Double-row slide-resistant pile earthquake damage dynamic evaluation method based on LSTM-entropy weight TOPSIS

Abstract

The invention discloses a double-row slide-resistant pile earthquake damage dynamic evaluation method based on LSTM-entropy weight TOPSIS, and relates to the technical field of earthquake damage dynamic evaluation, and the method comprises the steps: obtaining acceleration data and dynamic earth pressure data of a slide-resistant pile model under the action of earthquake waves; constructing a displacement prediction model on the basis of a long short-term memory (LSTM) network, and obtaining predicted displacement data in combination with the acceleration data and the dynamic earth pressure data; evaluating the prediction effect of the prediction displacement data according to the evaluation index; according to the method, a damage evaluation model is constructed according to an entropy weight method and an approximate ideal solution sorting method (TOPSIS), movable earth pressure data and predicted displacement data are combined, standardized index data and index weights are obtained, then weighted index data are obtained, the correlation closeness degree of the anti-slide pile damage state and the ideal damage state is calculated in combination with an Euclidean distance method, and the damage evaluation result of the anti-slide pile damage state and the ideal damage state is obtained. And determining a damage evaluation result according to the correlation close degree. According to the invention, the accuracy, precision and comprehensiveness of the earthquake damage evaluation method can be improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of seismic damage dynamic evaluation, and in particular to a double-row anti-slide pile seismic damage dynamic evaluation method based on LSTM-entropy weight TOPSIS. BACKGROUND

[0002] China is located in the intersection area of the Pacific Ring Seismic Belt and the Eurasian Seismic Belt. Due to the frequent seismic activity in this area, more than 20 earthquakes with a magnitude of 5 or above occur annually, so the seismic damage evaluation of this area is crucial. As a widely used supporting structure in slope engineering, the performance of anti-slide piles is directly related to the safety of the project, so the study of the seismic performance of anti-slide piles is an important part of seismic damage evaluation.

[0003] In the prior art, for the failure mechanism of anti-slide piles, Zhu Dan et al. revealed the failure mechanism of anti-slide piles under the action of different seismic parameters through systematic shaking table tests; Kong Siyu established a quantitative mapping relationship between structural crack characteristics and damage state based on machine learning algorithms; Mao Chenxi and Tang Hua respectively used machine learning and deep learning methods to realize intelligent evaluation of the damage degree of structures. For the damage identification and evaluation of anti-slide piles, Wang Peiyu et al. established a quantitative relationship between the residual displacement of the top of the anti-slide pile and the relative damage degree based on a three-dimensional finite difference model, and proposed a damage discrimination index and a damage calculation method for anti-slide piles; Tian Hongcheng combined the three-dimensional Fast Lagrangian Analysis of Continua (FLAC3D) platform to establish an anti-slide pile dynamic damage positioning method, and obtained a damage amount calculation formula; Su Hang established an anti-slide pile seismic damage evaluation index system by analyzing typical failure modes, and constructed a damage evaluation model based on the coupling theory of hierarchical-fuzzy mathematics. The existing technologies mainly focus on the damage mechanism of anti-slide piles, and there is less research on the identification of hidden damage of anti-slide piles under seismic action. However, such hidden damage can significantly reduce the long-term service performance, and there is less research on the collaborative application of deep learning algorithms and multi-criteria decision-making methods in the seismic damage state evaluation of anti-slide piles. Such a fusion method can consider both time-varying damage characteristics and multi-index synergistic effects, and can break through the limitations of traditional evaluation methods.

[0004] Therefore, it is urgent to establish a more scientific anti-slide pile seismic damage evaluation method to solve the problems of insufficient accuracy and comprehensiveness of existing evaluation methods, so as to realize accurate evaluation of the safety state of seismic damage structures and ensure the reliability of engineering seismic resistance. SUMMARY

[0005] The embodiment of the present application provides a double-row anti-slide pile seismic damage dynamic evaluation method based on LSTM-entropy weight TOPSIS.

[0006] The technical scheme of the embodiments of the present application is implemented as follows: In a first aspect, the embodiments of the present application provide a double-row anti-slide pile seismic damage dynamic evaluation method based on LSTM-entropy weight TOPSIS, which comprises the following steps: obtaining acceleration data and dynamic soil pressure data of an anti-slide pile model under the action of seismic waves; constructing a displacement prediction model based on an LSTM network, combining the acceleration data and the dynamic soil pressure data to obtain predicted displacement data; evaluating the prediction effect of the predicted displacement data according to evaluation indexes, wherein the evaluation indexes include mean square error, mean absolute error, mean absolute error ratio and determination coefficient; constructing a damage evaluation model according to an entropy weight method and TOPSIS, combining the dynamic soil pressure data and the predicted displacement data to obtain standardized index data and index weights; obtaining weighted index data according to the standardized index data and the index weights, calculating the correlation closeness between the damage state of the anti-slide pile and the ideal damage state by using the Euclidean distance method, and determining the damage evaluation result according to the correlation closeness.

[0007] The technical scheme provided by the present application firstly obtains acceleration data and dynamic soil pressure data of an anti-slide pile model under the action of seismic waves, and then constructs a displacement prediction model based on an LSTM network, and combines the acceleration data and the dynamic soil pressure data to obtain predicted displacement data. Since the displacement prediction model has stronger time sequence feature extraction capability and nonlinear fitting capability, the prediction accuracy can be improved by obtaining the predicted displacement data from the displacement prediction model, so that the long-term dependence characteristics of the seismic response data can be more effectively captured. Then, the prediction effect of the predicted displacement data is evaluated according to evaluation indexes, which include mean square error, mean absolute error, mean absolute error ratio and determination coefficient. The prediction effect of the predicted displacement data is comprehensively evaluated, which ensures the accuracy of subsequent damage evaluation using the predicted displacement data. Then, a damage evaluation model is constructed according to an entropy weight method and TOPSIS, and the dynamic soil pressure data and the predicted displacement data are combined to obtain standardized index data and index weights. This can effectively solve the problems of non-uniformity of multiple index dimensions and large numerical distribution span, ensure the objectivity and mathematical interpretability of weight distribution, and avoid the influence of subjective bias on the evaluation result. Finally, weighted index data is obtained according to the standardized index data and the index weights, the correlation closeness between the damage state of the anti-slide pile and the ideal damage state is calculated by using the Euclidean distance method, and the damage evaluation result is determined according to the correlation closeness. The closer the correlation closeness is to 1, the lower the seismic damage degree of the anti-slide pile is. The closer the correlation closeness is to 0, the higher the seismic damage degree of the anti-slide pile is. The technical scheme provided by the present application improves the accuracy, precision and comprehensiveness of the seismic damage evaluation method, and finally realizes accurate evaluation of the seismic damage state, ensuring the reliability of engineering seismic resistance.

[0008] Optionally, the displacement prediction model comprises an input layer, a hidden layer, a long short-term memory network layer, and an output layer, and the long short-term memory network layer comprises a forget gate, an input gate, and an output gate.

[0009] Optionally, the displacement prediction model is constructed based on an LSTM network, and the predicted displacement data is obtained by combining the acceleration data and the dynamic soil pressure data, including: preprocessing the acceleration data and the dynamic soil pressure data to obtain time series data; determining the parameter configuration of the displacement prediction model; receiving the time series data through the input layer and transmitting the time series data to the long short-term memory network layer; in the long short-term memory network layer, time series feature extraction is performed on the time series data; the long short-term memory network layer is connected to the full connection layer, and the time series feature extraction result is linearly transformed and mapped by a nonlinear activation function to generate the predicted displacement data by the output layer.

[0010] Optionally, the damage evaluation model is constructed according to the entropy weight method and the TOPSIS, and the standardized index data and the index weight are obtained by combining the dynamic soil pressure data and the predicted displacement data, including: the dynamic soil pressure data and the predicted displacement data are standardized by the range method to obtain the standardized index data, and the standardization processing includes positive index processing and reverse index processing; the index information entropy is calculated according to the standardized index data; and the index weight is calculated according to the index information entropy.

[0011] Optionally, the weighted index data is obtained according to the standardized index data and the index weight, and the correlation closeness between the anti-slide pile damage state and the ideal damage state is calculated by combining the Euclidean distance method, including: the positive ideal solution and the negative ideal solution are determined according to the weighted index data; the positive Euclidean distance and the negative Euclidean distance are calculated according to the positive ideal solution and the negative ideal solution by combining the Euclidean distance method; and the correlation closeness is calculated according to the positive Euclidean distance and the negative Euclidean distance.

[0012] In a second aspect, the embodiments of the present application provide an electronic device, comprising a memory and a processor, the memory stores a computer program capable of running on the processor, and the processor implements the steps of the above-mentioned one kind of double-row anti-slide pile seismic damage dynamic evaluation method based on LSTM-entropy weight TOPSIS when executing the program.

[0013] In a third aspect, the embodiments of the present application provide a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the steps of the above-mentioned one kind of double-row anti-slide pile seismic damage dynamic evaluation method based on LSTM-entropy weight TOPSIS.

[0014] The technical scheme provided by the embodiments of the present application has at least the following beneficial effects: The application provides a double-row anti-slide pile seismic damage dynamic evaluation method based on LSTM-entropy weight TOPSIS, first, acceleration data and dynamic soil pressure data of the anti-slide pile model under the action of seismic waves are obtained, then, a displacement prediction model is constructed based on an LSTM network, and prediction displacement data is obtained by combining the acceleration data and the dynamic soil pressure data, since the displacement prediction model has stronger time sequence feature extraction capability and nonlinear fitting capability, the prediction displacement data obtained from the displacement prediction model can improve the prediction accuracy, so that the long-term dependence characteristics of the seismic response data can be more effectively captured, then, the prediction effect of the prediction displacement data is evaluated according to an evaluation index, the evaluation index includes mean square error, mean absolute error, mean absolute error ratio and determination coefficient, the prediction effect of the prediction displacement data is comprehensively evaluated, and the accuracy of subsequent damage evaluation by using the prediction displacement data is ensured, then, a damage evaluation model is constructed according to the entropy weight method and TOPSIS, standardized index data and index weight are obtained by combining the dynamic soil pressure data and the prediction displacement data, the problem of non-uniform dimensions and large numerical distribution span of multiple indexes can be effectively solved, the objectivity and mathematical explainability of weight distribution are ensured, and the influence of subjective bias on the evaluation result is avoided, finally, weighted index data are obtained according to the standardized index data and the index weight, the correlation closeness of the anti-slide pile damage state and the ideal damage state is calculated by combining the Euclidean distance method, and the damage evaluation result is determined according to the correlation closeness, the closer the correlation closeness is to 1, the lower the seismic damage degree of the anti-slide pile is, and the closer the correlation closeness is to 0, the higher the seismic damage degree of the anti-slide pile is, the technical scheme provided by the application improves the accuracy, precision and comprehensiveness of the seismic damage evaluation method, and finally realizes accurate evaluation of the seismic damage state, and ensures the reliability of engineering seismic resistance. BRIEF DESCRIPTION OF DRAWINGS

[0015] In order to more clearly illustrate the technical solutions in the embodiments of the application, the following will briefly introduce the drawings needed to be used in the embodiment description. Obviously, the drawings in the following description are only some embodiments of the application, and for those skilled in the art, other drawings can also be obtained from these drawings without creative labor. Figure 1 A flowchart of a double-row anti-slide pile seismic damage dynamic evaluation method based on LSTM-entropy weight TOPSIS provided by the embodiments of the application; Figure 2 A schematic diagram of a seismic wave provided by the embodiments of the application; Figure 3 A result schematic diagram of denoising processing for acceleration data provided by the embodiments of the application; Figure 4 A result schematic diagram of denoising processing for dynamic soil pressure data provided by the embodiments of the application; Figure 5 A schematic diagram of an LSTM network algorithm effect comparison result provided by an embodiment of the present application is shown in the following table. Figure 6 A hardware entity schematic diagram of an electronic device provided by an embodiment of the present application is shown in the following table. DETAILED DESCRIPTION

[0016] To make the objectives, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described below in conjunction with the accompanying drawings of the embodiments of the present application. Obviously, the described embodiments are some of the embodiments of the present application but not all the embodiments of the present application. The following embodiments are used to illustrate the present application but not to limit the scope of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the scope of the present application.

[0017] In the following description, “some embodiments” are described, which describe a subset of all possible embodiments, but it can be understood that “some embodiments” can be the same subset or different subsets of all possible embodiments, and can be combined with each other without conflict.

[0018] It should be noted that the terms “first\second\third” involved in the embodiments of the present application are only to distinguish similar objects and do not represent a specific order of the objects. It can be understood that “first\second\third” can be interchanged with a specific order or sequence as allowed, so that the embodiments of the present application described here can be implemented in an order other than that illustrated or described here.

[0019] Those skilled in the art can understand that, unless otherwise defined, all terms (including technical terms and scientific terms) used herein have the same meaning as generally understood by those skilled in the art to which the embodiments of the present application belong. It should also be understood that terms such as those defined in general dictionaries should be understood as having meanings consistent with those in the context of the prior art, and should not be interpreted in an idealized or overly formal sense unless specifically defined as such.

[0020] The embodiments of the present application will be further described below in conjunction with the accompanying drawings.

[0021] In view of the problems existing in the field of dynamic evaluation of seismic damage of double-row anti-slide piles, the embodiments of the present application provide a dynamic evaluation method for seismic damage of double-row anti-slide piles based on LSTM-entropy weight TOPSIS.

[0022] The technical solutions of the present application will be introduced below. First, the method embodiments of the present application will be introduced.

[0023] Please refer to Figure 1 It illustrates a flowchart of a dynamic evaluation method for seismic damage of double-row anti-slide piles based on LSTM-entropy weighted TOPSIS provided in an embodiment of this application, as shown below. Figure 1 As shown, the method includes at least the following steps S110 to S150.

[0024] Step S110: Obtain the acceleration data and dynamic earth pressure data of the anti-slide pile model under the action of seismic waves.

[0025] In this embodiment, an earthquake damage performance test was conducted on double-row anti-slide piles. The test used an earthquake simulation shaking table from a local earthquake research institute, which has multi-directional seismic wave simulation capabilities and can emit seismic waves into the anti-slide pile model to simulate actual earthquake conditions. In one embodiment, the technical parameters of the earthquake simulation shaking table are shown in Table 1 below.

[0026] Table 1 (Technical parameters of the earthquake simulation shaking table) The earthquake simulation shaking table measures 400 mm in length and 600 mm in width; its maximum load capacity is 25 tons; its maximum acceleration is 1.7 m / s²; its maximum velocity is 1.5 m / s; and its maximum movable displacement is 250 mm.

[0027] Because the actual structural scale of the double-row anti-slide pile seismic damage performance test is very large, the feasibility of directly conducting a full-scale test is low. Therefore, based on similarity theory, a scaled-down anti-slide pile test model is made to adapt it to shaking table testing, which greatly improves the feasibility. The geometric dimensions of the anti-slide pile test model are proportionally scaled to the geometric dimensions of its prototype structure, and its material properties (e.g., elastic modulus, mass density, and Poisson's ratio) and time parameters (e.g., vibration frequency and period) maintain a strict similarity relationship with the prototype structure to ensure that the seismic damage performance test conducted based on the anti-slide pile test model can accurately reflect the response of the prototype structure in an earthquake. In one embodiment, detailed test similarity examples are shown in Table 2 below.

[0028] Table 2 (Trial Similarity Ratio) Because seismic damage performance tests involve scaling down real-world slopes using similarity theory and placing them within an anti-slide pile test model box, boundary issues can easily arise, affecting the accuracy and reliability of the test. Therefore, a sliding, flexible, and friction boundary treatment method is adopted. Specifically, plexiglass is installed on the left and right sides of the anti-slide pile test model box to form a sliding boundary, providing impact protection and facilitating observation. Polystyrene foam boards are placed on the front and rear inner walls as a damping layer to construct a flexible boundary, reducing seismic wave reflection interference. A 2-cm thick layer of gravel and coarse sand is laid at the bottom of the anti-slide pile test model box to form a friction boundary, increasing friction and reducing the slippage error between the anti-slide pile test model and the base plate. This ensures that the seismic damage performance test accurately reflects the condition of the prototype structure, providing a basis for seismic design and seismic damage assessment of engineering projects.

[0029] In this embodiment, the anti-slide pile test model box has geometric dimensions of 300 cm in length, 140 cm in width, and 114 cm in height. Its main structure is welded from 20 mm thick steel plates, and the four side walls are reinforced with steel plates of the same thickness. Furthermore, the observation surface of the anti-slide pile test model box is equipped with a 20 mm thick transparent fiberglass window, through which test personnel can monitor the dynamic response process of the model under seismic load in real time.

[0030] Furthermore, the anti-slide pile test model was generalized, and the sliding surface was set at a scale of 1:1.73 to simulate the sliding characteristics of the actual slope. During model preparation, similar raw materials were placed layer by layer into the model box, and compaction was used to ensure the density and uniformity of each layer to reproduce the physical properties of the actual slope. The raw materials included sand, bentonite, gypsum, talc, glycerin, and water. Sand was used to provide skeletal support and increase the internal friction angle and unit weight; talc and bentonite were used to adjust density, elastic modulus, and compressive strength; glycerin was used to reduce compressive strength; and water was used to regulate cohesion and moisture content. In addition, materials were configured according to similarity theory to construct the sliding body, bedrock, and anti-slide piles. In one embodiment, the specific proportions and physical quantities of the sliding body, bedrock, and anti-slide piles are shown in Table 3 below.

[0031] Table 3 (Specific proportions and physical quantities of sliding body, bedrock and anti-slide piles) Furthermore, acceleration sensors and earth pressure sensors are arranged at the designated locations on the anti-slide pile test model according to the design, so as to realize real-time monitoring of the dynamic response and earth pressure changes of the model under seismic waves or other loads, and to obtain acceleration data and dynamic earth pressure data.

[0032] In an optional embodiment, a seismic simulation shaking table is used to emit seismic waves in the anti-slide pile model to simulate actual earthquake conditions. For example, El Centro waves, which have a wide frequency representation, are used as seismic waves to simulate the complexities of actual earthquakes. Please refer to [reference needed].Figure 2 This document illustrates a seismic wave diagram provided in an embodiment of this application. The seismic wave is an EL Centro wave, with loading directions in the horizontal (X-direction) and vertical (Z-direction) directions. The ground motion intensities are 0.1g, 0.2g, and 0.4g (g is the gravitational acceleration, taken as 9.8 m / s²), totaling six loading conditions. Since white noise has a uniform power spectral density and can comprehensively reflect the dynamic response characteristics of the structure, a sinusoidal frequency sweep of 0.05g white noise is used before the seismic damage performance test to excite the various vibration modes of the model structure, thereby obtaining the initial dynamic characteristics. In the seismic damage performance test, a white noise frequency sweep test is performed after each level of seismic wave loading to provide benchmark data for changes in the model's dynamic characteristics. In a specific example, according to the loading regime, the test collected acceleration data and earth pressure data on the riverside, mountainside, and mountainside of the anti-slide piles. The acceleration data on the riverside of the anti-slide piles is... The earth pressure data is The acceleration data on the mountainside are: The earth pressure data is The acceleration data of the anti-slide piles on the mountainside are as follows: Earth pressure data is .

[0033] Step S120: Construct a displacement prediction model based on a long short-term memory network, and combine the acceleration data and the earth pressure data to obtain predicted displacement data.

[0034] In a specific embodiment, a displacement prediction model is first constructed based on an LSTM network. The LSTM network includes an input layer, a hidden layer, and an output layer. The core of the LSTM network is to regulate the flow and update of data through gating mechanisms such as the forget gate, the input gate, and the output gate. Specifically, the forget gate is used to filter redundant data, the input gate is used to determine the update of new data, and the output gate is used to extract the current hidden state. The calculation formulas for the output update of the forget gate, the input gate, and the output gate are expressed by the following formula (1): Formula (1); In the formula, This represents the output data of the forget gate; This represents the output data of the input gate; This indicates the output data of the output gate; The weight matrix representing the forget gate; This represents the weight matrix of the input gate; This represents the weight matrix of the output gate; The bias term representing the forget gate; This represents the bias term of the input gate; This represents the bias term of the output gate. represent The hidden state at any given moment; This indicates the hidden state of the previous time step; This represents the sigmoid activation function.

[0035] Furthermore, the displacement prediction model includes an input layer, a hidden layer, a long short-term memory (LSTM) network layer, and an output layer. The LSM network layer includes a forget gate, an input gate, and an output gate. The steps for obtaining predicted displacement data based on the displacement prediction model and combining acceleration and earth pressure data are as follows: First, the acceleration and earth pressure data are preprocessed to obtain time series data. Specifically, the preprocessing operations include denoising and time-segmentation. The original acceleration and earth pressure data typically contain noisy and irrelevant data. Denoising is performed to remove noise and irrelevant information, resulting in denoised data. For example, high-frequency noise can be eliminated through bandpass filtering using Fast Fourier Transform (FFT), and baseline correction can address baseline drift caused by low-frequency noise, ensuring the performance of the displacement prediction model. Finally, the acceleration data is converted into velocity information through integration, which serves as one of the inputs to the displacement prediction model. Please refer to [reference needed]. Figure 3 This illustration shows a schematic diagram of the result of denoising acceleration data according to an embodiment of this application, including acceleration data before denoising (i.e., the original data), acceleration data after filtering, and acceleration data after baseline correction. Please refer to... Figure 4 The illustration shows a schematic diagram of the result of denoising earth pressure data provided in an embodiment of this application, including earth pressure data before denoising (i.e., the original data), earth pressure data after filtering, and earth pressure data after baseline correction.

[0036] Furthermore, since the time series data collected by each sensor differ in length, to avoid the impact of inconsistent dataset size on the results, the acceleration and earth pressure data are truncated over time periods to obtain unified time series data. The technical solution provided in this application, by preprocessing the acceleration and earth pressure data, can more accurately reflect the true characteristics of the system, thereby improving prediction accuracy.

[0037] Furthermore, since the selection of hyperparameters is crucial to the training effect, convergence speed and generalization ability of the displacement prediction model, a set of hyperparameter configurations is determined through multiple iterations of optimization to achieve a balance between model performance and computational efficiency. The parameter settings of the displacement prediction model are shown in Table 4 below.

[0038] Table 4 (Parameter Settings for Displacement Prediction Model) Furthermore, the preprocessed time-series data is received through the input layer and passed to the Long Short-Term Memory (LSTM) network layer. The LTM network layer, through a unit structure containing forget gates, input gates, and output gates, performs temporal feature extraction on the preprocessed time-series data by combining cell states and previous hidden states. After processing by multiple levels of LTM network layers, the temporal feature extraction results are linearly transformed and mapped by nonlinear activation functions through a fully connected layer. Finally, the output layer generates the predicted displacement data.

[0039] Step S130: Evaluate the prediction effect of the predicted displacement data according to the evaluation indicators, which include mean square error, mean absolute error, mean absolute error ratio and coefficient of determination.

[0040] In specific embodiments, the prediction results (i.e., predicted displacement data) of the displacement prediction model are comprehensively judged through multiple evaluation indicators. Optionally, mean square error, mean absolute error, mean absolute error ratio, and coefficient of determination are used as evaluation indicators to assess the prediction effect of the predicted displacement data. Among them, mean square error is the mean of the squares of the deviations between the predicted displacement data and the actual displacement data. The closer the mean square error is to 0, the higher the prediction accuracy and the more sensitive it is to large errors. Mean absolute error is the average of the absolute values ​​of the errors between the predicted displacement data and the actual displacement data. Mean absolute error is not sensitive to outliers and has strong robustness. The smaller its value, the higher the accuracy of the displacement prediction model. Mean absolute error ratio is used to measure the relative error of the predicted displacement data. The smaller the mean absolute error ratio, the more accurate the displacement prediction model. Generally, when the mean absolute error ratio is greater than 20%, the prediction effect is considered unsatisfactory. Coefficient of determination is used to evaluate the interpretability and predictive ability of the predicted displacement data. Its value range is [0,1]. The closer the coefficient of determination is to 1, the better the data fitting effect. The formulas for calculating the mean square error, mean absolute error, mean absolute error ratio, and coefficient of determination are expressed by the following formula (2): Formula (2); In the formula, Indicates mean square error; Indicates the mean absolute error; Indicates the mean absolute error ratio; Indicates the coefficient of determination; Indicates the number of samples; Represents actual displacement data; This represents the predicted displacement data; This represents the average displacement.

[0041] In specific embodiments, the LSTM network algorithm is compared and analyzed with the traditional back propagation (BP) algorithm and the genetic algorithm-optimized back propagation neural network (GA-BP) algorithm. Please refer to [link / reference]. Figure 5 This diagram illustrates a comparison of the performance of an LSTM network algorithm according to an embodiment of this application. The horizontal axis represents the evaluation index, and the vertical axis represents the evaluation index value. The LSTM network algorithm is shown in the mean square error (MSE)... Mean absolute error ( ), mean absolute error ratio ( ) and coefficient of determination ( It significantly outperforms the BP algorithm and GA-BP algorithm in all four evaluation metrics, especially in the coefficient of determination (COP). In terms of performance indicators, the LSTM network algorithm's value is closer to 1, indicating that the displacement prediction model built based on the LSTM network algorithm has higher prediction accuracy and data fitting ability. The overall results show that the LSTM network algorithm can more effectively capture the long-term dependence characteristics of seismic response data, and has stronger temporal feature extraction and nonlinear fitting capabilities. In other words, the LSTM network algorithm is superior, and the displacement prediction model it builds has higher prediction accuracy and data fitting ability. It can more effectively capture the long-term dependence characteristics of seismic response data, and has stronger temporal feature extraction and nonlinear fitting capabilities.

[0042] In one embodiment, under the condition of 0.4X (where X is the design reference peak ground acceleration), data from a certain monitoring point is used as the input to the model training set to train the displacement prediction model. In the training results, the mean square error between the predicted displacement data and the actual displacement data is 0.0049353, which is close to 0. The overlap between the predicted and actual displacement data is extremely high, and the coefficient of determination between the predicted and actual displacement data is as high as 0.9995, indicating that the displacement prediction model can capture data characteristics and trends well. In another embodiment, in the training results of different sample points, the mean square error between the predicted and actual displacement data is also 0.0049353, which is close to 0. The overlap between the predicted and actual displacement data is extremely high, and the coefficient of determination between the predicted and actual displacement data is as high as 0.9985, indicating that the displacement prediction model can not only fit the training data, but also generalize well to new data.

[0043] Step S140: Construct a damage evaluation model based on the entropy weight method and the approximation ideal solution ranking method, and obtain standardized index data and index weights by combining the dynamic earth pressure data and the predicted displacement data.

[0044] In a specific embodiment, the key to earthquake damage assessment of anti-slide piles lies in establishing a scientific weight allocation system to accurately quantify the contribution of each indicator data to the damage state. Combining the entropy weight method and TOPSIS to construct a damage assessment model enables a comprehensive evaluation combining information entropy theory and multi-attribute decision analysis. This effectively solves the problems of inconsistent dimensions and large numerical distribution spans among multiple indicators. The entropy weight method, based on information entropy theory, dynamically allocates weights by quantifying the dispersion of indicator data distribution. It eliminates dimensional differences using range standardization and avoids subjective experience interference through a data-driven mechanism, ensuring the objectivity and mathematical interpretability of weight allocation. This method can objectively extract the contribution of key indicators from earth pressure data and predicted displacement data, thereby avoiding the influence of subjective bias on the evaluation results.

[0045] In a specific embodiment, the earth pressure data and predicted displacement data are first standardized using the range method to obtain standardized index data. A standardized index data matrix is ​​then constructed based on the standardized index data. The standardization process includes positive index processing and negative index processing. The earth pressure data and predicted displacement data include positive and negative indices. Positive index processing is performed on the positive indices, and negative index processing is performed on the negative indices. The calculation formula for the standardized index data is expressed by the following formula (3): Formula (3); In the formula, Indicates the first The first sample data The raw data for each indicator , ; Represents standardized indicator data; This represents the minimum value in the standardized indicator data; This represents the maximum value in the standardized indicator data.

[0046] Furthermore, the information entropy of the indicators is calculated based on the standardized indicator data. The information entropy of the indicators is used to measure the degree of disorder or information content of the standardized indicator data. The formula for calculating the information entropy of the indicators is expressed by the following formula (4): Formula (4); In the formula, This represents the entropy of the indicator information; This represents the constant used for normalization. ; Indicates the percentage of indicators. And regulations .

[0047] Furthermore, the indicator weights are calculated based on the indicator information entropy of each indicator. The formula for calculating the indicator weights is expressed by the following formula (5): Formula (5); In the formula, Indicates the weight of the indicator.

[0048] Step S150: Obtain weighted index data based on the standardized index data and the index weights, calculate the correlation between the damage state of the anti-slide pile and the ideal damage state using the Euclidean distance method, and determine the damage evaluation result based on the correlation.

[0049] In a specific embodiment, weighted index data is first obtained based on standardized index data and index weights. The standardized index data matrix is ​​then multiplied by the index weights of each index to obtain the weighted matrix. The calculation formula for the weighted index data is expressed by the following formula (6): Formula (6); In the formula, This represents weighted index data; Represents standardized indicator data; Indicates the weight of the indicator.

[0050] Furthermore, the correlation between the damage state and the ideal damage state of the anti-slide pile is calculated using the Euclidean distance method. The main steps are as follows: First, the positive and negative ideal solutions are determined based on the weighted index data. Then, the positive and negative Euclidean distances are calculated using the Euclidean distance method based on the positive and negative ideal solutions. The formulas for calculating the positive and negative Euclidean distances are expressed by the following formula (7): Formula (7); In the formula, Indicates positive Euclidean distance; Indicates negative Euclidean distance; This represents weighted index data; This represents the maximum value in the weighted index data; This represents the minimum value in the weighted index data.

[0051] Furthermore, the correlation closeness is calculated based on the positive and negative Euclidean distances. The formula for calculating the correlation closeness is expressed by the following formula (8): Formula (8); In the formula, This indicates the degree of closeness between the damage state of the anti-slide pile and the ideal damage state. Indicates positive Euclidean distance; This represents the negative Euclidean distance.

[0052] Finally, the damage assessment result is determined based on the relevance score. Specifically, the relevance score... The value ranges from 0 to 1, and the relevance is... The closer the value is to 1, the closer the weighted index data is to the positive ideal solution, indicating that the anti-slide pile performs better under comprehensive consideration and its seismic damage degree is lower; correlation closeness The closer the weighted index is to 0, the closer it is to the negative ideal solution, indicating that the anti-slide pile performs worse under comprehensive consideration and the corresponding degree of earthquake damage is higher.

[0053] In summary, the seismic damage dynamic evaluation method for double-row anti-slide piles based on LSTM-entropy weighted TOPSIS provided in this application first acquires the acceleration and earth pressure data of the anti-slide pile model under seismic wave action. Then, a displacement prediction model is constructed based on an LSTM network, and combined with the acceleration and earth pressure data to obtain predicted displacement data. Since the displacement prediction model has stronger temporal feature extraction and nonlinear fitting capabilities, the predicted displacement data obtained from the displacement prediction model can improve prediction accuracy, thereby more effectively capturing the long-term dependence characteristics of seismic response data. Finally, the prediction effect of the predicted displacement data is evaluated according to evaluation indicators, including mean square error, mean absolute error, mean absolute error ratio, and coefficient of determination. This comprehensive evaluation ensures the accuracy of subsequent damage assessment using the predicted displacement data. The method involves several steps: First, a damage assessment model is constructed based on the entropy weight method and TOPSIS. Combined with dynamic earth pressure data and predicted displacement data, standardized index data and index weights are obtained. This effectively solves the problems of inconsistent dimensions and large numerical distribution spans among multiple indicators, ensuring the objectivity and mathematical interpretability of weight allocation and avoiding the influence of subjective bias on the evaluation results. Finally, weighted index data is obtained based on the standardized index data and index weights. The correlation between the damage state of the anti-slide pile and the ideal damage state is calculated using the Euclidean distance method. The damage assessment result is determined based on the correlation correlation. The closer the correlation correlation is to 1, the lower the degree of earthquake damage to the anti-slide pile; the closer the correlation correlation is to 0, the higher the degree of earthquake damage to the anti-slide pile. The technical solution provided in this application improves the accuracy, precision, and comprehensiveness of the earthquake damage assessment method, ultimately achieving an accurate evaluation of the earthquake damage state and ensuring the reliability of the engineering seismic resistance.

[0054] It should be noted that, in the embodiments of this application, if the above-mentioned dynamic evaluation method for seismic damage of double-row anti-slide piles based on LSTM-entropy weight TOPSIS is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiments of this application, or the part that contributes to the related technology, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause an electronic device to execute all or part of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, mobile hard drives, read-only memory (ROM), magnetic disks, or optical disks. Thus, the embodiments of this application are not limited to any specific hardware and software combination.

[0055] Correspondingly, embodiments of this application provide a computer-readable storage medium storing a computer program thereon. When executed by a processor, this computer program implements the steps in the dynamic evaluation method for seismic damage of double-row anti-slide piles based on LSTM-entropy weighted TOPSIS described in any of the above embodiments. Correspondingly, embodiments of this application also provide a computer program product. When executed by a processor of an electronic device, this computer program product is used to implement the steps in the dynamic evaluation method for seismic damage of double-row anti-slide piles based on LSTM-entropy weighted TOPSIS described in any of the above embodiments.

[0056] Based on the same technical concept, this application provides an electronic device for implementing the above-described method embodiment of a dynamic evaluation method for seismic damage of double-row anti-slide piles based on LSTM-entropy weighted TOPSIS. Figure 6 This is a schematic diagram of the hardware entity of an electronic device provided in an embodiment of this application, such as... Figure 6 As shown, the electronic device 600 includes a memory 610 and a processor 620. The memory 610 stores a computer program that can run on the processor 620. When the processor 620 executes the program, it implements the steps in the dynamic evaluation method for seismic damage of double-row anti-slide piles based on LSTM-entropy weight TOPSIS as described in any embodiment of this application.

[0057] The memory 610 is configured to store instructions and applications executable by the processor 620, and can also cache data to be processed or already processed by the processor 620 and various modules in the electronic device (e.g., image data, audio data, voice communication data and video communication data), which can be implemented by flash memory or random access memory (RAM).

[0058] When the processor 620 executes the program, it implements the steps of a dynamic evaluation method for seismic damage of double-row anti-slide piles based on LSTM-entropy weighted TOPSIS, as described above. The processor 620 typically controls the overall operation of the electronic equipment 600.

[0059] The aforementioned processor can be at least one of the following: Application Specific Integrated Circuit (ASIC), Digital Signal Processor (DSP), Digital Signal Processing Device (DSPD), Programmable Logic Device (PLD), Field Programmable Gate Array (FPGA), Central Processing Unit (CPU), Controller, Microcontroller, and Microprocessor. It is understood that other electronic devices can also implement the functions of the aforementioned processor, and this application does not specifically limit the specific implementation.

[0060] The aforementioned computer storage media / memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), magnetic random access memory (FRAM), flash memory, magnetic surface memory, optical disc, or compact disc read-only memory (CD-ROM), etc.; or it can be various electronic devices that include one or any combination of the above-mentioned memories, such as mobile phones, computers, tablet devices, personal digital assistants, etc.

[0061] It should be noted that the descriptions of the storage medium and device embodiments above are similar to the descriptions of the method embodiments above, and have similar beneficial effects. For technical details not disclosed in the storage medium and device embodiments of this application, please refer to the descriptions of the method embodiments of this application for understanding.

[0062] It should be understood that the phrase "one embodiment" or "an embodiment" throughout the specification means that a specific feature, structure, or characteristic related to the embodiment is included in at least one embodiment of this application. Therefore, "in one embodiment" or "in an embodiment" appearing throughout the specification does not necessarily refer to the same embodiment. Furthermore, these specific features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. It should be understood that in the various embodiments of this application, the sequence numbers of the above-described processes do not imply a sequential order of execution; the execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application. The sequence numbers of the above-described embodiments are merely descriptive and do not represent the superiority or inferiority of the embodiments.

[0063] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0064] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods, such as: multiple units or components can be combined, or integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the various components shown or discussed can be through some interfaces, and the indirect coupling or communication connection between devices or units can be electrical, mechanical, or other forms.

[0065] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units. They may be located in one place or distributed across multiple network units. Some or all of the units may be selected to achieve the purpose of the embodiments of this application, depending on actual needs.

[0066] In addition, each functional unit in the various embodiments of this application can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the integrated unit can be implemented in hardware or in the form of hardware plus software functional units.

[0067] Alternatively, if the integrated units described above are implemented as software functional modules and sold or used as independent products, they can also be stored in a computer-readable storage medium. Based on this understanding, the technical solutions of the embodiments of this application, or the parts that contribute to related technologies, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause the device automatic test line to execute all or part of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, ROMs, magnetic disks, or optical disks.

[0068] The methods disclosed in the several method embodiments provided in this application can be arbitrarily combined without conflict to obtain new method embodiments.

[0069] The features disclosed in the several method or device embodiments provided in this application can be arbitrarily combined without conflict to obtain new method or device embodiments.

[0070] The above description is merely an embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A dynamic evaluation method for seismic damage of double-row anti-slide piles based on LSTM-entropy weighted TOPSIS, characterized in that, The method includes: Acquire acceleration and dynamic earth pressure data of the anti-slide pile model under seismic wave action; A displacement prediction model is constructed based on a long short-term memory network, and the predicted displacement data is obtained by combining the acceleration data and the earth pressure data. The prediction effect of the predicted displacement data is evaluated according to the evaluation indicators, which include mean square error, mean absolute error, mean absolute error ratio, and coefficient of determination. A damage evaluation model is constructed based on the entropy weight method and the approximation ideal solution ranking method. The standardized index data and index weights are obtained by combining the dynamic earth pressure data and the predicted displacement data. Weighted index data is obtained based on the standardized index data and the index weights. The correlation between the damage state of the anti-slide pile and the ideal damage state is calculated using the Euclidean distance method. The damage evaluation result is determined based on the correlation.

2. The method according to claim 1, characterized in that, The displacement prediction model includes an input layer, a hidden layer, a long short-term memory network layer, and an output layer. The long short-term memory network layer includes a forget gate, an input gate, and an output gate.

3. The method according to claim 2, characterized in that, The displacement prediction model constructed based on the Long Short-Term Memory network, combined with the acceleration data and the earth pressure data, yields predicted displacement data, including: The acceleration data and the earth pressure data are preprocessed to obtain time series data; Determine the parameter configuration of the displacement prediction model; The time-series data is received through the input layer and then transmitted to the long short-term memory network layer. In the long short-term memory network layer, temporal features are extracted from the time-series data; The temporal feature extraction results are linearly transformed and mapped by a nonlinear activation function through a fully connected layer, and the predicted displacement data is generated by the output layer.

4. The method according to claim 1, characterized in that, The damage evaluation model is constructed based on the entropy weight method and the approximate ideal solution ranking method. Combining the dynamic earth pressure data and the predicted displacement data, standardized index data and index weights are obtained, including: The dynamic earth pressure data and the predicted displacement data are standardized using the range method to obtain the standardized index data. The standardization process includes positive index processing and negative index processing. Calculate the index information entropy based on the standardized index data; The weight of the indicator is calculated based on the information entropy of the indicator.

5. The method according to claim 1, characterized in that, The step of obtaining weighted index data based on the standardized index data and the index weights, and calculating the correlation between the damage state of the anti-slide pile and the ideal damage state using the Euclidean distance method, includes: The positive and negative ideal solutions are determined based on the weighted index data. Based on the positive ideal solution and the negative ideal solution, and in conjunction with the Euclidean distance method, calculate the positive Euclidean distance and the negative Euclidean distance; The correlation proximity is calculated based on the positive Euclidean distance and the negative Euclidean distance.

6. An electronic device comprising a memory and a processor, the memory storing a computer program executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the method according to any one of claims 1 to 5.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the steps of the method according to any one of claims 1 to 5.

Images