statistical manner, and through data normalization processing, the weight of each evaluation index is calculated according to the coefficient of variation as the first objective weight The improved entropy value method determines the second objective weight by normalizing the data of each evaluation index in the research area and then shifting the evaluation index.

Preferably, the calculation method of the first objective weight is:

1). Normalize the data: make the data mapped between [0,1]; since the directionality of the evaluation indicators are all negative, the data is normalized as:

Among them, X represents the corresponding value of the j-th research area of the i-th evaluation index; X ^* represents the dimensionless matrix element obtained after normalizing the negative correlation index; max represents the j-th research area of the i-th evaluation index The maximum value within the range; min represents the minimum value within the range of the jth research area of the i-th evaluation index;

2). Sample proportion analysis: The proportion of each evaluation index in the total sample number is:

Among them, a _ij represents the element of the dimensionless matrix after normalization; X _ij represents the proportion importance of the jth research area of the i-th evaluation index;

为：

3).Sample mean

for:

Among them, x ₁ , x ₂ , and x _nl are elements that account for the proportion of the importance degree X _ij after judging the importance degree; n1 represents the number of study areas;

4). Calculate the sample standard deviation as:

Among them, _σi represents the standard deviation;

5). Calculate the coefficient of variation of each evaluation index as:

6). Determine the first objective weight W _1i of the evaluation index as:

Preferably, the calculation method of the second objective weight is:

其中，X_ij表示第i个评价指标的第j项研究区域的对应数值；X^* _ij表示对于负相关指标归一化后得到的无量纲矩阵元素；max_ij表示第i个评价指标的第j项研究区域范围内最大值；min_ij表示第i个评价指标的第j项研究区域范围内最小值；A1. Normalization processing:

Among them, X _ij represents the corresponding value of the j-th research area of the i-th evaluation index; X ^* _ij represents the dimensionless matrix element obtained after normalizing the negative correlation index; max _ij represents the j-th value of the i-th evaluation index The maximum value within the scope of the research area; min _ij represents the minimum value of the i-th evaluation index within the scope of the j-th research area;

A2. Evaluation index translation processing: X′ _ij = X ^* _ij + p;

Among them, p is the translation range of the evaluation index; X′ _ij represents the new matrix element obtained by the index translation of the dimensionless matrix element after normalization;

A3. Calculate the proportion of the j-th sample under the i-th evaluation index to the evaluation index:

Among them, n1 represents the total number of study areas;

A4. Calculate the information entropy of the i-th evaluation index:

A5. Calculate the information utility value of the i-th evaluation index: d _i =1-e _i ;

A6. Calculating the second objective weight of the i-th evaluation index is:

The subjective weight W _i , the first objective weight W _1i and the second objective weight W _2i are combined by using the method of equal weight weighted average, and the comprehensive weight of the evaluation index of the crack disease development degree of the rammed earth site is obtained as follows:

Preferably, in the step S3, the implementation method of obtaining the crack disease development level by using TOPSIS to approximate the ideal solution method is:

B1. Establish the evaluation matrix D1 of the original data;

B2. Standardize the data of the evaluation matrix D1;

B3. Construct a normalized decision matrix Z, where:

表示i个评价指标第j个研究区域标准化后的元素；Among them, Z _ij represents the elements in the normalized decision matrix Z;

Indicates the standardized elements of the jth research area of the i evaluation index;

B4. Construct a weighted normalized decision matrix V, where the element V _ij =λ _i ×Z _ij , λ _i is the comprehensive weight of the i-th evaluation index, Z _ij is the element of the normalized decision matrix Z; n represents the number of evaluation indicators ;

B5. Determine the positive ideal solution and the negative ideal solution are:

Positive ideal solution:

Negative ideal solution:

B6. Determine the distance between positive and negative ideal solutions and the degree of crack development in each research area:

为：

The distance from the degree of fissure development in the jth research area to the positive ideal solution V ⁺

for:

为：

The distance from the degree of fissure development in the jth research area to the negative ideal solution V ^-

for:

接近度即是评分值D；B7. Determine the degree of crack development in each research area and the closeness of positive and negative ideal solutions

The proximity is the score value D;

B8. When the score value D∈(0,0.2), the research area is a low-risk area; when the score value D∈[0.2,0.35], the research area is a low-risk area; when the score value D∈(0.35,0.5], The research area is a medium susceptibility area; when the score value D∈(0.5,1), the research area is a high susceptibility area.

Preferably, the BP neural network prediction model is constructed by Matlab programming software, and the total gap length C1, the average gap length C2, the maximum gap opening C3, the average gap opening C4, the bulk density C5, and the linear density C6 of the evaluation index are constructed. , annual average temperature C7, annual average rainfall C8, annual average evaporation C9, drought index C10, and annual average sunshine hours C11 are used as the input layer data, the average value D is used as the output data, and the input of the BP neural network prediction model The layer includes 11 neurons, the output layer includes 1 neuron, and the hidden layer includes 4 neurons. The training of the BP neural network prediction model uses the levenberg-marquardt algorithm; the data of the evaluation index of the research area is sample data, which is randomly selected for training. The sample is 70% sample data, the test sample is 15% sample data, and the forecast sample is 15% sample data.

Compared with prior art, the beneficial effect of the present invention:

1. Fully considering the problems of human subjective bias, data variability, and discreteness, the weighting method of combining three kinds of evaluations is used to carry out comprehensive weighting on the evaluation indicators of fissure disease.

2. Fully consider the relationship between the development degree of fissure disease in rammed earth sites and the response to natural environment characteristics, find out the dominant evaluation factors of internal and external factors, construct a reasonable scoring value division standard, and establish a grading system suitable for the development degree of fissure disease in rammed earth sites. Scientific and systematic evaluation of soil fissure disease control.

3. In the present invention, the subjective and objective comprehensive weighting-TOPSIS evaluation method combined with the BP neural network prediction model of machine learning can also perform quantitative evaluation, prediction and early warning of crack diseases in rammed earth ruins in other arid areas.

Description of drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings that need to be used in the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only These are some embodiments of the present invention. Those skilled in the art can also obtain other drawings based on these drawings without creative work.

Fig. 1 is a flowchart of the present invention.

Fig. 2 is a schematic diagram of the evaluation hierarchy model of the FAHP method of the present invention.

Fig. 3 is a structural diagram of the BP neural network prediction model of the present invention.

Fig. 4 is an analysis diagram of evaluation results of development degree of 7 kinds of fissure diseases in the present invention.

Fig. 5 is an analysis diagram of the low, middle and high susceptibility areas of the development degree evaluation results of 7 kinds of fissure diseases in the present invention.

Fig. 6 is a trend chart of fissure disease eigenvalues and natural environment eigenvalues in three research areas of the present invention, high, middle and low.

Fig. 7 is a graph showing the development trend of fissure disease in three research areas of the present invention, high, middle and low.

Fig. 8 is a comparison chart between the actual scoring value and the predicted scoring value of the training sample in the present invention.

Fig. 9 is a comparison chart between the actual scoring value and the predicted scoring value of the BP neural network prediction model of the present invention.

Detailed ways

The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some, not all, embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without making creative efforts belong to the protection scope of the present invention.

As shown in Figure 1, a method for evaluating and predicting the development degree of fissure disease in rammed earth ruins. This example takes the fissure disease of rammed earth ruins as an example, and investigates the area of fissure disease, gap length, gap opening, body density, and linear density. and its own factors, and collect the natural climate factors of the region, analyze the relationship between the cracks’ own factors and the response to climate factors, jointly quantitatively evaluate the degree of disease development, and predict the future development of cracks. The specific steps are as follows:

Step S1: Select factors directly related to the development of fissure disease to establish an evaluation index system. The evaluation indexes include: gap length, gap opening, fracture connectivity rate and natural environment characteristics.

Select the research area and evaluation index, and the evaluation index must be a factor directly related to the development of fissure disease. Among the evaluation indicators, the gap length includes the total gap length and the average gap length; the gap opening includes: the maximum gap opening and the average gap opening; the fracture connectivity rate includes: volume density, linear density; natural environment characteristics include: annual average temperature, Average rainfall, average annual evaporation, drought index, and average annual sunshine hours; the data of typical arid research areas and evaluation indicators are shown in Table 1.

Gap length refers to the occurrence of unloading cracks, deformation cracks, structural cracks, and construction process cracks in the body of the rammed earth site due to geological tectonic action (stress release), stress redistribution, fault and joint structure, and building process materials. The length developed on the body per unit area is hereinafter referred to as the gap length, including the total gap length (m) and the average gap length (m). Gap opening refers to the occurrence of unloading cracks, deformation cracks, structural cracks, and construction process cracks in the body of the rammed earth site due to geological tectonic action (stress release), stress redistribution, fault and joint structure, and building process materials. The width developed on the body per unit area is hereinafter referred to as the gap opening, including the maximum gap opening (cm) and the average gap opening (cm).

The fissure connectivity rate refers to the fact that unloading fissures, deformation fissures, structural fissures, and construction process fissures cause broken blocks on the rammed earth site body, cause other disease development, and aggravate the development of existing diseases, which affect the stability of the body. The degree of density produced on the bulk is hereinafter referred to as the fracture connectivity rate, including bulk density and linear density. Bulk density is the density of the total gap length of the unit area of the rammed earth site body resulting in the fracture connectivity rate, and the unit is mm ^-2 . The linear density is the density of the number of gap lengths that produce the connectivity rate of cracks per unit area of the rammed earth site body, and the unit is bar.

Table 1 Research areas and research indicators of typical arid rammed earth sites

The evaluation index system for the development degree of fissure disease in typical drought-type research areas is shown in Table 2. The directionality of each evaluation index is judged and explained through experience summary and induction.

Table 2 Evaluation index system for the development degree of fissure disease in typical drought-type research areas

The development of gap length and gap opening mainly refers to the unloading cracks, deformation cracks, structural cracks, and construction craft Fissure disease, which in turn affects the stability of the wall, produces weak structural surfaces, breaks blocks, and causes the serial development of other secondary diseases. The directionality of this evaluation index is negative.

Volume density and linear density are mainly due to the unloading cracks, deformation cracks, structural cracks, and construction process cracks that cause broken blocks on the body of the rammed earth site, cause the development of other diseases, and aggravate the development of existing diseases, which affect the stability of the body. The total gap length and the density of the number of gap lengths generated on the body per unit area. The greater the density, the more broken the wall and the worse the stability. The directionality of this evaluation index is negative.

The temperature difference between morning and evening is large in the arid region of Northwest China, with the average annual temperature range of 1.5-7.7°C, most of which are concentrated at around 4°C. The lower the temperature, the more serious the dry shrinkage of the cracks in the rammed earth site; as the temperature rises, the crack expansion increases, and due to the large temperature difference, the expansion and contraction changes rapidly and significantly, which in turn affects the development of cracks and promotes the serious development of crack diseases. The directionality of the annual average temperature index is negative.

The average annual rainfall ranges from 125 to 544 mm, which belongs to the arid region in the northwest. Due to the extreme climate in the arid region in the northwest, the rainfall is mostly concentrated rainfall, which has a severe impact on the wall of the rammed earth site, and the infiltration of rainwater makes Existing fissure diseases develop seriously and new fissure diseases are easy to produce, so the directionality of the average annual rainfall index is negative.

The average annual evaporation ranges from 875 to 2665 mm. Due to the dry climate in the arid region of Northwest China, the extremely dry climate after concentrated rainfall causes a sudden increase in evaporation, and the phenomenon of water and salt migration in the wall of the rammed earth ruins occurs, resulting in the crystallization of soluble salts. The salinization of the wall is caused, and the further crystal expansion phenomenon induces the development of cracks in the wall, so the directionality of the annual average evaporation index is negative.

The drought index is the ratio of annual evaporative capacity to annual precipitation, which represents the degree of climate drought. The range of variation is 1.62 to 9.6. The greater the drought index, the more severe the dryness. Because the wall of the rammed earth site has been exposed to the wild for a long time, it is dry and cold. Strong wind and sand make the surface of the wall prone to chapping, coupled with the infiltration of concentrated rainfall and rainwater, the development of crack diseases is serious, so the directionality of the drought index index is negative.

The annual average sunshine hours range from 2536 to 3221 hours. The overall sunshine radiation in the arid area of Northwest China is large and the average daily radiation time is long. The temperature stress difference caused by heat can easily cause thermal damage and thermal damage to the wall, which in turn aggravates the development of cracks and diseases. Therefore, the directionality of the annual average sunshine hours index is negative.

The positive index of the element makes the rammed earth site have the ability to resist deterioration. For example, the greater the index of compressive, tensile, and shear resistance elements, the better the mechanical properties and strength of the rammed earth site, and the stronger the wall’s ability to resist external damage. The earthen ruins are stronger. Negative indicators of elements will intensify the deterioration of rammed earth sites, such as rainfall indicators, solar radiation indicators, etc. The greater the negative index, the stronger the deterioration of rammed earth sites. The evaluation indicators selected this time are all negative indicators, and the weight of the negative evaluation indicators reflects their influence on the development degree of fissure disease, which is in line with the theme of evaluation of the degree of fissure development under the influence of natural environment characteristics. Due to the large amount of data and inconsistent units, the data can only be compared between the characteristics of different attributes (different units). Before the evaluation, the data needs to be processed without dimensionality. The positive and negative index calculation formulas in the data normalization processing Not the same, the choice of calculation formula also needs to distinguish between positive and negative indicators.

Step S2: Use the Fuzzy Analytic Hierarchy Process to construct the FAHP model based on the evaluation index system to calculate the subjective weight, use the multivariate instability index method to determine the first objective weight based on the evaluation index data of the study area, and use the improved entropy value method to calculate the second objective weight , the comprehensive weight of the evaluation index is obtained by processing the subjective weight, the first objective weight and the second objective weight by equal-weight weighted average.

Construct the multivariate instability index and the improved entropy method logical relationship model, determine the objective weight, and use the subjective and objective weighting method to determine the index combination weight.

By constructing FAHP model including target layer, criterion layer and index layer, the subjective weight is determined. Construct the FAHP model, as shown in Figure 2, divided into target layer: fracture disease development degree A; criterion layer: gap length B1, gap opening degree B2, fracture connectivity rate B3, environmental factors B4; index layer: total gap length C1, average Gap length C2, maximum gap opening C3, average gap opening C4, bulk density C5, linear density C6, annual average temperature C7, annual average rainfall C8, annual average evaporation C9, drought index C10 and annual average sunshine hours C11.

The fuzzy analytic hierarchy process (FAHP) is used to determine the subjective weight, and the experts score each evaluation index, and the fuzzy complementary judgment matrix is constructed, and the consistency of the weight value is checked by the compatibility of the fuzzy complementary judgment matrix. When dealing with complex decision-making problems, the AHP method often does not take into account the fuzziness of the judgment matrix when constructing the judgment matrix. The Fuzzy Analytic Hierarchy Process FAHP solves this problem very well. Its basic idea and steps are basically consistent with AHP, and fuzzy judgment is introduced into the evaluation system.

1. Establish a hierarchical analysis structure. as shown in picture 2.

2. Construct a fuzzy judgment matrix: compare and judge pairs of evaluation indicators, and quantitatively express the importance of one evaluation index over the other evaluation index. As shown in Table 3, the meaning of the scale a _ij of the fuzzy judgment matrix can be obtained. A=(a _ij ) _n×n , where a _ii =0.5; a _ij +a _ji =1, a _ij ≥0. n represents the number of selected evaluation indicators. For example, the criterion layer has four indicators B1, B2, B3, and B4, n=4, and the construction matrix is 4×4.

Table 3 Scale meaning of fuzzy judgment matrix

3. The subjective weight formula for solving the fuzzy judgment matrix A is as follows:

Among them, r _i represents the sum of the i-th evaluation index scale of fuzzy judgment matrix A, r _j represents the j column element corresponding to r _i ; r _ij represents the matrix element for constructing the subjective weight W _i .

4. Consistency check. Construct the feature matrix W of the fuzzy judgment matrix A, and check the consistency between the fuzzy judgment matrix A and the feature matrix W.

(1). Using the weight vector (W ₁ , W ₂ , W ₃ ..,W _i ,...W _n ) ^T of the fuzzy judgment matrix A to construct the feature matrix W=(W _ij ) _{n× n} , the element W _ij of the feature matrix W is:

(2). Check the consistency of the fuzzy judgment matrix A and the characteristic matrix W, construct the fuzzy complementary judgment matrix X, use the formula (5), when α=0.1 or so, that is, the smaller the α, it shows that the satisfaction of constructing the fuzzy judgment matrix A is higher High, the higher the consistency requirement.

The criterion layer fuzzy judgment matrix A is constructed by expert scoring, the feature matrix W and the fuzzy complementary judgment matrix X are calculated, and the calculation weight and consistency test are carried out, as shown in Table 4.

Table 4 Consistency check of the matrix and weights constructed by the criterion layer

The fuzzy judgment matrices of the B1, B2, and B3 criterion layers are respectively constructed through expert scoring, and their corresponding feature matrices and complementary judgment matrices are calculated, and the calculation weight and consistency check are performed, as shown in Table 5.

Table 5 Consistency check of matrix and weights constructed by index layers B1, B2, and B3

Construct the fuzzy judgment matrix of the B4 criterion layer through expert scoring, calculate its corresponding feature matrix and complementary judgment matrix, and carry out calculation weight and consistency check, as shown in Table 6.

Table 6 Matrix and Weight Consistency Test Constructed by Index Layer B4

Multiply the weights obtained from the criterion layer and the index layer to obtain the subjective weights of each evaluation index, as shown in Table 7.

Table 7 FAHP subjective weight

Calculate the first objective weight by using the multivariate instability index method: this method is to analyze the coefficient of variation among the evaluation indicators in the study area in a statistical way, and after normalizing the data, calculate the weight of each evaluation index factor based on the coefficient of variation Weight is the first objective weight, and the method is more objective, focusing on the variability analysis of the data itself. Proceed as follows:

1. Normalize the data. In order to remove the different dimensions and dimensional units between the data, so that the data are comparable to each other, the data will not cause "failure" after linear transformation, so that the data is mapped between [0,1]; because they are all directional is a negative index, so use the following formula:

Among them, X represents the corresponding value of the j-th research area of the i-th evaluation index. X* represents the dimensionless matrix element obtained after normalizing the negative correlation index. max represents the maximum value within the study area of the jth item of the ith evaluation index. min represents the minimum value of the i-th evaluation index within the j-th research area.

2. Sample proportion analysis. The analysis of each evaluation index is located in the proportion of the total sample number. The higher the proportion, the higher the severity of the impact of the evaluation index of this data on the development of fissure disease. The formula is:

Among them, a _ij represents the dimensionless matrix element after normalization. X _ij represents the proportion importance of the jth research area of the ith evaluation index.

为样本均值，反映数组中波动所围绕的中心程度，公式为：3.

is the sample mean, reflecting the center degree of fluctuations in the array, the formula is:

Among them, x ₁ , x ₂ , and x _nl are elements that account for the proportion of the importance degree X _ij after judging the importance degree; n1 represents the number of research areas, and the value is 14 in this embodiment.

的标准差来衡量数据的离散程度，公式为：4. Calculate the sample standard deviation. Sample deviation from the sample mean

The standard deviation to measure the degree of dispersion of the data, the formula is:

Among them, _σi represents the standard deviation.

5. Calculate the coefficient of variation of each evaluation index. The coefficient of variation indicates the sensitivity of each evaluation index factor to the development degree of fissure disease in rammed earth sites. The larger the coefficient of variation, the higher the probability that the evaluation index factor will affect the development of fissure disease. The formula is:

6. Determine the first objective weight of the evaluation index. As shown in Table 8, in the evaluation index system for the development degree of fissure disease in rammed earth ruins, the calculation method of the first objective weight of each evaluation factor is to divide the coefficient of variation of each evaluation index factor by the sum of the coefficient of variation of all factors, which is the fissure The weight of each index factor for disease evaluation, the formula is:

Table 8 Multivariate instability index method to determine the first objective weight

The second objective weight is calculated using the improved entropy value method. Entropy is a measure of uncertain information, which reflects the degree of differentiation of evaluation indicators from the perspective of index dispersion. The smaller the entropy value, the greater the degree of dispersion, and the greater the influence of the evaluation index in the evaluation, that is, the greater the weight. After the data is normalized, the evaluation index is translated to determine the second objective weight. The calculation steps are as follows:

1. Normalization processing. use

Among them, X _ij represents the corresponding value of the j-th research area of the i-th evaluation index; X ^* _ij represents the dimensionless matrix element obtained after normalizing the negative correlation index; max _ij represents the j-th value of the i-th evaluation index The maximum value within the scope of the research area; min _ij represents the minimum value of the i-th evaluation indicator within the scope of the j-th research area.

2. Evaluation index translation processing. After some evaluation indicators are normalized, the value may be small or zero. In order to unify the calculation, the normalized value is shifted to eliminate this situation.

X′ _ij ＝X ^* _ij +p formula (13)

Among them, p is the translation range of the evaluation index, which is taken as 0.1. X′ _ij represents the new matrix elements obtained by index translation of dimensionless matrix elements after normalization. Some indicators are 0 or smaller after the normalization process, which will affect the accuracy of the entropy value during the calculation process, so a unified translation process is performed.

3. Calculate the proportion of the j-th sample under the i-th evaluation index to the evaluation index:

Among them, n1 represents the total number of study areas.

4. Calculate the information entropy of the i-th evaluation index:

Among them, y _ij represents the dimensionless data using the proportion method, that is, the proportion of the evaluation index.

5. Calculate the information utility value of the i-th evaluation index.

d _i =1-e _i formula (16)

6. Calculate the second objective weight of the jth evaluation index, as shown in Table 9, the formula is as follows:

Table 9 Entropy value method to determine the objective weight

The fuzzy analytic hierarchy process (FAHP) determines the subjective weight, and the multivariate instability index method and the improved entropy value method respectively determine the first objective weight and the second objective weight. The variable instability index method focuses on the analysis of data variability, and the improved entropy value method pays more attention to the degree of data dispersion. The three methods have their own advantages and disadvantages. In order to avoid subjective deviations, objective deviations in data quality or completeness, etc., equal weights are used. The weighted average method combines the three methods to obtain the comprehensive weight of the evaluation index of the crack disease development degree of the rammed earth site, as shown in Table 10, using the following formula:

Among them, W _i represents the subjective weight, W _1i represents the weight of the multivariate instability index, W _2i represents the weight of the entropy method, and n represents the number of evaluation indicators.

Advantages of combined weights: ①Three kinds of weight combinations take into account the advantages of both subjective and objective weights, and avoid their own shortcomings; The importance of each evaluation factor; ③Compared with single weight+TOPSIS or other 2 weight+TOPSIS evaluation methods, the 3 weights+TOPSIS evaluation methods are closer to the actual situation of crack disease development, and the evaluated objects are well distinguished It has a certain guiding effect on the later protection measures of fissure disease.

Table 10 Determining the combination weight of subject and object

From the ranking of comprehensive weights, it can be seen that the fracture connectivity rate has the highest impact on the severity of fracture disease, especially the body density; the degree of influence of fracture opening and length on the development of fracture disease is more important, especially the total fracture length and maximum fracture opening, because The infiltration of rainfall will make the fissures further develop into gullies, and finally a through surface will be formed to destroy the wall. Among the natural environmental factors, the temperature index is more important, because the rammed earth site has always been in a dry and rainless area, and the temperature and sunlight radiation have a great impact on the cracks in the site. Disease development is highly affected. It shows that the comprehensive weight after subjective and objective weighting is reasonable, suitable and representative, and can be used to evaluate the development of fissure disease in rammed earth sites in arid research areas.

Step S3: Use TOPSIS to approximate the ideal solution method to evaluate the development level of fissure disease: calculate the Euclidean distance between each evaluation plan and the positive and negative ideal solution according to the comprehensive weight determined in step S2, and use the Euclidean distance to evaluate the development of fissure disease in multiple rammed earth sites in the arid Northwest region The degree is evaluated to obtain the developmental grade of fissure disease.

Using the TOPSIS approach to the ideal solution method to determine the Euclidean distance between each evaluation plan and the positive and negative ideal solutions, that is, to judge the reasonableness based on the distance to the closest idealization degree, and to evaluate the development degree of cracks and diseases in several rammed earth sites in the arid Northwest region, Sort to get the disease development grade.

1. Use Table 1 to establish the evaluation matrix of the original data. There are 14 plans to be evaluated, 11 evaluation indicators, and the original evaluation matrix D is established.

2. Data standardization of evaluation matrix D. The evaluation indicators are extremely large and extremely small indicators. According to the previous judgment, the directionality of the evaluation indicators is negative, and the data is dedimensionalized according to formula (12).

3. Construct a normalized decision matrix Z, using the following formula:

表示i个评价指标第j个研究区域标准化后的元素，nl表示研究区个数。Among them, Z _ij represents the elements in the normalized decision matrix Z;

Indicates the standardized elements of the jth research area of the i evaluation index, and nl indicates the number of research areas.

4. Construct a weighted normalized decision matrix V, where V _ij =λ _j ×Z _ij , where λ _i is the comprehensive weight of the evaluation index, _Zij is the element of the normalized decision matrix Z, and n represents the number of evaluation indexes.

5. Determine the positive and negative ideal solutions. The larger the value of element V _ij in the weighted normalized decision matrix V, the better the evaluation scheme j.

Positive ideal solution:

Negative ideal solution:

7. Determine the distance between the positive and negative ideal solutions and the optimal evaluation scheme for the development degree of fissure disease in each area to be studied. The distance S ⁺ _i of each evaluation scheme to the positive ideal solution V ⁺ and the distance S ^- _i to the negative ideal solution V ^- use the following formula:

8. Determine the closeness between the development degree of fissure disease in each research area and the positive and negative ideal solutions, as shown in Table 11. The 14 study area crack disease development degree scores are close to the positive ideal solution and the farthest from the negative ideal solution, so this score value is the best, the best evaluation scheme. To judge the optimal evaluation plan, the distance ^between the evaluation plan and the positive and negative ideal solutions must be considered _at the same time. The larger the value of the distance S ⁺ _i from the evaluation plan, the farther the distance from the optimal solution is. The farther the solution distance is; the most understood is the smaller the distance S ⁺ _i value and the larger the distance S ^- _i value. Set the proximity scale as C _i , and sort by relative proximity. The larger the value of C _i is, the higher the overall level is. The proximity C _i is between 0 and 1. When C _i =1, the performance level is the highest, reaching the optimal state; when C _i =0, there is no performance, and it is in a highly disordered and chaotic state. Calculated as follows:

Table 11 The optimal plan and positive and negative ideal solutions for the development degree of fissure disease in the study area

Specifically, the evaluation of the development degree of fissure disease in rammed earth sites is divided into four evaluation grades: less prone, low prone, medium prone, and high prone. The selection of the scoring value D interval and the explanation of the protective measures are shown in Table 12. It can be seen from Table 12 that when the scoring value D∈(0,0.2), the research area is a non-incidence area; when the scoring value D∈[0.2,0.35], the research area is a low-incidence area; when the scoring value D∈(0.35, 0.5], the research area is a medium-prone area; when the score value D ∈ (0.5,1), the research area is a high-prone area. The correspondence between the four evaluation grades of the crack disease development degree of rammed earth ruins and the score value D should be fully considered To understand the response relationship between the natural environment characteristics and the degree of development of fissure disease, to explore the dominant evaluation factors of internal and external factors, to select the optimal evaluation scheme, to establish a division standard for reasonable scoring values and evaluation grades, and to establish a grading system for the degree of development of fissure disease in rammed earth sites. Make a scientific and systematic assessment of the treatment of cracks in rammed earth.

Table 12 Criteria for grading and dividing the development degree of fissure disease in rammed earth sites

According to the TOPSIS approximation ideal solution method, the degree of fracture development of rammed earth sites in the study area is scored. According to the score value D Table 12, the fracture evaluation grades of the 14 research areas can be divided as shown in Table 13.

Table 13 Score value and ranking of the development degree of fissure disease in the study area

From Table 14 the score value D and evaluation grade of the development degree of fissure disease in the study area, it can be seen that through seven evaluation methods ① FAHP (w _i ) + multivariate instability index (w _1i ) + entropy value method (w _2i ) comprehensive weight + TOPSIS, ②FAHP(w _i )+TOPSIS, ③Multivariate instability index objective weight (w _1i )+TOPSIS, ④Entropy method (w _2i )+TOPSIS, ⑤FAHP(w _i )+Multivariate instability index (w _1i ) comprehensive weight + TOPSIS, ⑥ FHAP (w _i ) + entropy value method (w _2i ) comprehensive weight + TOPSIS, ⑦ multivariate instability index (w _1i ) + entropy value method (w _2i ) comprehensive weight + score value calculated by TOPSIS D for comparative analysis, different methods to calculate different weights or weight combinations combined with TOPSIS to calculate the scoring value to obtain the evaluation level. It can be seen from the score value D of the seven evaluation methods in the study area in Figure 4 that most of the study area is in the middle-susceptibility area and high-susceptibility area, a small part is in the low-susceptibility area, and there are no research points in the non-incidence area, which is consistent with the development of fissure disease The actual situation of the degree, the division of scoring value and the evaluation level are applicable to this evaluation method. The modes of the 7 evaluation levels in the 14 study areas are shown in Table 15, which are medium-prone area, low-prone area, high-prone area, medium-prone area, high-prone area, low-prone area, and medium-prone area Area, medium-prone area, medium-prone area, medium-prone area, medium-prone area, medium-prone area, medium-prone area, high-prone area, subjective weight + first objective weight + second objective weight combined with TOPSIS The modes of the seven evaluation grades are in line with the actual development of fissure disease and have certain application value. It can be seen from the score value chart of the seven evaluation methods in the study area in Figure 5 that according to the score values of the study area, the low-prone area to the high-prone area is sorted, and the subjective weight, the first objective weight, and the second objective weight are combined with TOPSIS The curve of the evaluation method has been in the middle of all the curves and is relatively stable. The volatility and dispersion of the curves of other evaluation methods are relatively large. Based on this, the evaluation method of subjective weight, first objective weight, and second objective weight combined with TOPSIS is obtained. It is suitable for evaluating the development degree of fissure disease in arid areas.

Table 14 Score value D and evaluation grade of the development degree of fissure disease by 7 evaluation methods in the study area

Continuing from the table

Table 15 Evaluation grade mode

Select Jiuquan, Zhangye, and Yongchang from the high-risk areas, medium-prone areas, and low-prone areas of the evaluation results for research and analysis, and compare the linear density, maximum gap opening, body density, and average gap length through Figure 6 and Figure 7 , average annual temperature, and drought index fissure disease characteristic value found: Jiuquan>Zhangye>Yongchang, which is consistent with the actual situation of the development degree of fissure disease; according to the combination of subjective and objective weights combined with the TOPSIS evaluation results, the high-prone area (Jiuquan) and the medium-prone area (Zhangye) and the low-prone area (Yongchang) are similar to the actual development scale of fissure disease, and the weight of annual average temperature and drought index as environmental factors is relatively large, which also shows that natural environmental factors play an important role in the development of fissure disease. From both quantitative and qualitative perspectives, it can be determined that the evaluation method of subjective weight + first objective weight + second objective weight combined with TOPSIS is suitable for the evaluation of the development degree of fissure disease.

Through the grade evaluation of the development degree of fissure disease in rammed earth sites in 14 research areas, it can be seen that most of the fissure diseases are in the middle-prone section, and none of the fissure diseases are in the low-risk section, indicating that the development rate and connectivity rate of wall fissure diseases are relatively high , but the risk is controllable, and most of them should take corresponding and reasonable protective measures. Datong and Jiuquan fissures are in the high-susceptibility zone, indicating that sufficient attention should be paid, and scientific and systematic protection measures should be taken before the fissure develops into a connected damage surface; Yongchang and Menyuan fissure disease are in the low-susceptibility zone, indicating that the development rate of fissure disease is slightly higher , in order to prevent the further development of the disease, it is enough to take corresponding protective measures.

From the ranking of subjective and objective comprehensive weights, we can see the importance level of each evaluation factor. The weight ranking of fracture connectivity ranks first. The density of fracture bodies in Datong, Jiuquan, and Honggu are 0.12mm ^-2 , 0.11mm ^-2 , and 0.07mm ^- respectively ² This also reflects the reason why the three are in the high susceptibility segment of fissure disease from the side. Therefore, the combination of subjective and objective comprehensive weights combined with the TOPSIS evaluation method can reveal the real situation of the development of crack diseases in rammed earth sites in the arid area of Northwest China, find out the main internal and external factors affecting the development of diseases, and then take corresponding protection measures to protect the site body scientifically and systematically. This method has certain accuracy and rationality when applied to the evaluation of crack disease development in rammed earth sites.

For the classification standard and evaluation grade of the development degree of fissure disease in rammed earth sites, the response relationship between the natural environment characteristics and the development degree of fissure disease is fully considered, and the evaluation factors dominated by internal and external factors are explored, the optimal evaluation scheme is selected, and a reasonable classification standard for scoring values is constructed. A discriminant standard was introduced to deal with the score values of the development degree of fissure disease in 14 study areas, and it was judged which area had the most serious development of fissure disease.

Step S4: Use machine learning BP neural network to build a BP neural network prediction model, use all the evaluation indicators of the evaluation index system as input data, and the evaluation results of step S3 as output data to predict the future development trend of crack diseases in multiple rammed earth ruins in arid areas of Northwest China Make predictions and verify the prediction results.

Combining the subjective and objective combined weight-TOPSIS evaluation results, the BP neural network machine learning method is applied to the protection of cracks in soil ruins, and the future development trend of cracks in rammed earth ruins in the arid area of Northwest China is predicted and early-warning, aiming to better predict the existing According to the trend of crack development and the possibility of crack damage in the wall body, corresponding preventive protection measures are put forward to prevent the protection of earthen sites from happening.

采用Matlab编程软件构建BP神经网络预测模型，表1中的研究区域为样本数据，随机选取训练样本约70％，检验样本约15％，预测样本约15％，见表16所示。其中，将评价指标C1-C11作为输入层数据、平分值D作为输出数据，训练算法选择levenberg-marquardt算法。数据样本训练、检验、预测阶段的输出结果如表16所示，通过训练阶段、检验阶段及预测阶段的实际评分值与预测评分值对比分析，检验预测的效果和精度，BP神经网络预测模型的输出结果如表17所示。The BP neural network prediction model for fissure disease in rammed earth ruins in the arid area of Northwest China, as shown in Figure 3, is divided into input layer, hidden layer, and output layer according to the structure. The first layer of the input layer includes 11 neurons, and the first layer of the output layer includes 1 neuron. Neurons, hidden layer 1 layer includes 4 neurons, hidden layer neurons are generally used:

Matlab programming software is used to construct the BP neural network prediction model. The research area in Table 1 is the sample data. About 70% of the training samples, about 15% of the test samples, and about 15% of the prediction samples are randomly selected, as shown in Table 16. Among them, the evaluation index C1-C11 is used as the input layer data, the split value D is used as the output data, and the training algorithm selects the levenberg-marquardt algorithm. The output results of the data sample training, testing, and prediction stages are shown in Table 16. Through the comparison and analysis of the actual scoring value and the predicted scoring value in the training stage, testing stage, and prediction stage, the effect and accuracy of the prediction are tested, and the BP neural network prediction model The output results are shown in Table 17.

Table 16 Evaluation grades of crack disease evaluation indicators of rammed earth sites in the arid area of Northwest China

Table 17 BP neural network prediction model output results

Check the results of the BP neural network prediction model. It can be seen from Figure 8 that the regression curve fitting effect between the actual score value D of the training sample and the score value T of the BP neural network prediction model is very good, and the correlation coefficient between the two is R=0.99062 . It can be seen from Figure 9 that the BP neural network prediction is performed on the overall sample, and the actual score value D and the predicted score value T are fitted, and the correlation coefficient of the regression curve between the two is R=0.96546.

Further test the accuracy of the BP neural network prediction model, introduce three evaluation methods of mean square error (MSE), absolute variance (R ² ) and relative root mean square error (RRMSE), and conduct accuracy checks on the training phase, testing phase and prediction phase. Test, the specific formula is as follows, and the results are shown in Table 18:

Table 18 Output detection results of BP neural network prediction model

Among them, D is the combination of subjective and objective weights - the actual score value of TOPSIS on the development degree of crack disease, T is the prediction score value of the BP neural network prediction model, and N is the number of samples in the training, testing and prediction stages. The values are 10 and 2 respectively. , 2. The smaller the mean square error (MSE) and the absolute relative root mean square error (RRMSE) value, the larger the absolute variance ^R2 , indicating the higher accuracy of the BP neural network prediction model. It can be seen from Table 18 that the correlation coefficients between the actual score value D and the predicted score value T are highly correlated in the training, testing, and prediction stages, the MSE and RRMSE values are relatively small, and the ^R2 values are close to 1, which is relatively large. Overall, it shows that the BP neural network prediction model established on the basis of the subjective and objective combination weight-TOPSIS evaluation method has high accuracy, and can be applied to the evaluation of fissure disease in rammed earth sites in the arid area of Northwest China, which has certain rationality and suitability.

The above descriptions are only preferred embodiments of the present invention, and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included in the scope of the present invention. within the scope of protection.

Claims (10)

1. A method for evaluating and predicting development degree of fracture diseases of rammed earth ruins is characterized by comprising the following steps:

step S1: selecting factors directly related to crack disease development to establish an evaluation index system, wherein the evaluation indexes comprise: the length of the gap, the opening degree of the gap, the communication rate of the gap and the characteristics of the natural environment;

step S2: calculating subjective weight according to an evaluation index system by using a fuzzy analytic hierarchy process, calculating first objective weight by using a multivariate unstable index method according to data of the evaluation index system of a research area, calculating second objective weight by using an improved entropy value method, and processing the subjective weight, the first objective weight and the second objective weight by equal weight weighted average to obtain comprehensive weight of the evaluation index;

the calculation method of the second objective weight comprises the following steps:

A1. normalization treatment:

wherein, X _ij The occupation importance degree of the j research area of the ith evaluation index is represented; x ^* _ij Expressing dimensionless matrix elements obtained after normalization of the negative correlation indexes; max _ij The maximum value in the range of the j study area of the ith evaluation index; min _ij A minimum value within the range of the j-th study region representing the ith evaluation index;

A2. evaluation index translation processing: x _i ′ _j ＝X ^* _ij +p；

Wherein, p is the translation amplitude of the evaluation index; x _i ′ _j Representing new matrix elements obtained by index translation of the dimensionless matrix elements after normalization;

A3. calculating the proportion of the jth sample in the ith evaluation index:

wherein n1 represents the total number of study areas;

A4. calculating the information entropy of the ith evaluation index:

A5. calculating the information utility value of the ith evaluation index: d is a radical of _i ＝1-e _i ；

A6. Calculating a second objective weight of the ith evaluation index as:

and step S3: evaluating the crack disease development grade by using a TOPSIS approximate ideal solution method: calculating Euclidean distances between each evaluation scheme and the positive and negative ideal solutions according to the comprehensive weight determined in the step S2, and evaluating the development degree of the fracture diseases of the rammed earth relics of the northwest arid regions through the Euclidean distances to obtain the development grade of the fracture diseases;

and step S4: and (4) constructing a BP neural network prediction model by using a machine learning BP neural network, predicting the future development trends of the rammed earth relic fracture diseases in the northwest drought regions by using all evaluation indexes of an evaluation index system as input data and the evaluation result of the step S3 as output data, and verifying the prediction results.

2. The method for evaluating and predicting the development degree of the fracture disease of the rammed earth site according to claim 1, wherein the gap length in the evaluation index comprises a total gap length and an average gap length; the gap opening comprises a maximum gap opening and an average gap opening; the fracture connectivity comprises bulk density and linear density; the natural environmental characteristics include the annual average temperature, annual average rainfall, annual average evaporation, drought index and annual average sunshine hours.

3. The method for evaluating and predicting the development degree of the fracture diseases of the rammed earth site according to claim 2, wherein the directionality of each evaluation index is judged to be negative.

4. The method for evaluating and predicting the development degree of the rammed earth site fissure disease according to claim 2 or 3, wherein the implementation method of the fuzzy analytic hierarchy process is as follows:

(1) Constructing an FAHP model: the crack disease development degree index layer comprises a target layer, a criterion layer and an index layer, wherein the target layer is the crack disease development degree A; the criterion layer comprises a gap length B1, a gap opening B2, a gap communication rate B3 and an environmental factor B4; the index layer comprises a total gap length C1, an average gap length C2, a maximum gap opening C3, an average gap opening C4, a bulk density C5, a linear density C6, an annual average temperature C7, an annual average rainfall C8, an annual average evaporation capacity C9, a drought index C10 and an annual average sunshine duration C11;

(2) Construction of fuzzy judgment matrix A = (a) by expert scoring _ij ) _n×n Wherein, the element a _ii ＝0.5；a _ij +a _ji ＝1，a _ij Not less than 0; n represents the number of selected evaluation indexes;

(3) Calculating subjective weights of all evaluation indexes according to the fuzzy judgment matrix A;

wherein is constant

r _i Sum of evaluation index scales r representing ith row of the blur determination matrix A _j Is represented by the formula _i The corresponding j rows of elements; r is a radical of hydrogen _ij Subjective weighting W of expression construction _i A matrix element of (a);

(4) And (3) checking consistency:

by weight vector (W) ₁ ,W ₂ ,W ₃ ..,W _i ,...W _n ) ^T Constructing a feature matrix W = (W) _ij ) _n×n Element W of the feature matrix W _ij Comprises the following steps:

checking the consistency of the fuzzy judgment matrix A and the characteristic matrix W:

construction of fuzzy complementary judgment matrix

Taking a threshold value alpha =0.1; the smaller the threshold alpha is, the higher the satisfaction degree of the fuzzy judgment matrix A is, and the higher the consistency requirement is.

5. The method for evaluating and predicting the development degree of the rammed earth site fissure disease according to claim 4, wherein the method for calculating the subjective weight comprises the following steps: calculating weights of the gap length B1, the gap opening B2, the gap communication rate B3 and the natural environment characteristic B4 of the criterion layer by constructing a fuzzy judgment matrix A1 about the gap length B1, the gap opening B2, the gap communication rate B3 and the natural environment characteristic B4 in the criterion layer through expert scoring; respectively constructing a fuzzy judgment matrix A2 of an index layer of a gap length B1, a fuzzy judgment matrix A3 of the index layer of a gap opening B2, a fuzzy judgment matrix A4 of the index layer of a gap communication rate B3 and a fuzzy judgment matrix A5 of the index layer of a natural environment characteristic B4 by expert scoring, calculating the weights of the total gap length C1 and the average gap length C2 of the evaluation indexes of the index layer according to the fuzzy judgment matrix A2, calculating the weights of the maximum gap opening C3 and the average gap opening C4 of the evaluation indexes of the index layer according to the fuzzy judgment matrix A3, calculating the weights of the volume density C5 and the linear density C6 of the evaluation indexes of the index layer according to the fuzzy judgment matrix A4, and calculating the weights of the annual average air temperature C7, the annual average rainfall C8, the annual average evaporation capacity C9, the drought index C10 and the annual average sunshine duration C11 of the evaluation indexes of the index layer according to the fuzzy judgment matrix A5; and multiplying the weight of the evaluation index of the index layer and the weight of the corresponding criterion layer to obtain the subjective weight of each evaluation index.

6. The method for evaluating and predicting the development degree of the fracture diseases of the rammed earth ruins according to claim 1 or 5, wherein the multivariate unsteady index method is used for analyzing the variation coefficient among the evaluation indexes in a research area in a statistical measurement mode, performing data normalization processing, and calculating the weight of each evaluation index according to the variation coefficient to serve as a first objective weight; the improved entropy method determines a second objective weight by normalizing the evaluation index data of each research area and then performing translation processing on the evaluation index.

7. The method for evaluating and predicting the development degree of the rammed earth ruin fissure disease according to claim 6, wherein the first objective weight is calculated by:

1) Normalizing the processed data: so that the data maps between [0,1 ]; since the directivities of the evaluation indexes are all negative, the data is normalized as:

wherein X represents the correspondence of the jth study region of the ith evaluation indexA numerical value; x ^* Expressing dimensionless matrix elements obtained after normalization of the negative correlation indexes; max represents the maximum value within the range of the j study region of the ith evaluation index; min represents the minimum value in the range of the j research area of the ith evaluation index;

2) Sample fraction analysis: the evaluation indexes are as follows:

wherein, U _ij Representing elements of the normalized dimensionless matrix; x _ij The occupation importance degree of the j research area of the ith evaluation index is represented;

3) Sample mean value

Comprises the following steps:

wherein x is ₁ 、x ₂ 、x _nl Is the degree of importance X of the ratio _ij Judging the elements with the importance degree ratio; n1 represents the number of regions of interest;

4) Calculate the sample standard deviation as:

wherein σ _i Represents the standard deviation;

5) Calculating the coefficient of variation of each evaluation index as:

6) Determining a first objective weight W of the evaluation index _1i Comprises the following steps:

8. the method for evaluating and predicting the development degree of the fracture diseases of the rammed earth ruins according to claim 7, wherein the subjective weight W is calculated by using an equal-weight weighted average method _i First objective weight W _1i And a second objective weight W _2i And combining to obtain the evaluation index of the development degree of the fracture disease of the rammed earth site, wherein the comprehensive weight is as follows:

9. the method for evaluating and predicting the development degree of the fissure diseases of the rammed earth ruins according to claim 2, 3, 5, 7 or 8, wherein the method for obtaining the development grade of the fissure diseases by approaching an ideal solution through TOPSIS in the step S3 comprises the following steps:

B1. establishing an evaluation matrix D1 of original data;

B2. normalizing the data of the evaluation matrix D1;

B3. constructing a normalized decision matrix Z, wherein:

wherein, Z _ij Representing elements in a normalized decision matrix Z;

B4. constructing a weighted normalized decision matrix V, wherein the element V _ij ＝λ _i ×Z _ij ，λ _i Is the integrated weight of the ith evaluation index, Z _ij Elements of a normalized decision matrix Z; n represents the number of evaluation indexes;

B5. determining the positive ideal solution and the negative ideal solution as follows:

the positive ideal solution:

negative ideal solution:

B6. determining the distance between the positive and negative ideal solutions and the fissure development degree of each research area:

j study area crack development degree to positive ideal solution V ⁺ S distance of ⁺ _j Comprises the following steps:

the development degree of the fissure of the jth research area reaches the negative ideal solution V ^- S distance of ^- _j Comprises the following steps:

B7. determining the closeness of the fissure development degree of each research area to the positive and negative ideal solution

Proximity is the score value D;

B8. when the score value D belongs to (0, 0.2), the research area is a non-germination area; when the score value D belongs to [0.2,0.35], the research area is a low-susceptibility area; when the score value D belongs to (0.35, 0.5), the research region is a medium-easy region, and when the score value D belongs to (0.5, 1), the research region is a high-easy region.

10. The method for evaluating and predicting the development degree of the rammed earth site fissure disease according to claim 9, wherein the BP neural network prediction model is constructed by Matlab programming software, data of total gap length C1, average gap length C2, maximum gap opening C3, average gap opening C4, body density C5, linear density C6, annual average temperature C7, annual average rainfall C8, annual average evaporation C9, drought index C10 and annual average sunshine duration C11 of evaluation indexes are used as input layer data, a bisection value D is used as output data, the input layer of the BP neural network prediction model comprises 11 neurons, the output layer comprises 1 neuron and the hidden layer comprises 4 neurons, and a levenberg-marquardt algorithm is selected for training of the BP neural network prediction model; data of evaluation indexes of the research area are sample data, training samples are randomly selected to be 70% of the sample data, inspection samples are 15% of the sample data, and prediction samples are 15% of the sample data.

Images