Earth and rockfill dam rolling quality prediction method based on improved firefly optimization random forest algorithm

Abstract

The invention relates to an earth and rockfill dam rolling quality prediction method based on an improved firefly optimization random forest algorithm. The method comprises the following steps: S1, constructing an original data set of earth and rockfill dam compaction quality evaluation indexes; s2, analyzing and quantifying the uncertainty of the rolling quality influence parameters of the earth and rockfill dam; s3, normalization and feature extraction of the input data are carried out; s4, Ntree and Mtry parameters of a random forest algorithm are optimized based on improved fireflies; and S5, training a random forest model based on the optimized Ntree and Mtry parameters, and predicting the rolling quality of the whole bin surface of the earth and rockfill dam. According to the method, the accuracy, robustness and efficiency of earth and rockfill dam rolling quality prediction can be improved.

Description

Technical Field

[0001] The invention belongs to the technical field of water conservancy and hydropower engineering, and particularly relates to a method for predicting the rolling quality of an earth-rock dam based on an improved firefly optimization random forest algorithm. Background Art

[0002] As a water retaining structure with simple structure, low cost and strong adaptability, earth-rock dams have been widely used in water conservancy and hydropower projects. In recent years, with the increasing scale of earth-rock dam projects in China, the society has paid particular attention to the safety and construction quality control of dams. Among them, the rolling quality control of earth-rock dams is a key link to ensure the safety of dam projects. However, traditional quality control mainly relies on trial pit tests to obtain evaluation data through dry density. This method has many deficiencies: Firstly, it is difficult to comprehensively reflect the overall compaction quality of the dam body through limited sampling points, which affects the accuracy of evaluation; Secondly, the trial pit test takes a long time and delays the on-site construction progress; Thirdly, this method is for post-evaluation and cannot identify and feedback the control of on-site rolling quality in time.

[0003] To solve the above problems, many scholars have developed a real-time monitoring system for the rolling quality of dams. For example, a real-time monitoring system based on technologies such as GPS and GPRS can automatically collect and monitor construction parameters during the rolling process, including the number of rolling passes, rolling thickness, and vibration force state, etc., which provides an effective means for the real-time control of construction quality. In addition, the Compaction Value (CV) has been proposed as a real-time characterization index for compaction quality and has been verified by engineering. However, although the real-time monitoring system has remarkable effects in improving construction quality control, the above parameters are difficult to directly reflect the compaction quality, and the uncertainty of parameters caused by random sampling is still a challenge affecting the reliability of dam body quality assessment.

[0004] Currently, related research uses artificial intelligence algorithms to construct the mapping relationship between the rolling quality influence parameters and the rolling quality characterization index, and obtains the full-face prediction values of the rolling quality characterization index (such as dry density and compaction degree), so as to realize the full-face prediction and evaluation of the rolling quality of the dam. Existing traditional compaction quality evaluation models mostly adopt linear or nonlinear regression and neural network methods. The above traditional methods can better fit the relationship between each parameter and dry density, but usually ignore the influence of parameter uncertainty factors, and the fitting accuracy and generalization ability still need to be improved, resulting in limited evaluation accuracy. Therefore, there is an urgent need for an algorithm that can comprehensively consider parameter uncertainty to more comprehensively and accurately predict compaction quality. Summary of the Invention

[0005] The purpose of the invention is to provide a method for predicting the rolling quality of an earth-rock dam based on an improved firefly optimization random forest algorithm, which can improve the accuracy, robustness and efficiency of the prediction of the rolling quality of the earth-rock dam.

[0006] To achieve the above object, the technical solution adopted by the present invention is: a method for predicting the rolling quality of an earth-rock dam based on an improved firefly optimization random forest algorithm, comprising the following steps:

[0007] Step S1: Construct an original data set of evaluation indexes for the compaction quality of the earth-rock dam;

[0008] Step S2: Analyze and quantify the uncertainty existing in the parameters affecting the rolling quality of the earth-rock dam;

[0009] Step S3: Perform normalization and feature extraction of the input data;

[0010] Step S4: Based on the Ntree and Mtry parameters of the improved firefly optimization random forest algorithm;

[0011] Step S5: Based on the optimized Ntree and Mtry parameters, train a random forest model and predict the rolling quality of the entire warehouse surface of the earth-rock dam.

[0012] Further, in the step S1, rolling parameters, material source parameters and meteorological elements are obtained respectively through the real-time monitoring system for earth-rock dam rolling, pit test and meteorological on-site supervision, and an original data set of evaluation indexes for the compaction quality of the earth-rock dam is constructed; the evaluation index system for the compaction quality of the earth-rock dam includes three layers. The first layer is the target layer of the evaluation, that is, the dry density of the evaluation index for the compaction quality of the earth-rock dam; the second layer is the parameter categories affecting the dry density of the compaction quality evaluation index, including rolling parameters, material source parameters and meteorological elements; the third layer is the specific parameters affecting the dry density of the compaction quality evaluation index. The rolling parameters include rolling speed, number of rolling passes and rolling thickness. The material source parameters include P5 content, total material moisture content and dam material gradation. The meteorological elements include temperature and humidity.

[0013] Further, the step S1 specifically includes the following steps:

[0014] Step S11: Use the real-time monitoring system for earth-rock dam rolling to obtain the rolling parameters at any position; adopt GPS positioning technology and line segment generation technology to obtain the number of rolling passes, rolling speed and rolling thickness;

[0015] Step S12: Use a nuclear densitometer or the core cutter method to detect the dry density at the construction site and obtain the dry density data of the detection points;

[0016] Step S13: Detect the gradation of the soil and stone materials by using particle analysis tests; obtain d10, d30, and d60 through the particle size distribution curve; d10 is the effective particle size; d30 is the median particle size; d60 is the control particle size; calculate Cu and Cc through d10, d30, and d60; the smaller the coefficient of uniformity Cu, the more uniform the composition of the soil particles, and vice versa, the more non-uniform the composition of the soil particles; the coefficient of curvature Cc reflects the distribution range of the soil particles.

[0017] Step S14: Obtain the moisture content of the dam materials of the earth-rock dam by using moisture content tests; calculate the soil property indexes including dry density, saturation degree, and void ratio based on the moisture content of the soil as the basic data.

[0018] Step S15: Obtain the temperature and humidity of the earth-rock dam surface by using meteorological on-site monitoring.

[0019] Step S16: Conduct a correlation analysis between the construction parameters and the dry density, select the index with a larger significance as the influencing factor, so as to obtain the evaluation index system for the compaction quality of the earth-rock dam.

[0020] Furthermore, the specific method for conducting the correlation analysis between the construction parameters and the dry density is as follows: Use the Pearson correlation analysis method of SPSS software to analyze the correlation between each index parameter and the dry density. On this basis, conduct a relevant significance T-test, so as to establish a correlation coefficient matrix diagram between each index parameter, and select the index with a larger significance as the main influencing factor through the correlation analysis index; the calculation formula of the correlation coefficient is as follows:

[0021]

[0022] where r is the correlation coefficient, n is the number of samples, x i is the i-th value of the dry density influencing factor, is the average value of the dry density influencing factor, value is the i-th value of the measured dry density, is the average value of the measured dry density.

[0023] Furthermore, the specific steps of the said Step S2 include the following steps:

[0024] Step S21: Sort each source parameter to construct a pseudo time series S j :

[0025] S j =sort(x j ), j = 1, 2, …, n (2)

[0026] where sort() represents sorting the elements, and n is the number of samples;

[0027] Step S22: Set the sample entropy parameters: embedding dimension m = 2; similarity tolerance r = 0.2×SD, where SD is the standard deviation of the sequence, and time delay τ = 1;

[0028] Step S23: Perform phase space reconstruction on the pseudo time series S of length n j to construct vectors of dimension m, obtaining the following sequence vectors:

[0029] X(i) = [x(i), x(i + 1), …, x(i + m - 1)] (i = 1, 2, …, N - m + 1) (3)

[0030] Step S24: Calculate the sample entropy value as follows:

[0031]

[0032] where is the number with Euclidean distance less than r;

[0033] Step S25: Repeat the above steps to obtain the sample entropy values of all material source parameters, and judge the uncertainty degree of the parameters according to the magnitude of the entropy values.

[0034] Further, the specific steps of the said Step S3 include the following steps:

[0035] Step S31: The parameters affecting the rolling quality of the rockfill dam include rolling parameters, material source parameters, and meteorological parameters, and different data have different units and scales; before training the rolling quality prediction model, normalize the original input data and then import it into the rolling quality prediction model for training and verification; adopt the minimum-maximum scaling to normalize the time series data:

[0036]

[0037] where is the original input value; is the normalized original input value; is the minimum value of the original input value; is the maximum value of the original input value;

[0038] Step S32: Set the first moment and the second central moment of the normalized original input value and the original input value as the input to improve the characteristics of the input data fluctuation information; for each x i The first moment and the second central moment of the input number x (x1, x2,..., x n ) are:

[0039]

[0040] where n is the length of the data for extracting data features, that is, calculating the first moment and the second central moment; and are the first moment and the second central moment.

[0041] Furthermore, in step S4, the position of the firefly represents the number of features Mtry and the number of trees Ntree of the random forest; the position of the firefly is randomly initialized, with the mean squared error MSE as the fitness. The smaller the fitness, the closer it is to the target; the position of the firefly is updated, and the random forest is trained. When the set fitness value is reached or the maximum number of iterations is reached, the training ends, and the optimal position is output, that is, the optimal number of features Mtry and the number of trees Ntree of the random forest.

[0042] Furthermore, step S4 specifically includes the following steps:

[0043] Step S41: Initialize the firefly algorithm; set parameters: Initialize the parameters of the firefly algorithm, including population size, maximum number of iterations, initial brightness, attraction coefficient, and step size factor; generate the initial population: randomly generate a group of firefly individuals, and each individual represents a solution, that is, the parameter combination of Ntree and Mtry of the RF algorithm;

[0044] Step S42: Introduce dynamic inertia weight and adaptive factor; during the movement of the firefly, introduce dynamic inertia weight and adaptive factor to enable the algorithm to have stronger local search ability during later convergence and avoid falling into local optimal solutions;

[0045] Step S43: Introduce differential evolution strategy; for each firefly individual, execute the differential evolution strategy; by comparing the fitness of the candidate solution and the current solution, select the solution with better fitness to retain and enter the next generation;

[0046] Step S44: Firefly movement and update; based on the objective function: the prediction error of the dam rolling quality, calculate the brightness of each firefly; according to the brightness difference and the introduced dynamic inertia weight and adaptive factor, calculate the movement vector of the firefly individual and update the individual position;

[0047] Step S45: Update the new position and brightness value of each firefly individual;

[0048] Step S46: Use the improved firefly algorithm to adaptively optimize the parameters of the RF algorithm;

[0049] Step S47: Use the optimal parameter combination to train the random forest model and apply it to the prediction of dam rolling quality.

[0050] Further, in step S5, based on the optimal Ntree and Mtry parameters obtained by the improved firefly algorithm, the random forest model is trained; subsequently, using the rolling parameters, material source parameters, and meteorological parameters as inputs and the rolling quality dry density as the output, the data of the validation set is imported into the trained model to predict the rolling quality of the entire dam surface of the earth-rock dam and perform inverse normalization to obtain the calculated output value and compare it with the original output value; four indicators are selected to evaluate the performance of the model, including the standard deviation error E SD and the absolute mean error E abs extreme value error E max and goodness of fit R new :

[0051]

[0052] where and are the original output value and the calculated output value of the prediction model respectively; and are the absolute average values of the original output value and the calculated output value of the prediction model respectively; y actu_max and y pred_max are the maximum values of the original output value and the calculated output value of the prediction model respectively.

[0053] The present invention also provides an earth-rock dam rolling quality prediction system based on an improved firefly optimization random forest algorithm, including a memory, a processor, and computer program instructions stored on the memory and executable by the processor. When the processor runs the computer program instructions, the above method can be implemented.

[0054] Compared with the prior art, the present invention has the following beneficial effects:

[0055] 1. Prediction is carried out through the black-box model - random forest algorithm, considering the influence of parameter uncertainty, and sample entropy analysis is used to quantify the parameter uncertainty, making the prediction analysis more efficient and reliable.

[0056] 2. The firefly algorithm is improved by combining dynamic inertia weight, adaptive factor, and differential evolution algorithm to solve the problems that the traditional firefly algorithm may fall into local optimum when facing complex and high-dimensional problems, affecting the search efficiency and solution quality.

[0057] 3. The improved firefly algorithm is used to optimize the hyperparameters of the random forest, effectively improving the prediction efficiency of the model and ensuring the generalization ability and accuracy level of the model. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 is a diagram of the evaluation index system for the compaction quality of the earth-rock dam in the embodiment of the present invention; ​​

[0059] Figure 2 It is the flowchart of optimizing the random forest algorithm by the improved firefly algorithm in the embodiment of the present invention;

[0060] Figure 3 It is the implementation flowchart of the roller compaction quality prediction method for earth-rock dams based on the improved firefly optimized random forest algorithm in the embodiment of the present invention; Specific implementation manners

[0061] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0062] It should be noted that the following detailed description is exemplary and is intended to provide further illustration of the present application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present application belongs.

[0063] It should be noted that the terms used herein are only for describing specific implementation manners and are not intended to limit the exemplary implementation manners according to the present application. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0064] As Figure 3 shown, this embodiment provides a roller compaction quality prediction method for earth-rock dams based on the improved firefly optimized random forest algorithm, including the following steps:

[0065] Step S1: Construct the original data set of the evaluation index for the compaction quality of the earth-rock dam.

[0066] Specifically, the rolling parameters, material source parameters, and meteorological elements are obtained through the real-time monitoring system for earth-rock dam rolling, pit tests, and meteorological on-site supervision respectively, and the original data set of the evaluation index for the compaction quality of the earth-rock dam is constructed. As Figure 1 shown, the evaluation index system for the compaction quality of the earth-rock dam includes three layers. The first layer is the target layer of the evaluation, that is, the evaluation index for the compaction quality of the earth-rock dam (dry density); the second layer is the parameter category that affects the dry density of the compaction quality evaluation index, including rolling parameters, material source parameters, and meteorological elements; the third layer is the specific parameters that affect the dry density of the compaction quality evaluation index. The rolling parameters include rolling speed, number of rolling passes, and rolling thickness. The material source parameters include P5 content, total material moisture content, and dam material gradation. The meteorological elements include temperature and humidity.

[0067] The specific steps of step S1 include the following steps:

[0068] Step S11: Obtain the compaction parameters at any position by using the real-time monitoring system for earth-rock dam compaction. Compaction parameters are important indicators for controlling compaction quality. The compaction passes, compaction speed, and compaction thickness are obtained by using GPS positioning technology and line generation technology. Generally, it is required that the compaction speed is between 1 km / h and 3 km / h, the compaction passes are at least 8 times, and the compaction thickness is about 30 cm.

[0069] Step S12: Detect the dry density at the construction site by using a nuclear density gauge or the sand replacement method to obtain the dry density data of the detection points. Dry density is an indicator characterizing compaction quality. At the same time, the magnitude of dry density will affect the quality performance such as the anti-seepage and shear strength of the earth-rock dam.

[0070] Step S13: Detect the gradation of the soil and rock materials by using particle analysis tests such as the sieve analysis method. Soil is an aggregate composed of particles with different shapes and sizes, and a grain group is a collection of particles with similar engineering properties; through the particle size distribution curve, d10, d30, and d60 can be obtained; d10 is called the effective grain size. Generally speaking, the d10 of sandy soil is positively correlated with its permeability; while the d10 of cohesive soil is negatively correlated with its plasticity; d30 is the median grain size; d60 is the controlling grain size; based on d10, d30, and d60, Cu and Cc can be calculated through calculation; the smaller the coefficient of uniformity Cu, the more uniform the soil particle composition, and vice versa, the more uneven the soil particle composition; the coefficient of curvature Cc reflects the distribution range of soil particles.

[0071] Step S14: Obtain the moisture content of the earth-rock dam materials by using moisture content tests such as the drying method. Moisture content is a basic physical index reflecting the dry and wet state degree of soil. Based on the moisture content of the soil as the basic data, soil property indexes such as dry density, saturation, and void ratio can be calculated. At the same time, moisture content is an important basis for evaluating the engineering properties of soil and rock materials and an important index for studying their physical and mechanical properties.

[0072] Step S15: Obtain the temperature and humidity of the earth-rock dam surface by using meteorological on-site monitoring. Climate changes such as external temperature and humidity have a certain impact on the compaction quality of the earth-rock dam. For example, humidity can affect the change of moisture content, thereby indirectly affecting the compaction quality.

[0073] Step S16: Conduct a correlation analysis between construction parameters and dry density, and select the index with a larger significance as the influencing factor, so as to obtain the index system of the earth-rock dam compaction quality evaluation model. Specifically: use the Pearson correlation analysis method of SPSS software to analyze the correlation between each index parameter and dry density. On this basis, conduct a relevant significance T test, so as to establish a correlation coefficient matrix diagram between each index parameter, and select the index with a larger significance as the main influencing factor through the correlation analysis index. The calculation formula of the correlation coefficient is as follows:

[0074]

[0075] Among them, r is the correlation coefficient, n is the number of samples, and x i is the i-th value of the dry density influence factor, is the average value of the dry density influence factor, the value is the i-th value of the measured dry density, is the average value of the measured dry density.

[0076] Step S2: Analyze and quantify the uncertainty existing in the parameters affecting the rolling quality of the earth-rock dam.

[0077] The uncertainty of the dry density influencing factors mainly comes from two aspects. On the one hand, it is the variability of uncontrollable factors. Taking the moisture content as an example, although the moisture content has reached the control standard through tests, during the actual construction process, the construction measures to ensure the moisture content cannot accurately control the moisture content, resulting in a large variability of the moisture content of the dam materials on the entire bin surface. Therefore, the moisture content of the dam materials has uncertainty. On the other hand, it is the randomness in the process of evaluating the compaction quality. The on-site material source parameters can only obtain determined values at limited test pits. For the untested points on the bin surface, normal random numbers are generated by MATLAB for simulation. Therefore, the finally obtained dry density has a certain degree of random uncertainty. Affected by the unevenness of the dam material properties, even with the same rolling parameters, different compaction qualities may be caused. Therefore, the sample entropy theory is applied to analyze it, and the uncertainty of the material source parameters is expressed as "sample entropy" to achieve the purpose of quantifying the parameter uncertainty.

[0078] Applying "sample entropy" to process the material source parameters with uncertainty specifically includes the following steps:

[0079] Step S21: Sort each material source parameter to construct a pseudo-time series S j :

[0080] S j = sort(x j ), j = 1, 2, …, n (2)

[0081] Among them, sort() represents sorting the elements, and n is the number of samples.

[0082] Step S22: Set the sample entropy parameters: embedding dimension m = 2; similarity tolerance r = 0.2×SD (SD is the standard deviation of the sequence), delay time τ = 1 (static data).

[0083] Step S23: Perform phase space reconstruction on the pseudo-time series S j with length n to construct a vector with dimension m, and obtain the following sequence vectors:

[0084] X(i) = [x(i), x(i + 1), …, x(i + m - 1)] (i = 1, 2, …, N - m + 1) (3)

[0085] Step S24: Calculate the sample entropy value as follows:

[0086]

[0087] where, is the number of Euclidean distances less than r.

[0088] Step S25: Repeat the above steps to obtain the sample entropy values of all material source parameters, and judge the degree of uncertainty of the parameters according to the magnitudes of the entropy values.

[0089] Step S3: Perform normalization and feature extraction of the input data.

[0090] The said Step S3 specifically includes the following steps:

[0091] Step S31: The parameters affecting the rolling quality of the rockfill dam include rolling parameters, material source parameters, and meteorological parameters, and different data have different units and scales; before training the rolling quality prediction model, normalize the original input data and then import it into the rolling quality prediction model for training and verification; adopt the minimum - maximum scaling to normalize the time - series data:

[0092]

[0093] where, is the original input value; is the normalized original input value; is the minimum value of the original input value; is the maximum value of the original input value.

[0094] Step S32: Set the first - order moment and the second - order central moment of the normalized original input value and the original input value as the input to improve the characteristics of the input data fluctuation information; for each x i The number of inputs x(x1, x2,..., x n )'s first - order moment and second - order central moment are:

[0095]

[0096] where, n is the length of the data for extracting features, that is, calculating the first - order moment and the second - order central moment; and are the first - order moment and the second - order central moment.

[0097] Step S4: Based on the Ntree and Mtry parameters of the improved firefly - optimized random forest algorithm.

[0098] The basic idea of optimizing the random forest algorithm with the improved firefly algorithm is that the position of the firefly represents the number of features Mtry and the number of trees Ntree of the random forest; randomly initialize the position of the firefly, use the mean squared error (MSE) as the fitness, and the smaller the fitness, the closer to the target; update the position of the firefly, train the random forest, and when the set fitness value is reached or the maximum number of iterations is reached, end the training and output the optimal position (i.e., the optimal number of features Mtry and the number of trees Ntree of the random forest).

[0099] As Figure 2 shown, step S4 specifically includes the following steps:

[0100] Step S41: Initialize the firefly algorithm; set parameters: initialize the parameters of the firefly algorithm, including population size, maximum number of iterations, initial brightness, attraction coefficient, step size factor, etc.; generate the initial population: randomly generate a group of firefly individuals, and each individual represents a solution (i.e., the parameter combination of Ntree and Mtry of the RF algorithm).

[0101] Step S42: Introduce dynamic inertia weight and adaptive factor; during the movement of the firefly, introduce dynamic inertia weight and adaptive factor to enable the algorithm to have stronger local search ability during later convergence and avoid falling into local optimal solutions.

[0102] Step S43: Introduce differential evolution strategy; for each firefly individual, execute the differential evolution strategy; by comparing the fitness of the candidate solution and the current solution, select the solution with better fitness to retain and enter the next generation.

[0103] Step S44: Firefly movement and update; based on the objective function (here it is the prediction error of the dam rolling quality), calculate the brightness of each firefly; according to the brightness difference and the introduced dynamic inertia weight and adaptive factor, calculate the movement vector of the firefly individual and update the individual position.

[0104] Step S45: Update the new position and brightness value of each firefly individual.

[0105] Step S46: Use the improved firefly algorithm to adaptively optimize the parameters of the RF algorithm.

[0106] Step S47: Use the optimal parameter combination to train the random forest model and apply it to the prediction of the dam rolling quality.

[0107] Step S5: Based on the optimized Ntree and Mtry parameters, train the random forest model and predict the rolling quality of the entire full-face of the earth-rock dam.

[0108] Based on the optimal Ntree and Mtry parameters obtained by the improved firefly algorithm, the random forest model is trained. Subsequently, with the rolling parameters, material source parameters, and meteorological parameters as inputs and the rolling quality dry density as the output, the data of the validation set is imported into the trained model to predict the rolling quality of the entire dam surface of the earth-rock dam, and inverse normalization is performed to obtain the calculated output value and compare it with the original output value. Four indicators are selected to evaluate the performance of the model, including the standard deviation error E SD , the absolute mean error E abs , the extreme value error E max , and the goodness of fit R new :

[0109]

[0110] Among them, and are the original output value and the calculated output value of the prediction model respectively; and are the absolute average values of the original output value and the calculated output value of the prediction model respectively; y actu_max and y pred_max are the maximum values of the original output value and the calculated output value of the prediction model respectively.

[0111] According to the "Design Requirements and Construction Parameters of Dam Materials" in the dam project, the compliance rate of the dry density of the gravelly soil material in the core wall area (i.e., ρ d ≥2.18 g / cm 3 ) shall not be less than 97%. Therefore, the compliance rate R of the dry density of the gravelly soil material in the core wall area is obtained by analyzing the generated cloud map of the dry density distribution of the compaction quality. (The solution equation is shown in Formula 5).

[0112] R = A' / A × 100% (12)

[0113] Among them, R is the compliance rate of the compaction degree of the entire dam surface in the core wall area of the earth-rock dam; A' is the area of the entire dam surface; A is the area where the compaction degree of the entire dam surface reaches the index.

[0114] If R < 97%, the compaction quality of this dam surface is unqualified;

[0115] If R ≥ 97%, the compaction quality of this dam surface is qualified.

[0116] In this embodiment, the MATLAB software is used to implement the construction of a rolling quality prediction model based on the improved firefly optimization random forest algorithm, and the process is as Figure 3 shown.

[0117] First, set the basic parameters of the model, set the calculation time step and calculation time; according to step S1, construct the original data set of the evaluation index for the compaction quality of the earth-rock dam; according to step S2, analyze and quantify the uncertainty existing in the influencing parameters of the rolling quality of the earth-rock dam; according to step S3, normalize the input data and extract features; according to step S4, determine the Ntree and Mtry parameters of the random forest algorithm based on the improved firefly optimization algorithm; finally, based on the optimized Ntree and Mtry parameters, train the random forest model and predict the rolling quality of the entire filling surface of the earth-rock dam.

[0118] The method for predicting the rolling quality of an earth-rock dam based on the improved firefly optimization random forest algorithm provided by the present invention integrates multi-source monitoring data and uncertainty factors, and realizes effective and accurate prediction of compaction quality through the combination of intelligent optimization and machine learning algorithms. This method uses sample entropy to more accurately analyze the existing uncertainty of parameters, improves the firefly algorithm through dynamic weight adjustment, adaptive factor and differential evolution algorithm, couples the improved firefly algorithm and the random forest algorithm, enhances the search ability of the random forest algorithm and its adaptability to parameter uncertainty, and can effectively improve the accuracy and applicability of the construction quality control of earth-rock dams, providing a new technical means for the safety and stability of the project.

[0119] This embodiment also provides a system for predicting the rolling quality of an earth-rock dam based on the improved firefly optimization random forest algorithm, including a memory, a processor, and computer program instructions stored on the memory and capable of being run by the processor. When the processor runs the computer program instructions, the above method can be implemented.

[0120] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0121] The present application is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram, and the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate for implementation in the process Figure 1one or more processes and / or blocks Figure 1 means for the functions specified in one or more blocks

[0122] These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to work in a particular manner, such that the instructions stored in the computer-readable memory produce a manufacture including an instruction means that implements the functions in the process Figure 1 one or more processes and / or blocks Figure 1 the functions specified in one or more blocks

[0123] These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus, such that a series of operational steps are performed on the computer or other programmable apparatus to produce a computer-implemented process, whereby the instructions executed on the computer or other programmable apparatus provide steps for implementing the functions in the process Figure 1 one or more processes and / or blocks Figure 1 the functions specified in one or more blocks

[0124] As described above, it is only the preferred embodiments of the present invention, and is not a limitation of the present invention in other forms. Any person skilled in the art may use the technical content disclosed above to make changes or modifications into equivalent embodiments with equivalent changes. However, any simple modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the technical solution content of the present invention still fall within the protection scope of the technical solution of the present invention.

Claims

1. A method for predicting the rolling quality of an earth-rock dam based on an improved firefly optimization random forest algorithm, characterized in that, It includes the following steps: Step S1: Construct the original data set of the compacting quality evaluation index of the earth-rock dam; Step S2: Analyze and quantify the uncertainty existing in the influencing parameters of the rolling quality of the earth-rock dam; Step S3: Conduct normalization and feature extraction of the input data; Step S4: Based on the Ntree and Mtry parameters of the improved firefly optimization random forest algorithm; Step S5: Based on the optimized Ntree and Mtry parameters, train the random forest model and predict the rolling quality of the entire filling surface of the earth-rock dam.

2. A method for predicting the rolling quality of an earth-rock dam based on an improved firefly optimization random forest algorithm according to claim 1, characterized in that, In the said Step S1, the rolling parameters, material source parameters and meteorological elements are obtained through the real-time monitoring system for the rolling of the earth-rock dam, the trial pit test and the meteorological on-site supervision respectively, and the original data set of the compacting quality evaluation index of the earth-rock dam is constructed; the compacting quality evaluation index system of the earth-rock dam includes three layers. The first layer is the target layer of the evaluation, that is, the dry density of the compacting quality evaluation index of the earth-rock dam; The second layer is the parameter category affecting the dry density of the compacting quality evaluation index, including rolling parameters, material source parameters and meteorological elements; the third layer is the specific parameters affecting the dry density of the compacting quality evaluation index. The rolling parameters include rolling speed, number of rolling passes and rolling thickness. The material source parameters include the content of P5, the moisture content of the whole material and the dam material gradation. The meteorological elements include temperature and humidity.

3. A method for predicting the compaction quality of an earth-rock dam based on an improved firefly optimization random forest algorithm according to claim 2, characterized in that The said Step S1 specifically includes the following steps: Step S11: Use the real-time monitoring system for the rolling of the earth-rock dam to obtain the rolling parameters at any position; use the GPS positioning technology and the line segment generation technology to obtain the number of rolling passes, rolling speed and rolling thickness; Step S12: Use the nuclear densitometer or the core cutter method to detect the dry density at the construction site and obtain the dry density data of the detection points; Step S13: Use the particle analysis test to detect the gradation of the soil and rock materials; through the particle size distribution curve, obtain d10, d30, d60; d10 is the effective particle size; d30 is the median particle size; d60 is the control particle size; through d10, d30, d60, calculate Cu and Cc; the smaller the coefficient of uniformity Cu, the more uniform the composition of the soil particles, otherwise, the more uneven the composition of the soil particles; the curvature coefficient Cc reflects the distribution range of the soil particles; Step S14: Use the moisture content test to obtain the moisture content of the dam materials of the earth-rock dam; based on the moisture content of the soil as the basic data, calculate the soil property indexes including dry density, degree of saturation and void ratio; Step S15: Use the meteorological on-site supervision to obtain the temperature and humidity of the filling surface of the earth-rock dam; Step S16: Conduct a correlation analysis between the construction parameters and the dry density, and select the indexes with greater significance as the influencing factors, so as to obtain the compacting quality evaluation index system of the earth-rock dam.

4. A method for predicting the compaction quality of an earth-rock dam based on an improved firefly optimization random forest algorithm according to claim 3, characterized in that The specific method for conducting the correlation analysis between the construction parameters and the dry density is: use the Pearson correlation analysis method of the SPSS software to analyze the correlation between each index parameter and the dry density. On this basis, conduct a relevant significance T test, so as to establish a correlation coefficient matrix diagram between each index parameter, and select the indexes with greater significance as the main influencing factors through the correlation analysis index; the calculation formula of the correlation coefficient is as follows: Among them, r is the correlation coefficient, n is the number of samples, and x i is the i-th value of the dry density influence factor, is the average value of the dry density influence factor, value is the i-th value of the measured dry density, is the average value of the measured dry density.

5. A method for predicting the compaction quality of an earth-rock dam based on an improved firefly optimization random forest algorithm according to claim 1, characterized in that, The said Step S2 specifically includes the following steps: Step S21: Sort each source parameter to construct a pseudo time series S j : S j = sort(x j ), j = 1, 2, …, n (2) Among them, sort() represents sorting the elements, and n is the number of samples; Step S22: Set the sample entropy parameters: embedding dimension m = 2; similarity tolerance r = 0.2×SD, where SD is the standard deviation of the sequence, and delay time τ = 1; Step S23: Perform phase space reconstruction on the pseudo time series S of length n j to construct a vector of dimension m, obtaining the following sequence vector: X(i) = [x(i), x(i + 1), …, x(i + m - 1)] (i = 1, 2, …, N - m + 1) (3) Step S24: Calculate the sample entropy value as follows: Among them, is the number of Euclidean distances less than r; Step S25: Repeat the above steps to obtain the sample entropy values of all material source parameters, and judge the uncertainty degree of the parameters according to the magnitude of the entropy values.

6. A method for predicting the rolling quality of an earth-rock dam based on an improved firefly optimization random forest algorithm according to claim 1, characterized in that The specific steps of step S3 are as follows: Step S31: The parameters affecting the rolling quality of the rockfill dam include rolling parameters, material source parameters, and meteorological parameters, and different data have different units and scales; before training the rolling quality prediction model, normalize the original input data and then import it into the rolling quality prediction model for training and verification; use the minimum-maximum scaling to normalize the time series data: Among them, is the original input value; is the normalized original input value; is the minimum value of the original input value; is the maximum value of the original input value; Step S32: Set the first moment and the second central moment of the normalized original input values and the original input values as the input to improve the characteristics of the input data fluctuation information; for each x i The number of inputs x (x1, x2,..., x n ) of the first moment and the second central moment are: Among them, n is the data feature for extraction, that is, the data length for calculating the first moment and the second central moment; and are the first moment and the second central moment.

7. A method for predicting the rolling quality of an earth-rock dam based on an improved firefly optimization random forest algorithm according to claim 1, characterized in that, In step S4, the position of the firefly represents the number of features Mtry and the number of trees Ntree of the random forest; randomly initialize the firefly position, with the mean squared error MSE as the fitness, and the smaller the fitness, the closer it is to the target; update the firefly position, train the random forest, and end the training when the set fitness value is reached or the maximum number of iterations is reached, and output the optimal position, that is, the optimal number of features Mtry and the number of trees Ntree of the random forest.

8. A method for predicting the rolling quality of an earth-rock dam based on an improved firefly optimization random forest algorithm according to claim 7, characterized in that, The specific steps of step S4 are as follows: Step S41: Initialize the firefly algorithm; set parameters: initialize the parameters of the firefly algorithm, including population size, maximum number of iterations, initial brightness, attraction coefficient, and step size factor; generate the initial population: randomly generate a group of firefly individuals, and each individual represents a solution, that is, the parameter combination of Ntree and Mtry of the RF algorithm; Step S42: Introduce dynamic inertia weight and adaptive factor; during the movement of the firefly, introduce dynamic inertia weight and adaptive factor to make the algorithm have stronger local search ability during later convergence and avoid falling into local optimal solutions; Step S43: Introduce differential evolution strategy; for each firefly individual, execute the differential evolution strategy; by comparing the fitness of the candidate solution and the current solution, select the solution with better fitness to be retained in the next generation; Step S44: Firefly movement and update; based on the objective function: the prediction error of the dam rolling quality, calculate the brightness of each firefly; according to the brightness difference and the introduced dynamic inertia weight and adaptive factor, calculate the movement vector of the firefly individual and update the individual position; Step S45: Update the new position and brightness value of each firefly individual; Step S46: Use the improved firefly algorithm to adaptively optimize the parameters of the RF algorithm; Step S47: Use the optimal parameter combination to train the random forest model and apply it to the prediction of the dam rolling quality.

9. A method for predicting the rolling quality of an earth-rock dam based on an improved firefly optimization random forest algorithm according to claim 1, characterized in that, In the step S5, the random forest model is trained based on the optimal Ntree and Mtry parameters obtained by the improved firefly algorithm. Subsequently, with the rolling parameters, material source parameters, and meteorological parameters as inputs and the rolling quality dry density as the output, the data of the validation set is imported into the trained model to predict the rolling quality of the entire full face of the earth-rock dam, and inverse normalization is performed to obtain the calculated output value and compare it with the original output value. Four indicators are selected to evaluate the performance of the model, including the standard deviation error E SD , the absolute mean error E abs , the extreme value error E max and the goodness of fit R new : Wherein, and are the original output value and the calculated output value of the prediction model respectively; and are the absolute average values of the original output value and the calculated output value of the prediction model respectively; y actu_max and y pred_max are the maximum values of the original output value and the calculated output value of the prediction model respectively.

10. A roller compaction quality prediction system for earth-rock dams based on an improved firefly optimization random forest algorithm, characterized in that, It includes a memory, a processor, and computer program instructions stored on the memory and capable of being run by the processor. When the processor runs the computer program instructions, the method described in any one of claims 1-9 can be implemented.

Images