Optimization decision-making method for slope stability of open pit coal mine waste dump

Abstract

The invention discloses an opencast coal mine waste dump slope stability optimization decision-making method, which comprises the steps of obtaining a test data set based on factor information influencing the slope stability of an opencast coal mine waste dump; constructing a feedforward neural network model, and performing data feature extraction and nonlinear processing on the test data set; optimizing the weight and the threshold value of the feedforward neural network model by using a genetic algorithm; using a mean square error as a loss function to train and verify the feedforward neural network model to obtain a trained feedforward neural network model; and utilizing the trained feedforward neural network model to predict waste dump data to obtain a slope stability decision scheme. According to the method, the efficiency and accuracy of predicting the slope stability of the open pit coal mine waste dump are improved, the slope stability of the open pit coal mine waste dump can be analyzed more reliably, and powerful support is provided for disaster prevention and stability prediction of the open pit coal mine waste dump.

Description

Technical Field

[0001] The invention relates to the technical field of slope stability analysis, and in particular to an optimization decision method for slope stability of an open-pit coal mine dump. Background Art

[0002] In open-pit coal mines, the spoil dump is an area used to store and process excavated soil and rock. The slope stability of the spoil dump is an important issue because the instability of the slope may lead to serious consequences such as debris flow, landslide and accidents. The slope stability of the spoil dump of an open-pit coal mine refers to whether the earth slope can maintain sufficient stability during long-term operation to prevent landslides, collapses and other disasters. The slope stability of the spoil dump is a very important safety issue in coal mining engineering, involving personnel and equipment safety issues as well as environmental protection issues. The evaluation of the slope stability of the spoil dump needs to consider multiple factors, including the mechanical properties of the soil, slope, soil dumping method, water level, geological and topographic conditions, etc. The factors that affect the slope stability of the spoil dump are mainly: groundwater and atmospheric precipitation lead to an increase in the amount of water in the spoil, which reduces the strength of the spoil dump; the height of the spoil pile creates a huge load on the base; the internal composition of the spoil affects its shear resistance.

[0003] There are many problems in the slope of open-pit coal mine dumps. The geological environment in which coal mines are located may be very complex, with various rock layers, faults, fragile geological structures, etc. These geological conditions have a significant impact on the stability of the slope, which may lead to slope deformation, collapse, landslides and other problems. The dumps of open-pit coal mines usually need to handle a large amount of excavated soil and rocks. Large-scale accumulation of soil and rocks may put pressure on the stability of the slope and increase the risk of slope sliding and collapse. At present, the common dump slope reinforcement technologies include geosynthetics, earthwork reinforcement and slope finishing, soil improvement technology, drainage system construction, rock anchors and face protection nets, monitoring and early warning systems, etc. These measures are costly to construct and maintain, which may affect the economic benefits of coal mining enterprises. For the slope stability of the dump, groundwater flow conditions play a vital role in the stability of the dump slope. The rise in groundwater levels caused by heavy monsoon rainfall is usually considered to be the cause of the unstable dump slope. Rainfall that penetrates into weathered soils triggers circular slope failure due to loss of cohesion, and surface water flow also affects the stability of the dump, increasing the seepage force during prolonged rainfall. Therefore, reducing slope sliding and the impact of water flow on the stability of dump slopes play an important role. Summary of the invention

[0004] In view of the above-mentioned deficiencies in the prior art, the present invention provides an open-pit coal mine dump slope stability optimization decision method, which improves the stability of the open-pit coal mine dump slope by predicting the stability of the open-pit coal mine dump slope.

[0005] In a first aspect, the present invention provides a method for optimizing the slope stability of an open-pit coal mine dump, comprising:

[0006] Based on the information of factors affecting the slope stability of the open-pit mine dump, a test data set is obtained;

[0007] Construct a feedforward neural network model to extract data features and perform nonlinear processing on the test data set;

[0008] Use genetic algorithm to optimize the weights and thresholds of feedforward neural network model;

[0009] Using mean square error as loss function, the feedforward neural network model is trained and verified to obtain a trained feedforward neural network model;

[0010] The trained feedforward neural network model is used to predict the dump data and obtain the slope stability decision-making plan.

[0011] According to an embodiment of the present invention, the information of factors affecting the slope stability of the open-pit mine dump includes: soil quality of the dump, rainfall in the dump, slope of the dump, and various indicators of drainage facilities in the dump.

[0012] According to one embodiment of the present invention, the feedforward neural network model includes an input layer, a one-dimensional convolutional layer, a hidden layer and an output layer.

[0013] According to one embodiment of the present invention, a method for extracting data features and performing nonlinear processing on a test data set includes:

[0014] Standardize the test data set to obtain a standard data set;

[0015] Different convolution kernels are set in the one-dimensional convolution layer to identify and extract local features of the standard data set;

[0016] The nonlinear relationship of the standard data set is captured by a nonlinear activation function in the hidden layer.

[0017] According to one embodiment of the present invention, a method for optimizing weights and thresholds of a feedforward neural network model using a genetic algorithm includes:

[0018] Randomly generate several groups of weight values ​​and thresholds in the network model as evaluation data sets;

[0019] The fitness of the data in the evaluation data set is calculated by the fitness function;

[0020] Through genetic operations, the data with the highest fitness is selected for crossover mutation to generate a new replacement data set;

[0021] Through iteration and convergence, the weight and threshold data with the highest fitness are selected as the weight and threshold of the feedforward neural network model.

[0022] According to one embodiment of the present invention, a method for training and verifying a feedforward neural network model using a mean square error as a loss function includes:

[0023] Initialize the weights and threshold data of the feedforward neural network model;

[0024] Feed the neural network model with batches to input the test dataset;

[0025] The test data set is propagated forward through each layer of the feedforward neural network model. The nodes of each layer perform weighted summation of the output of the previous layer, add a bias, and use an activation function to generate the output of the current layer.

[0026] Calculate the error between the output result and the actual label, use the mean square error as the loss function, and calculate the average error of each batch;

[0027] The gradient of the function with respect to each parameter is calculated through the back-propagation algorithm, and the network parameters are adjusted using the optimizer;

[0028] The forward propagation, loss calculation, and backpropagation process are repeated until the training cycle is completed.

[0029] According to one embodiment of the present invention, the method for training and verifying a feedforward neural network model using mean square error as a loss function further includes:

[0030] Use the test dataset to obtain prediction results through forward propagation;

[0031] Calculate the mean squared error between the predicted results and the actual labels;

[0032] Compare the validation loss with the training loss to determine whether the feedforward neural network model is overfitting and obtain the validation result;

[0033] According to the verification results, the network structure and training parameters of the feedforward neural network model are adjusted.

[0034] According to one embodiment of the present invention, the test data set includes a training set and a test set, and the ratio of the training set to the test set is 7:3.

[0035] A second aspect of the present invention provides a non-transitory computer-readable storage medium storing a computer program, which, when executed by a processor, implements a method for optimizing the decision-making of slope stability of an open-pit coal mine spoil dump.

[0036] The third aspect of the present invention provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements a method for optimizing the decision-making of the slope stability of an open-pit coal mine spoil dump when executing the computer program.

[0037] The technical solution of the present invention includes at least the following technical effects:

[0038] Through the construction of a deep learning model, the present invention can accurately analyze and predict the stability of the slope of an open-pit coal mine dump, and in particular can identify the most stable slope angle.

[0039] The application of this model can significantly improve the safety management of open-pit coal mine dumps, reduce potential safety risks, and improve the efficiency of dump design and monitoring.

[0040] The fast and accurate prediction results provided by the model can be directly used for decision support, helping mine managers to adjust operating strategies in a timely manner to ensure production safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 A flow chart of an open-pit coal mine dump slope stability optimization decision method provided by an embodiment of the present invention;

[0042] Figure 2 A schematic diagram of the slope stability angle of the open-pit coal mine dump provided by an embodiment of the present invention;

[0043] Figure 3 An IFNN structure diagram of a slope angle prediction model provided by an embodiment of the present invention;

[0044] Figure 4 The feedforward network layer structure of the IFNN model provided by the embodiment of the present invention;

[0045] Figure 5 A diagram of the data conversion process of a model data set provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0046] The present disclosure will now be discussed with reference to several exemplary embodiments. It should be understood that these embodiments are discussed only to enable those skilled in the art to better understand and thus implement the present disclosure, rather than implying any limitation on the scope of the present disclosure.

[0047] See also Figure 1-5 As shown in FIG. 1 , an optimization decision method for slope stability of an open-pit coal mine dump provided in an embodiment of the present application is provided, wherein, Figure 1 As shown, the method includes:

[0048] Based on the information of factors affecting the slope stability of the open-pit mine dump, a test data set is obtained;

[0049] Construct a feedforward neural network model to extract data features and perform nonlinear processing on the test data set;

[0050] Use genetic algorithm to optimize the weights and thresholds of feedforward neural network model;

[0051] Using mean square error as loss function, the feedforward neural network model is trained and verified to obtain a trained feedforward neural network model;

[0052] The trained feedforward neural network model is used to predict the dump data and obtain the slope stability decision-making plan.

[0053] Specifically, if Figure 2 As shown, four factors related to the slope stability of the dump are collected, and some stable open-pit coal mine dump information matching the present invention is collected at home and abroad to obtain its sample set. The method includes:

[0054] Investigate the slope information of different open-pit coal mine dumps based on oblique photogrammetry technology, based on the characteristics of UAV aerial survey technology, which is short in time, high in data collection accuracy, and high in three-dimensional visualization of results. Utilize UAV laser point cloud, advanced remote sensing sensor technology, telemetry and remote control technology, GPS differential positioning technology, etc., to obtain digital surface models and digital elevation models of open-pit coal mine dumps that have been dumped in place in a short time.

[0055] According to the acquired digital elevation model data, the terrain data, water network data and catchment area data of the open-pit coal mine dump are obtained, the slope and earthwork volume information of the open-pit coal mine dump that has been dumped in place are obtained, and the area information of the top platform and slope platform of the dump is obtained. Combined with the local climate characteristics, the rainfall data of the area where the dump is located is counted. Combined with the local geological conditions, the waste in the dump area is counted.

[0056] Data statistics were collected for different open-pit coal mine dumps, including soil shear strength, soil saturation, critical slope, and seepage per unit time. Based on oblique photogrammetry technology and other data acquisition methods, four factors were statistically analyzed and sorted out for 100 stable open-pit coal mine dumps at home and abroad.

[0057] The four influencing factors of soil shear strength, soil saturation, critical slope and seepage rate per unit time are expressed in a formula.

[0058] Specifically, four factors affecting the slope stability of open-pit coal mine dumps are analyzed. The indicators of dump soil quality, dump rainfall, dump slope and dump drainage facilities are formulated, mainly including:

[0059] When the shear stress at a point on any plane in the soil is equal to the shear strength of the soil, the point is in a critical state on the verge of destruction, and this critical state is called the "limit equilibrium state". The shear strength of soil refers to the ability of soil to resist shear failure. Its formula can be derived from the Coulomb criterion, and its basic form is:

[0060] τ=c+σtanφ;

[0061] In the formula, τ represents the shear strength of the soil, c represents the cohesion of the soil, and σ represents the normal stress of the soil. represents the internal friction angle of the soil.

[0062] Soil saturation refers to the ratio of the moisture content in the soil to the maximum water holding capacity of the soil, expressed as a percentage. The maximum water holding capacity of the soil refers to the maximum moisture content that the soil can maintain under natural conditions. When the rainfall is greater than the maximum water holding capacity of the soil, the excess water will seep into the groundwater or flow away, and the soil moisture content will not increase, but the soil saturation will decrease. Therefore, the formula for the change in soil saturation caused by rainfall can be expressed as:

[0063] Soil saturation = (rainfall / maximum soil water holding capacity) × 100%;

[0064] The calculation formula of the slope angle and critical value of the spoil dump can be derived based on the soil stability conditions. Specifically, the critical slope can be calculated using parameters such as the internal friction angle, cohesion, and bulk density of the soil. The internal friction angle is the most important parameter affecting the stability of soil. It reflects the friction between the particles inside the soil, while the cohesion C reflects the bonding force between the particles inside the soil. In actual engineering, these two parameters can be used to calculate the critical slope. The specific calculation formula is as follows:

[0065]

[0066] In the formula, It represents the tangent value of the internal friction angle, soil bulk density represents the unit volume weight of the soil, soil depth represents the depth of the soil, soil cohesion represents the bonding force between the internal particles of the soil, and soil width represents the width of the soil.

[0067] Darcy's Law describes the flow of water in soil, and dump drainage facilities are designed to control and direct the flow of water. Therefore, there is a certain relationship between Darcy's Law and dump drainage facilities. Darcy's Law is used to describe the flow of water in sandy soils, and its formula is expressed as:

[0068] Q = K*A*(h / L);

[0069] Where: Q is the seepage rate per unit time, K is the permeability coefficient, A is the cross-sectional area of ​​the water flow, h is the total head loss, and L is the length of the seepage path.

[0070] The data of the four factors obtained were normalized to ensure the input of the model.

[0071]

[0072] In the formula, norm is the normalized result of the influencing factors, x i is the original data of the influencing factor, max is the maximum value of the influencing factor, min is the minimum value of the influencing factor, and the pixel value is adjusted to the (0, 1) interval after normalization.

[0073] The data of the four factors obtained are organized into serialized numerical strings, and the 100 sample sets obtained are classified into training set and validation set with a classification ratio of 7:3.

[0074] The feedforward neural network (IFNN) model was constructed. By consulting the data, the four factors that affect the slope stability of the open-pit coal mine dump are not related to each other, and the comprehensive impact on the slope stability is caused. Because the dump has stronger nonlinear, time-varying and non-stationary characteristics, the uncertainty analysis method is used to analyze the stability of the dump.

[0075] The feedforward neural network is composed of a single-layer input layer, multiple hidden layers and a single-layer output layer. The input layer neurons are mainly used to receive various input information, the hidden layer neurons are mainly used for spatial mapping transformation of input information, and the output layer neurons are mainly used for output response. Inside the deep feedforward neural network, the parameters are propagated forward from the input layer to the output layer in a unidirectional manner. Unlike other neural networks, no directed loop is formed inside. The improved deep feedforward neural network (IFNN) model includes an input layer with 4 nodes, a one-dimensional convolutional layer, two hidden layers and an output layer. After the one-dimensional convolutional layer of the convolutional neural network is introduced into the input layer of the deep feedforward neural network, it is used to extract local features. By using the one-dimensional convolutional layer, the local feature map of such data can be better extracted and used as the input of the next layer of the network. As the depth of the convolutional layer increases, the neural network will learn features of different dimensions and use the ReLU function for nonlinear processing. The present invention uses the mean square error function as the loss function of the model to calculate the degree of difference between the true value and the model output, thereby judging the degree of convergence of the model. It is generally believed that the smaller the value of the mean square error, the closer the predicted value of the model is to the true value. During the model training process, in order to avoid the loss function falling into the local minimum, the present invention uses the Adam optimizer in the PyTorch deep learning library to adaptively adjust the learning rate to further ensure the convergence of the model. The training of deep learning models requires a large amount of data. Therefore, based on literature analysis and on-site exploration, this patent obtains sample data of 100 spoil dump models, 70% of the samples are used as training data sets, and 30% of the samples are used as verification data sets.

[0076] Improved IFNN network model based on Pytorch design: Figure 3 , 4 , 5, the specific steps include:

[0077] IFNN is used as the basic network architecture. The four parameters of soil shear strength, soil saturation, critical slope and seepage per unit time correspond to the input features of the output nodes, which are recorded as X=[x1,x2,x3,x4]. A one-dimensional convolutional layer of the convolutional neural network is introduced after the input layer. A total of three layers of one-dimensional convolutional layers are introduced. The number of output filters is set to 10, 20 and 30 respectively, and the size of the convolution kernel is set to 3×1. Taking the Heidaigou open-pit coal mine as an example, the specific process is that the preprocessed data set is filled in the "same" way to become 16 layers; then, after data conversion, a 4×4 grayscale image is obtained and passed to the input layer. The data conversion process is as follows: Figure 4As shown in the figure, the input layer will be convolved with a convolution kernel of size 3×1 to complete the dimension increase of the input feature layer; next, the nonlinear activation function ReLU is used for nonlinear processing; the hidden layer is located between the input layer and the output layer, after the convolution layer, and is composed of multiple neurons, which is used to extract and transform the input data. Among them, Layer1 extracts the original data into dimensionless feature indicators, and further extracts the features of the data through the algorithm conversion between Layer1 and Layer2, and finally passes the results to the output layer. The output layer is used to output the final optimal slope angle result.

[0078] The activation function is used to control the output of neurons and improve the learning ability of IFNN for nonlinear problems. The ReLU function improves the training speed of the neural network and helps the network adjust parameters more stably. Therefore, this model uses the ReLU function as the activation function to make the nonlinear relationship between the FNN learning results and its influencing factors.

[0079] f(x)=max(0,x);

[0080] Where f(x) is the output value of the neuron; x is the input value of the neuron.

[0081] IFNN propagates through continuous iterations and finally obtains the model output.

[0082] z (l) =W (l) f l-1 (z (l-1) )+b (l) ;

[0083] a (l) =f l (W (l) a (l-1) +b (l) );

[0084] In the formula, z is the static input of the neuron; l represents the number of layers of IFNN; W is the weight matrix; f is the activation function; b is the bias parameter; and a is the output of the neuron. In IFNN, the weight matrix W is a parameter for stretching the feature vector, and its value is a random number in the range of 0 to 1. During the model training process, the W of each network layer is updated through the back propagation of the feedforward neural network to improve the calculation accuracy:

[0085]

[0086] Where W new is the updated weight matrix; α is the learning rate; is the loss function; Y and are the output vector of the true value and the output vector of the predicted value of the slope results in the training set, respectively.

[0087] Genetic algorithm is used to optimize the feedforward neural network model. The weights and thresholds of the feedforward neural network are random, which has a certain impact on the convergence speed and is not easy to converge. Taking advantage of the strong global search ability of the genetic algorithm, its weights and thresholds are optimized to improve the network convergence speed and enhance the accuracy and stability of the prediction model. The genetic algorithm is a global optimization random search algorithm that simulates the selection, crossover and mutation process in the biological genetic process through genetic operators to obtain the optimal individual. The steps are: initialize the population. The individual encoding is a real number encoding, which consists of the connection weights between the input layer and the hidden layer, the hidden layer threshold, the connection weights between the hidden layer and the output layer, and the output layer threshold; calculate the fitness function, obtain the initial weights and thresholds of the neural network according to the individual, train the neural network with training data, and predict the system output. The absolute value of the error between the predicted output and the expected output is used as the individual fitness; perform a selection operation and select the roulette algorithm as the genetic optimization algorithm; perform a crossover operation, and select the real crossover method for the crossover operation. The feedforward neural network model was optimized using a genetic algorithm. The genetic algorithm parameter settings were: population size was 10, evolutionary generations were 50, crossover probability was 0.4, and mutation probability was 0.2.

[0088] The network model is trained using statistical data. The specific steps include:

[0089] Select the loss function. Set the initial learning rate of IFNN to 0.01. During the training process, the model continuously adjusts the weight matrix and bias parameters to make the training output closer to the true value and reduce the loss during the model training process. This model uses the mean squared error (MSE) function as the loss function of the model to calculate the difference between the true value and the model output, so as to judge the degree of convergence of the model. It is generally believed that the smaller the MSE value, the closer the model's predicted value is to the true value:

[0090]

[0091] In the formula, n represents the number of samples in each training batch; y i and are the true value and predicted value of the model slope result in each training batch, respectively.

[0092] Select the optimizer. During the model training process, in order to prevent the loss function from falling into the local minimum, this model uses the Adam optimizer in the PyTorch deep learning library to adaptively adjust the learning rate (α) to further ensure the convergence of the model. The Adam optimizer combines the Momentum algorithm and the RMSProp algorithm, so that it can handle situations with large amounts of data or parameters and improve the model calculation efficiency.

[0093]

[0094] In the formula, θ t represents the current time step; θ t-1 Indicates the time step of the previous iteration; m t represents the first-order moment estimate of the corrected gradient; represents the second-order moment estimate of the corrected gradient; ε is a hyperparameter of the Adam optimizer, and its default value is 1×10 -8 .

[0095] The present invention has the following advantages and beneficial effects:

[0096] The present invention designs a new feedforward neural network structure (IFNN) by improving the IFNN network model, and predicts the stability of the slope of the open-pit coal mine dump based on the new network structure. The new feedforward neural network consists of an input layer, a convolution layer, a feedforward network layer, and an output layer. The feedforward neural network first adjusts the weight matrix and bias parameters in the algorithm by setting parameters such as activation function, learning rate, hidden layer, etc., which are used to simulate various nonlinear functions, and then optimizes the algorithm, optimizes the feedforward neural network model using a genetic algorithm, optimizes its weights and thresholds, improves the network convergence speed, and enhances the accuracy and stability of the prediction model. The IFNN model analyzes the dump soil quality, dump rainfall, dump slope, and dump drainage facilities input into the model through the three parts of the selection of activation function, the selection and optimization of the feedforward neural network algorithm, and the training of the feedforward neural network, and finally obtains the predicted result of the slope and the real result display. The model aims to analyze the stability of the slope of the open-pit coal mine dump, and ultimately output the slope result that makes the slope of the open-pit coal mine dump the most stable. As an optimization decision-making method for the stability of the slope of the open-pit coal mine dump, it provides a deep learning model result selection for the stability of the slope of the open-pit coal mine dump.

[0097] The present invention formulates several factors that affect the stability of the slope of the open-pit coal mine dump, and constructs a deep learning model for the complex nonlinear relationship that affects its stability. Ultimately, the most stable angle result for the slope angle of the open-pit coal mine dump can be obtained, which can be quickly and efficiently applied to the verification of the stability angle of the slope of the open-pit coal mine dump, and is of great significance to the safety of the open-pit coal mine dump and the safety of coal mine production.

[0098] The present invention also provides a non-transient computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, a method for optimizing the decision-making of the slope stability of an open-pit coal mine spoil dump is implemented.

[0099] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements an optimization decision method for slope stability of an open-pit coal mine spoil dump when executing the computer program.

[0100] Those of ordinary skill in the art will appreciate that the modules and algorithm steps described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0101] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described devices and equipment can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0102] In the embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the modules is only a logical function division. There may be other division methods in actual implementation, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or modules, which can be electrical, mechanical or other forms.

[0103] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical modules, that is, they may be located in one place or distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the embodiments of the present invention.

[0104] In addition, each functional module in the embodiment of the present invention may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module.

[0105] If the functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the energy-saving signal sending / receiving method of each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a ROM, a RAM, a magnetic disk, or an optical disk.

[0106] The above description is only a preferred embodiment of the present application and an explanation of the technical principles used. Those skilled in the art should understand that the scope of the invention involved in the present application is not limited to the technical solution formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the inventive concept. For example, the above features are replaced with (but not limited to) technical features with similar functions disclosed in the present application.

[0107] It should be understood that the size of the sequence number of each step in the content of the invention and the embodiments of the present invention does not absolutely mean the 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 embodiment of the present invention. For the purpose of example and description, the foregoing description of the implementation of the present disclosure has been given. The foregoing description is not exhaustive and is not intended to limit the present disclosure to the exact form disclosed. Various deformations and modifications may exist according to the above teachings, or various deformations and modifications may be obtained from the practice of the present disclosure. These embodiments are selected and described to illustrate the principles of the present disclosure and its practical application, so that those skilled in the art can use the present disclosure in various embodiments and various modifications suitable for the specific purpose conceived.

Claims

1. A method for optimizing the slope stability of an open-pit coal mine dump, characterized in that: The methods include: Based on the information of factors affecting the slope stability of the open-pit mine dump, a test data set is obtained; Construct a feedforward neural network model to extract data features and perform nonlinear processing on the test data set; Use genetic algorithm to optimize the weights and thresholds of feedforward neural network model; Using mean square error as loss function, the feedforward neural network model is trained and verified to obtain a trained feedforward neural network model; The trained feedforward neural network model is used to predict the dump data and obtain the slope stability decision-making plan.

2. The method for optimizing the slope stability of an open-pit coal mine dump as claimed in claim 1, characterized in that: The factors that affect the slope stability of the open-pit mine spoil dump include: soil quality of the spoil dump, rainfall in the spoil dump, slope of the spoil dump and various indicators of the drainage facilities of the spoil dump.

3. The method for optimizing the slope stability of an open-pit coal mine dump as claimed in claim 1, characterized in that: The feedforward neural network model includes an input layer, a one-dimensional convolutional layer, a hidden layer and an output layer.

4. The method for optimizing the slope stability of an open-pit coal mine dump as claimed in claim 3, characterized in that: Methods for data feature extraction and nonlinear processing of test data sets include: Standardize the test data set to obtain a standard data set; Different convolution kernels are set in the one-dimensional convolution layer to identify and extract local features of the standard data set; The nonlinear relationship of the standard data set is captured by a nonlinear activation function in the hidden layer.

5. The method for optimizing the slope stability of an open-pit coal mine dump as claimed in claim 1, characterized in that: Methods for optimizing weights and thresholds of feedforward neural network models using genetic algorithms include: Randomly generate several groups of weight values ​​and thresholds in the network model as evaluation data sets; The fitness of the data in the evaluation data set is calculated by the fitness function; Through genetic operations, the data with the highest fitness is selected for crossover mutation to generate a new replacement data set; Through iteration and convergence, the weight and threshold data with the highest fitness are selected as the weight and threshold of the feedforward neural network model.

6. The method for optimizing the slope stability of an open-pit coal mine dump as claimed in claim 1, characterized in that: Methods for training and validating feedforward neural network models using mean squared error as the loss function include: Initialize the weights and threshold data of the feedforward neural network model; Feed the neural network model with batches to input the test dataset; The test data set is propagated forward through each layer of the feedforward neural network model. The nodes of each layer perform weighted summation of the output of the previous layer, add a bias, and use an activation function to generate the output of the current layer. Calculate the error between the output result and the actual label, use the mean square error as the loss function, and calculate the average error of each batch; The gradient of the function with respect to each parameter is calculated through the back-propagation algorithm, and the network parameters are adjusted using the optimizer; The forward propagation, loss calculation, and backpropagation process are repeated until the training cycle is completed.

7. The method for optimizing the slope stability of an open-pit coal mine dump as claimed in claim 1, characterized in that: The method of training and validating the feedforward neural network model using mean square error as the loss function also includes: Use the test dataset to obtain prediction results through forward propagation; Calculate the mean squared error between the predicted results and the actual labels; Compare the validation loss with the training loss to determine whether the feedforward neural network model is overfitting and obtain the validation result; According to the verification results, the network structure and training parameters of the feedforward neural network model are adjusted.

8. The method for optimizing the slope stability of an open-pit coal mine dump as claimed in claim 1, characterized in that: The test data set includes a training set and a test set, and the ratio of the training set to the test set is 7:

3.

9. A non-transitory computer-readable storage medium, characterized in that: The non-transitory computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method for optimizing the decision-making of the slope stability of an open-pit coal mine spoil dump as described in any one of claims 1 to 8 is implemented.

10. An electronic device, characterized in that: The method comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the method for optimizing the decision-making method for the slope stability of an open-pit coal mine spoil dump as described in any one of claims 1 to 8 is implemented.

Images