High and steep slope near area particle peak velocity prediction method and device and electronic equipment

Abstract

The invention provides a high and steep slope near area particle peak velocity prediction method and device and electronic equipment, and relates to the technical field of blasting excavation, and the method comprises the steps: determining a first variable parameter according to an attenuation index between a current measurement point and a current explosion source and an equivalent explosive quantity, a second variable parameter is determined according to the attenuation index between the current measuring point and the current explosion source and the explosion source distance, and a third variable parameter is determined according to the preset coefficient between the current measuring point and the current explosion source and the mass point peak velocity obtained by the current measuring point; and inputting the first variable parameter, the second variable parameter, the third variable parameter and the current blasting parameter to a preset CFOA-BP-ANN model to obtain a particle peak velocity predicted value. The mass point peak velocity of all the protected objects is monitored, resources are saved, the blasting vibration monitoring efficiency and quality of the protected objects are improved, the blasting effect can be improved and blasting vibration can be reduced by optimizing blasting parameters, the site construction efficiency is improved, and the safety of surrounding facilities is ensured.

Description

Technical Field

[0001] The invention relates to the technical field of blasting excavation, and in particular to a method, a device and an electronic device for predicting the peak velocity of particles in an area adjacent to a high and steep slope. Background Art

[0002] Compared with mechanical excavation, blasting excavation is the most effective excavation method. However, while blasting excavation has many advantages, it will inevitably produce various harmful effects such as blasting vibration, blasting flying rocks, blasting poisonous gas, etc. Among them, blasting vibration is one of the most serious hazards, which will affect the surrounding geological environment and structures. Therefore, when blasting excavation is carried out, if there are important facilities and equipment nearby, it is necessary to monitor their vibration velocity. According to the "Safety Regulations for Blasting GB6722-2014", different protected objects correspond to different particle peak velocity specification values, such as adobe houses (PPV≦0.45cm / s), houses, bridges (PPV≦2.0cm / s), and traffic tunnels (PPV≦12cm / s). Therefore, blasting vibration monitoring of protected objects is of great significance.

[0003] However, due to the conditions of instruments and resources, it is often impossible to conduct comprehensive vibration monitoring of all protected objects in actual engineering blasting. The peak velocity of particles in the vicinity of the blasting area under different blasting parameters cannot be accurately predicted, which not only wastes resources but also reduces the efficiency and quality of blasting vibration monitoring of the protected objects, making it impossible to guarantee the on-site construction efficiency and the safety of surrounding facilities.

[0004] In summary, there is currently no technical solution that can solve the above technical problems, and there is no method, device or electronic equipment for predicting the peak velocity of particles in the vicinity of a high and steep slope. Summary of the invention

[0005] The present invention provides a method, device and electronic equipment for predicting the peak velocity of particles in the vicinity of a high and steep slope, which can accurately predict the peak velocity of particles in any vicinity of a blasting area under different blasting parameters by combining a novel fishing optimization algorithm (CFOA), a BP neural network (BP-ANN), a Sadovsky empirical formula (SAS) and the least squares transformation principle.

[0006] In a first aspect, the present invention provides a method for predicting the peak velocity of particles in an area adjacent to a high and steep slope, comprising:

[0007] The first variable parameter is determined according to the attenuation index between the current measuring point and the current explosion source and the equivalent charge amount, the second variable parameter is determined according to the attenuation index between the current measuring point and the current explosion source and the explosion source distance, and the third variable parameter is determined according to the preset coefficient between the current measuring point and the current explosion source and the peak velocity of the particle obtained at the current measuring point;

[0008] Input the first variable parameter, the second variable parameter, the third variable parameter, and the current blasting parameters into the preset CFOA - BP - ANN model to obtain the predicted value of the peak particle velocity output by the preset CFOA - BP - ANN model;

[0009] The preset CFOA - BP - ANN model is determined after training the initial CFOA - BP - ANN model based on the sample first parameter, sample second parameter, sample third parameter, sample blasting parameters, and sample predicted values;

[0010] The sample first parameter, sample second parameter, and sample third parameter are determined after taking the logarithm of both sides of the Sadovskii empirical formula using the least - squares method and performing equivalent transformations.

[0011] According to the method for predicting the peak particle velocity in the vicinity of a high - steep slope provided by the present invention, the current blasting parameters include the aperture size, bench height, minimum burden size, toe burden size, hole spacing size, row spacing size, hole depth value, stemming depth value, charge length, stemming length, hole inclination angle, hole density coefficient, decoupling coefficient, explosive specific consumption, single - hole charge amount, maximum charge amount per delay section, total charge amount, hole - to - hole delay value, row - to - row delay value, and total number of holes.

[0012] According to the method for predicting the peak particle velocity in the vicinity of a high - steep slope provided by the present invention, taking the logarithm of both sides of the Sadovskii empirical formula using the least - squares method and performing equivalent transformations includes:

[0013]

[0014] Among them, the sample first parameter is The sample second parameter is αln R, the sample third parameter is ln P PV - lnK, where K is a preset coefficient between the sample measurement point and the sample explosion source, Q is the sample equivalent charge amount, R is the distance from the sample explosion source, α is the attenuation index between the sample measurement point and the sample explosion source, and PPV is the peak particle velocity of the sample.

[0015] According to the method for predicting the peak particle velocity in the vicinity of a high - steep slope provided by the present invention, before inputting the first variable parameter, the second variable parameter, the third variable parameter, and the current blasting parameters into the preset CFOA - BP - ANN model, the method further includes:

[0016] Using the preset CFOA (Cuckoo Search Optimization Algorithm) to optimize the hyperparameters of the BP neural network to obtain the initial CFOA - BP - ANN model;

[0017] The hyperparameters include the learning rate η, the regularization parameter λ, the number of layers L of the neural network, the number of neurons j in each hidden layer, the number of learning epochs Epoch, the size of the mini-batch data Batch Size, the optimizer, the weight initialization method, and the type of neuron activation function.

[0018] According to the method for predicting the peak particle velocity in the vicinity of a high-steep slope provided by the present invention, the preset fishing optimization algorithm CFOA includes an exploration stage and a development stage;

[0019] The use of the preset fishing optimization algorithm CFOA to optimize the hyperparameters of the BP neural network to obtain the initial CFOA-BP-ANN model includes:

[0020] In the exploration stage, the capture rate parameter is used to simulate fishermen to establish advantages through autonomous search and disturbance of the water area, and group enclosure is used as an auxiliary search method. If the capture situation at the current location is the same as the capture speed parameter, local search is performed; otherwise, follow search is performed.

[0021] In the development stage, a Gaussian distribution is used to simulate the update process.

[0022] According to the method for predicting the peak particle velocity in the vicinity of a high-steep slope provided by the present invention, the use of the capture rate parameter to simulate fishermen to establish advantages through autonomous search and disturbance of the water area, and group enclosure as an auxiliary search method. If the capture situation at the current location is the same as the capture speed parameter, local search is performed; otherwise, follow search is performed, includes:

[0023]

[0024] Where α is the capture rate parameter, CF is the current evaluation quantity, and MaxCF is the maximum evaluation value;

[0025]

[0026] Where is a group of fishermen with 3 or 4 people, C is the center point, and are the positions of the i-th fisherman in the j-th dimension after the (T + 1)-th and T-th updates in the group; r2 is the speed of the fisherman approaching the center, and its value range is (0, 1); r3 is the offset of the movement, and its value range is (-1, 1).

[0027] According to the method for predicting the peak particle velocity in the vicinity of a high-steep slope provided by the present invention, the use of a Gaussian distribution to simulate the update process includes:

[0028]

[0029] Where GD is the Gaussian distribution function, σ is the population variance, CF is the current evaluation quantity, MaxCF is the maximum evaluation value, Gbest is the global optimal position, r4 is a random number of 1, 2, or 3, and mean(Fisher) is the matrix of the mean values of each dimension of the fisherman center. is the position of the i-th fisherman at the (T + 1)-th update.

[0030] According to the method for predicting the peak particle velocity in the adjacent area of a high-steep slope provided by the present invention, before inputting the first variable parameter, the second variable parameter, the third variable parameter, and the current blasting parameter into the preset CFOA-BP-ANN model, the method further includes:

[0031] Standardize the first variable parameter, the second variable parameter, the third variable parameter, and the current blasting parameter to obtain the standardized first variable parameter, the second variable parameter, the third variable parameter, and the current blasting parameter.

[0032] In a second aspect, a device for predicting the peak particle velocity in the adjacent area of a high-steep slope is provided, including:

[0033] A determination unit, which is used to determine the first variable parameter according to the attenuation index and the equivalent charge amount between the current measuring point and the current blasting source, determine the second variable parameter according to the attenuation index and the blasting source distance between the current measuring point and the current blasting source, and determine the third variable parameter according to the preset coefficient between the current measuring point and the current blasting source and the peak particle velocity obtained at the current measuring point;

[0034] An input unit, which is used to input the first variable parameter, the second variable parameter, the third variable parameter, and the current blasting parameter into the preset CFOA-BP-ANN model to obtain the predicted value of the peak particle velocity output by the preset CFOA-BP-ANN model;

[0035] The preset CFOA-BP-ANN model is determined after training the initial CFOA-BP-ANN model according to the sample first parameter, the sample second parameter, the sample third parameter, the sample blasting parameter, and the sample predicted value;

[0036] The sample first parameter, the sample second parameter, and the sample third parameter are determined after equivalent transformation by taking the logarithm of both sides of the Sadovskii empirical formula using the least squares method.

[0037] In a third aspect, an electronic device is provided, including a memory, a processor, and a computer program stored on the memory and executable on the processor, and when the processor executes the program, it implements the method for predicting the peak particle velocity in the adjacent area of a high-steep slope.

[0038] The present invention can achieve the monitoring of the particle peak velocity of all protected objects, which not only saves resources but also improves the efficiency and quality of blasting vibration monitoring of protected objects. Moreover, it can also improve the blasting effect and reduce blasting vibration by optimizing blasting parameters, improve the on-site construction efficiency and ensure the safety of surrounding facilities.

[0039] The present invention introduces a new meta-heuristic optimization algorithm - the new fish swarm optimization algorithm CFOA to optimize the hyperparameters of the BP neural network, thereby establishing a CFOA-BP-ANN prediction model, making the CFOA-BP-ANN prediction model have higher prediction accuracy. Using the CFOA-BP-ANN prediction model can achieve vibration prediction at any measurement point, which not only saves monitoring costs and resources but also improves the efficiency and quality of monitoring. Brief Description of the Drawings

[0040] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0041] Figure 1 It is a flow schematic diagram of the method for predicting the particle peak velocity in the area adjacent to a high and steep slope provided by the present invention;

[0042] Figure 2 It is a technical roadmap of the method for predicting the particle peak velocity in the area adjacent to a high and steep slope provided by the present invention;

[0043] Figure 3 It is a fitness convergence curve diagram of different algorithms provided by the present invention;

[0044] Figure 4 It is a comparison diagram of the calculation time of different algorithms provided by the present invention;

[0045] Figure 5 It is a schematic diagram of collective search provided by the present invention;

[0046] Figure 6 It is a schematic diagram of the BP neural network provided by the present invention;

[0047] Figure 7 It is a comparison diagram of R2 of different models provided by the present invention;

[0048] Figure 8 It is a comparison diagram of RMSE of different models provided by the present invention;

[0049] Figure 9It is a schematic diagram of the monitoring points provided by the present invention;

[0050] Figure 10 It is a schematic structural diagram of a device for predicting the peak particle velocity in the vicinity of a high-steep slope provided by the present invention;

[0051] Figure 11 It is a schematic structural diagram of an electronic device provided by the present invention. Detailed implementation manners

[0052] To make the objectives, technical solutions and advantages of the present invention clearer, the technical solutions in the present invention will be clearly and completely described below with reference to the accompanying drawings in the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present invention without making creative efforts shall fall within the protection scope of the present invention.

[0053] Figure 1 It is a schematic flowchart of a method for predicting the peak particle velocity in the vicinity of a high-steep slope provided by the present invention. The method for predicting the peak particle velocity in the vicinity of a high-steep slope includes:

[0054] Step 101: Determine a first variable parameter according to the attenuation index and equivalent charge amount between the current measurement point and the current blast source, determine a second variable parameter according to the attenuation index and blast source distance between the current measurement point and the current blast source, and determine a third variable parameter according to the preset coefficient between the current measurement point and the current blast source and the peak particle velocity obtained at the current measurement point;

[0055] Step 102: Input the first variable parameter, the second variable parameter, the third variable parameter, and the current blasting parameter into a preset CFOA-BP-ANN model to obtain a predicted value of the peak particle velocity output by the preset CFOA-BP-ANN model;

[0056] The preset CFOA-BP-ANN model is determined after training an initial CFOA-BP-ANN model according to sample first parameters, sample second parameters, sample third parameters, sample blasting parameters, and sample predicted values;

[0057] The sample first parameter, sample second parameter, and sample third parameter are determined after taking the logarithm of both sides of the Sadovskii empirical formula using the least squares method and performing equivalent transformation.

[0058] In step 101, since the first sample parameter, the second sample parameter, and the third sample parameter are determined after taking the logarithm of both sides of the Sadovskii empirical formula and performing equivalent transformation using the least squares method, and the determination methods of the first variable parameter, the second variable parameter, and the third variable parameter correspond to those of the first sample parameter, the second sample parameter, and the third sample parameter, the first variable parameter is determined according to the attenuation exponent (α) and the equivalent charge (Q) between the current measuring point and the current blast source. This involves performing logarithmic transformation and equivalent transformation on the Sadovskii empirical formula to obtain the first variable parameter related to the equivalent charge and the attenuation exponent; the second variable parameter is determined according to the attenuation exponent (α) and the blast source distance (R) between the current measuring point and the current blast source. This also requires performing logarithmic transformation and equivalent transformation on the Sadovskii empirical formula to obtain the second variable parameter related to the blast source distance and the attenuation exponent; the third variable parameter is determined according to the preset coefficient (K) between the current measuring point and the current blast source and the peak particle velocity (PPV) obtained at the current measuring point. Here, the preset coefficient may be related to the terrain and geological conditions, and the peak particle velocity is obtained through actual measurement.

[0059] Optionally, the Sadovskii empirical formula is reduced in dimension by the least squares principle to obtain new variables and add the corresponding values of all blasting parameters, and then all data are standardized. Finally, it is used as the input of the CFOA-BP-ANN model for the next step of model training. Specifically, in blast vibration prediction, the Sadovskii empirical formula is one of the most commonly used formulas, mainly involving the peak particle velocity PPV, the equivalent charge Q, and the blast source distance R, as shown below:

[0060]

[0061] In the formula, PPV is the peak particle velocity; Q is the equivalent charge; R is the blast source distance; K and α are the coefficient and attenuation exponent related to the terrain and geological conditions between the measuring point and the blast source.

[0062] By the least squares principle, taking the logarithm of both sides of the above formula, we can get:

[0063]

[0064] After taking the logarithm of both sides of the Sadovskii empirical formula using the least squares method and performing equivalent transformation, it includes:

[0065]

[0066] Among them, the first sample parameter is The second sample parameter is αln R, the third sample parameter is ln P PV - ln K, where K is a preset coefficient between the sample measurement point and the sample explosion source, Q is the sample equivalent charge, R is the distance from the sample explosion source, α is the attenuation index between the sample measurement point and the sample explosion source, and PPV is the peak particle velocity of the sample.

[0067] Optionally, the current blasting parameters include hole diameter size, bench height, minimum burden size, toe burden size, hole spacing size, row spacing size, hole depth value, stemming depth value, charge length, stemming length, drilling inclination angle, hole density coefficient, decoupling coefficient, explosive specific consumption, single-hole charge, maximum charge per delay section, total charge, hole-to-hole delay value, row-to-row delay value, and total number of holes. The 23 parameters of the current blasting parameters, the first variable parameter, the second variable parameter, and the third variable parameter are jointly used as the model input.

[0068] Optionally, before inputting the first variable parameter, the second variable parameter, the third variable parameter, and the current blasting parameters into the preset CFOA - BP - ANN model, the method further includes:

[0069] Performing standardization processing on the first variable parameter, the second variable parameter, the third variable parameter, and the current blasting parameters to obtain the standardized first variable parameter, second variable parameter, third variable parameter, and current blasting parameters.

[0070] To eliminate the influence of dimensions, it is necessary to perform standardization processing on all inputs. The standardization processing formula is:

[0071]

[0072] where x is the parameter for standardization processing; xmin is the minimum value in each group of parameters; xmax is the maximum value in each group of parameters.

[0073] In step 102, the first variable parameter, the second variable parameter, the third variable parameter, and the current blasting parameters (such as hole diameter size, bench height, minimum burden size, etc.) determined in step 101 are input into the preset CFOA - BP - ANN model to obtain the predicted value of the peak particle velocity output by the preset CFOA - BP - ANN model. The CFOA - BP - ANN model is a prediction model that combines the new fish - swarm optimization algorithm (CFOA), the BP neural network (BP - ANN), and the Sadovskii empirical formula (SAS). This model optimizes the hyperparameters of the BP - ANN through CFOA to improve the prediction accuracy.

[0074] Optionally, the CFOA-BP-ANN model can comprehensively consider various blasting parameters and influencing factors of the peak particle velocity, so as to more accurately predict the peak particle velocity in any adjacent blasting area. By introducing the CFOA algorithm to optimize the hyperparameters of the BP neural network, the prediction accuracy and training efficiency of the model can be significantly improved. In an optional embodiment, other factors related to blasting vibration (such as explosive energy distribution, geological structure, etc.) can be considered as additional input parameters and incorporated into the CFOA-BP-ANN model to further improve the prediction accuracy. The CFOA algorithm can also be improved or other optimization algorithms (such as genetic algorithm, particle swarm algorithm, etc.) can be introduced to further optimize the hyperparameters of the BP neural network and improve the prediction performance of the model.

[0075] The present invention predicts the peak particle velocity by comprehensively considering multiple parameters (such as attenuation exponent, equivalent charge amount, distance from the blast source, preset coefficient, etc.) between the current measurement point and the current blast source, and combining complex blasting parameters (such as hole diameter size, bench height, minimum resistance line size, etc.), using a specially trained CFOA-BP-ANN model. This prediction method that comprehensively considers multiple influencing factors can significantly improve the prediction accuracy and reliability compared with traditional single-factor or simple empirical formula prediction.

[0076] When constructing the CFOA-BP-ANN model, the present invention uses the least squares method to perform an equivalent transformation on the Sadovsky empirical formula to obtain the first sample parameter, the second sample parameter, and the third sample parameter, which are used as important inputs for model training. At the same time, the CFOA (Catching Fish Optimization Algorithm) is used to optimize the hyperparameters of the BP neural network to obtain the optimal initial CFOA-BP-ANN model. This model training method not only improves the generalization ability of the model but also reduces the risk of model overfitting. During the prediction process, only the first variable parameter, the second variable parameter, the third variable parameter, and the current blasting parameters need to be input into the preset CFOA-BP-ANN model to quickly obtain the predicted value of the peak particle velocity, which not only simplifies the prediction process but also shortens the prediction time and improves the prediction efficiency.

[0077] Since the present invention comprehensively considers multiple influencing factors and uses advanced machine learning algorithms for model training and prediction, it has a wide range of applications and can be used in different types of blasting projects and scenarios. It also has good scalability and flexibility, and the parameters can be adjusted and the model can be optimized according to actual needs.

[0078] Figure 2It is the technical roadmap of the method for predicting the peak particle velocity in the vicinity of high-steep slopes provided by the present invention. The present invention aims to optimize the blasting parameters for vibration reduction during the excavation of high-steep slopes, thereby reducing the impact of blasting vibration on nearby structures and improving the efficiency of blasting vibration monitoring. A method for accurately predicting the peak particle velocity of any blasting area in the vicinity under different blasting parameters is proposed, which combines the novel fishing optimization algorithm (CFOA), BP neural network (BP-ANN), Sadovsky empirical formula (SAS), and the least squares transformation principle. In this method, CFOA is used to optimize the hyperparameters of BP-ANN, and then the least squares method is used to reduce the dimension of SAS to establish new variables. Then, the data obtained by standardizing all blasting parameters is used as the input of the prediction model and brought into the CFOA-BP-ANN model for training and testing. Combined with the measured data for correction, the prediction of the peak particle velocity of any blasting area in the vicinity is finally completed. This method can effectively predict the vibration of blasting on surrounding structures and geology, improve the monitoring efficiency and quality of blasting vibration, and secondly, improve the blasting effect and reduce blasting vibration by optimizing blasting parameters, improve the on-site construction efficiency and ensure the safety of surrounding facilities.

[0079] The present invention introduces a novel metaheuristic optimization algorithm based on human behavior - the novel fishing optimization algorithm CFOA to optimize the hyperparameters of the BP neural network, thereby establishing a CFOA-BP-ANN prediction model. CFOA is mainly divided into an exploration stage and a development stage, and the development stage includes an independent search method and a collective search method. Compared with recently proposed optimization algorithms, such as the sparrow search algorithm SSA, slime mold algorithm SMA, whale optimization algorithm GOA, mayfly optimization algorithm MA, and artificial bee colony algorithm ABC, the CFOA algorithm shows better robustness and optimization performance, and the comparison of the fitness convergence curve and optimization time is as Figure 3 and Figure 4 shown. Therefore, introducing CFOA can improve the hyperparameter optimization efficiency of the BP neural network, making the CFOA-BP-ANN prediction model have higher prediction accuracy.

[0080] Optionally, the preset fishing optimization algorithm CFOA includes an exploration stage and a development stage. Using the preset fishing optimization algorithm CFOA to optimize the hyperparameters of the BP neural network to obtain the initial CFOA-BP-ANN model includes:

[0081] In the exploration stage, the capture rate parameter is used to simulate fishermen to establish advantages through autonomous search and disturbing the water area. Group encirclement is used as an auxiliary search method. If the capture situation at the current location is the same as the capture speed parameter, local search is performed; otherwise, following search is performed.

[0082] In the development stage, the Gaussian distribution is used to simulate the update process.

[0083] When the new fishing optimization algorithm CFOA is optimizing, it systematically divides the entire optimization process into three different strategies: independent search, grouped capture, and collective capture. In CFOA, each fisherman corresponds to a search agent, and its model is as follows:

[0084]

[0085] Among them, Fisher is a matrix of N search positions in the d-dimensional space, and its initialization is as follows:

[0086] Fisher i,j =(ub j -lb j )*r + lb j (1)

[0087] In the exploration stage, fishermen initially mainly establish advantages by independent search and disturbing the water area, and group encirclement is used as an auxiliary search method. The conversion in this mode is simulated through the capture rate parameter.

[0088] The simulation of fishermen establishing advantages by independent search and disturbing the water area, with group encirclement as an auxiliary search method. If the capture situation at the current location is the same as the capture speed parameter, then local search is performed; otherwise, follow-up search is carried out, including:

[0089]

[0090] Among them, α is the capture rate parameter, CF is the current evaluation quantity, and MaxCF is the maximum evaluation value;

[0091]

[0092] Among them, Fishermen in groups of 3 or 4, C is the center point, and are the positions of the i-th fisherman in the j-th dimension in the group after the (T + 1)-th and T-th updates; r2 is the speed at which the fisherman approaches the center, and its value range is (0, 1); r3 is the offset of the movement, and its value range is (-1, 1).

[0093] In the exploitation stage, Gaussian distribution is used for simulation. The use of Gaussian distribution to simulate the update process includes:

[0094]

[0095] Among them, GD is the Gaussian distribution function, σ is the population variance, CF is the current evaluation quantity, MaxCF is the maximum evaluation value, Gbest is the global optimal position, r4 is a random number of 1, 2, or 3, and mean(Fisher) is the matrix of the mean values of each dimension of the fishermen center. is the position of the i-th fisherman at the (T + 1)-th update.

[0096] Figure 5 This is the schematic diagram of collective search provided by the present invention. Finally, by writing code in Python, the hyperparameters of the BP neural network can be optimized.

[0097] Optionally, before inputting the first variable parameter, the second variable parameter, the third variable parameter, and the current blasting parameter into the preset CFOA-BP-ANN model, the method further includes:

[0098] Using the preset fishing optimization algorithm CFOA to optimize the hyperparameters of the BP neural network to obtain the initial CFOA-BP-ANN model;

[0099] The hyperparameters include the learning rate η, the regularization parameter λ, the number of layers L of the neural network, the number of neurons j in each hidden layer, the number of learning epochs Epoch, the size of the mini-batch data Batch Size, the optimizer, the weight initialization method, and the type of neuron activation function. If defined manually, there is a certain subjectivity and it does not necessarily represent the optimal combination, which will seriously affect the training efficiency and accuracy of the BP neural network. Therefore, CFOA can well solve this problem and greatly improve the training efficiency of the model.

[0100] Optionally, the ranges of these hyperparameters are generally: learning rate η (0.0001 - 0.1), regularization parameter (0 - 0.1), number of layers L of the neural network (1 - 10), number of neurons j (10 - 1000), number of learning epochs Epoch (10 - 1000), size of the mini-batch data Batch Size (16 - 256), optimizer (SGD, Adam, RMSprop), weight initialization method (random initialization, Xavier initialization, He initialization).

[0101] The types of neuron activation functions are as follows:

[0102] Sigmoid:

[0103]

[0104] Tanh hyperbolic tangent:

[0105]

[0106] ReLU:

[0107]

[0108] Leaky ReLU:

[0109]

[0110] ELU:

[0111]

[0112] PreLU:

[0113]

[0114] Softmax:

[0115]

[0116] Where e refers to the natural constant, x is the independent variable input, a and Z are definable coefficients, and K refers to the number of iterations.

[0117] Figure 6 is the schematic diagram of the BP neural network provided by the present invention. The BP neural network is mainly divided into an input layer, a hidden layer, and an output layer. An activation function needs to be introduced when connecting neurons. Its main parameters are mainly weights and biases, and each connection has independent weights and biases. Taking Tanh as an example, each neuron can be expressed as:

[0118]

[0119] In the formula, w is the weight and b is the bias.

[0120] To prevent the neural network from overfitting, a regularization term needs to be added to modify the loss function of the neural network. Taking L2 regularization as an example, the modified loss function is:

[0121]

[0122] Where m is the number of epochs of learning, n is the total number of samples in one training, is the predicted value output by the model, y i represents the true value of the prediction target, η is the learning rate, and w represents the weight connecting to the previous layer of the neural network.

[0123] Finally, to establish the CFOA - BP - ANN model, the following steps are required: specify the upper and lower limit ranges of each hyperparameter, then set the optimization target dimension dim = 9 of the CFOA, take the neural network model as the objective function of the CFOA, and take the root mean square error RMSE of the prediction as the fitness value of the CFOA. RMSE is expressed as:

[0124]

[0125] Among them, is the predicted value output by the model, y i represents the true value of the prediction target.

[0126] Optionally, set the population number N of CFOA to be 100 - 500, the number of iterations MaxCF to be 100 - 500, and it has been verified that when N = 200 and the number of iterations MaxCF = 200, the efficiency and precision benefit values are the largest. Finally, in order to evaluate the model performance and prediction accuracy, the coefficient of determination R2 and the root mean square error RMSE are used to evaluate the model. Among them, RMSE is the same as above, and R2 is:

[0127]

[0128] Among them, is the predicted value output by the model, y i represents the true value of the prediction target.

[0129] Figure 7 is the R 2 comparison chart of different models provided by the present invention, Figure 8 is the RMSE comparison chart of different models provided by the present invention. In order to reflect the advantages of the model, this method is also compared with other existing models. The closer R 2 is to 1, the better the prediction effect, and the smaller the RMSE, the better the prediction effect. The comparison effect of R 2 and RMSE is as Figure 7 and Figure 8 shown.

[0130] Figure 9 is the schematic diagram of the monitoring points provided by the present invention. Due to limited instrument equipment and human resources, it is impossible to conduct vibration monitoring at all positions in the blasting area adjacent to the excavation of high-steep slopes during actual vibration monitoring. Therefore, the final idea is to predict the vibration velocity at any other position through the vibration monitoring data at some positions. First, it is necessary to monitor some positions and record the corresponding peak particle velocity and corresponding blasting parameters. This data needs to be used as the training set and test set of the CFOA - BP - ANN model. Assume that the schematic diagrams of the initially monitored measuring point positions and the planned monitoring point positions are as Figure 9 shown. By recording the data of the first 4 measuring points, as the training set and test set of the CFOA - BP - ANN model, and then using the coefficient of determination R 2 and the root mean square error RMSE to evaluate the model. When R 2When it is greater than 0.85 and the accuracy meets the engineering standards, the vibration prediction of any other measuring points can be realized. This not only saves the monitoring cost and resources, but also improves the efficiency and quality of monitoring.

[0131] Those skilled in the art understand that in an alternative embodiment, by using the measured peak particle velocity data, the present invention can also combine sensor technology and a data acquisition system to establish a real-time blasting vibration monitoring system. This system can continuously monitor the vibration conditions generated during the blasting process, record and analyze the data in real time, set reasonable vibration warning thresholds according to the structural characteristics and geological conditions of the surrounding structures, and when the monitored vibration data exceeds the warning threshold, the system can automatically trigger the warning mechanism to notify the on-site construction personnel and the management personnel of the surrounding facilities in a timely manner.

[0132] In another alternative embodiment, by using the peak particle velocity data and combining blasting parameters and geological conditions, an evaluation model of the impact of blasting vibration on the surrounding structures and geology is established. This model can predict the degree of vibration impact on the surrounding environment under different blasting schemes, and optimize and adjust the blasting scheme according to the prediction results of the evaluation model to reduce the vibration impact on the surrounding environment. For example, adjust parameters such as blasting charge, blasting sequence, and blasting time to reduce the vibration amplitude and duration.

[0133] In another alternative embodiment, by using preset automated monitoring technologies and data analysis methods, the monitoring efficiency and quality are improved. For example, by adopting unattended monitoring stations and remote data transmission technologies, remote real-time monitoring and analysis of blasting vibration are realized, and the monitoring data is visually displayed in the form of charts, curves, etc., which is convenient for construction personnel and management personnel to intuitively understand the blasting vibration situation. At the same time, a monitoring report is automatically generated to provide a basis for subsequent adjustment and optimization of the blasting scheme.

[0134] In another alternative embodiment, according to the monitoring results and the prediction results of the evaluation model, the on-site construction personnel are guided to carry out blasting operations. By adjusting the blasting parameters and monitoring strategies, while ensuring the construction efficiency, the vibration impact on the surrounding environment is reduced, and the safety management of the surrounding facilities is strengthened to ensure that the surrounding facilities are not damaged during the blasting operation. For example, reinforce important facilities and set up safety isolation belts, etc.

[0135] Figure 10It is a schematic structural diagram of the device for predicting the peak particle velocity in the vicinity of a high-steep slope provided by the present invention. The device for predicting the peak particle velocity in the vicinity of a high-steep slope includes: a determination unit 1. The determination unit is used to determine a first variable parameter according to the attenuation index and equivalent charge amount between the current measurement point and the current explosion source, determine a second variable parameter according to the attenuation index and explosion source distance between the current measurement point and the current explosion source, and determine a third variable parameter according to the preset coefficient between the current measurement point and the current explosion source and the peak particle velocity obtained at the current measurement point. The working principle of the determination unit 1 can refer to the aforementioned step 101 and will not be elaborated here.

[0136] The device for predicting the peak particle velocity in the vicinity of a high-steep slope further includes an input unit 2. The input unit is used to input the first variable parameter, the second variable parameter, the third variable parameter, and the current blasting parameter into a preset CFOA-BP-ANN model to obtain the predicted value of the peak particle velocity output by the preset CFOA-BP-ANN model. The working principle of the input unit 2 can refer to the aforementioned step 102 and will not be elaborated here.

[0137] The preset CFOA-BP-ANN model is determined after training the initial CFOA-BP-ANN model according to the sample first parameter, sample second parameter, sample third parameter, sample blasting parameter, and sample predicted value;

[0138] The sample first parameter, sample second parameter, and sample third parameter are determined after taking the logarithm of both sides of the Sadovsky empirical formula using the least squares method and performing equivalent transformation.

[0139] The present invention can realize the monitoring of the peak particle velocity of all protected objects, which not only saves resources but also improves the monitoring efficiency and quality of the blasting vibration of the protected objects. Moreover, it can also improve the blasting effect and reduce the blasting vibration by optimizing the blasting parameters, improve the on-site construction efficiency, and ensure the safety of surrounding facilities;

[0140] The present invention introduces a new meta-heuristic optimization algorithm - the new fish swarm optimization algorithm CFOA to optimize the hyperparameters of the BP neural network, thereby establishing a CFOA-BP-ANN prediction model, making the CFOA-BP-ANN prediction model have higher prediction accuracy. Using the CFOA-BP-ANN prediction model, the vibration prediction of any measurement point can be realized, which not only saves the monitoring cost and resources but also improves the monitoring efficiency and quality.

[0141] Figure 11 It is a schematic structural diagram of the electronic device provided by the present invention. As Figure 11As shown in the figure, the electronic device may include: a processor 110, a communications interface 120, a memory 130, and a communication bus 140. Among them, the processor 110, the communications interface 120, and the memory 130 complete communication with each other through the communication bus 140. The processor 110 may call the logical instructions in the memory 130 to execute the method for predicting the peak particle velocity in the vicinity of a high and steep slope. The method includes: determining a first variable parameter according to the attenuation index and the equivalent charge amount between the current measurement point and the current blast source, determining a second variable parameter according to the attenuation index and the blast source distance between the current measurement point and the current blast source, and determining a third variable parameter according to the preset coefficient between the current measurement point and the current blast source and the peak particle velocity obtained at the current measurement point; inputting the first variable parameter, the second variable parameter, the third variable parameter, and the current blasting parameters into a preset CFOA-BP-ANN model to obtain a predicted value of the peak particle velocity output by the preset CFOA-BP-ANN model; the preset CFOA-BP-ANN model is determined after training the initial CFOA-BP-ANN model according to the sample first parameter, the sample second parameter, the sample third parameter, the sample blasting parameters, and the sample predicted value; the sample first parameter, the sample second parameter, and the sample third parameter are determined after taking the logarithm of both sides of the Sadovskii empirical formula using the least squares method and performing equivalent transformation.

[0142] In addition, the logical instructions in the above-mentioned memory 130 may be implemented in the form of software functional units and, when sold or used as an independent product, may be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, may be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs that can store program codes.

[0143] On the other hand, the present invention also provides a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute a method for predicting the peak particle velocity in the vicinity of a high-steep slope provided by the above-mentioned various methods. The method includes: determining a first variable parameter according to the attenuation index and equivalent charge amount between the current measurement point and the current explosion source, determining a second variable parameter according to the attenuation index and explosion source distance between the current measurement point and the current explosion source, and determining a third variable parameter according to the preset coefficient between the current measurement point and the current explosion source and the peak particle velocity obtained at the current measurement point; inputting the first variable parameter, the second variable parameter, the third variable parameter and the current blasting parameter into a preset CFOA-BP-ANN model to obtain a predicted value of the peak particle velocity output by the preset CFOA-BP-ANN model; the preset CFOA-BP-ANN model is determined after training the initial CFOA-BP-ANN model according to the sample first parameter, sample second parameter, sample third parameter, sample blasting parameter and sample predicted value; the sample first parameter, sample second parameter and sample third parameter are determined after taking the logarithm of both sides of the Sadovskii empirical formula by using the least squares method and performing equivalent transformation.

[0144] On another aspect, the present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it realizes the method for predicting the peak particle velocity in the vicinity of a high-steep slope provided by the above-mentioned various methods. The method includes: determining a first variable parameter according to the attenuation index and equivalent charge amount between the current measurement point and the current explosion source, determining a second variable parameter according to the attenuation index and explosion source distance between the current measurement point and the current explosion source, and determining a third variable parameter according to the preset coefficient between the current measurement point and the current explosion source and the peak particle velocity obtained at the current measurement point; inputting the first variable parameter, the second variable parameter, the third variable parameter and the current blasting parameter into a preset CFOA-BP-ANN model to obtain a predicted value of the peak particle velocity output by the preset CFOA-BP-ANN model; the preset CFOA-BP-ANN model is determined after training the initial CFOA-BP-ANN model according to the sample first parameter, sample second parameter, sample third parameter, sample blasting parameter and sample predicted value; the sample first parameter, sample second parameter and sample third parameter are determined after taking the logarithm of both sides of the Sadovskii empirical formula by using the least squares method and performing equivalent transformation.

[0145] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art can understand and implement it without creative efforts.

[0146] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the essence of the above technical solution, or the part that contributes to the prior art, can be embodied in the form of a software product, which can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0147] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for predicting the peak particle velocity in the vicinity of a high-steep slope, characterized in that, Including: Determine the first variable parameter according to the attenuation exponent and equivalent charge amount between the current measurement point and the current explosion source, determine the second variable parameter according to the attenuation exponent and explosion source distance between the current measurement point and the current explosion source, and determine the third variable parameter according to the preset coefficient between the current measurement point and the current explosion source and the particle peak velocity obtained at the current measurement point; Input the first variable parameter, the second variable parameter, the third variable parameter and the current blasting parameter into the preset CFOA-BP-ANN model to obtain the predicted value of the particle peak velocity output by the preset CFOA-BP-ANN model; The preset CFOA-BP-ANN model is determined after training the initial CFOA-BP-ANN model according to the sample first parameter, sample second parameter, sample third parameter, sample blasting parameter and sample predicted value; The sample first parameter, sample second parameter and sample third parameter are determined after taking the logarithm of both sides of the Sadovsky empirical formula using the least squares method and performing equivalent transformation; 2. The method for predicting the peak particle velocity in the vicinity of a high and steep slope according to claim 1, wherein The current blasting parameters include hole diameter size, bench height, minimum burden size, toe burden size, hole spacing size, row spacing size, hole depth value, stemming depth value, charge length, stemming length, drill hole inclination, hole density coefficient, decoupling coefficient, charge per unit volume, charge per hole, maximum charge per delay section, total charge amount, hole-by-hole delay value, row-by-row delay value and total number of holes; 3. The method for predicting the peak particle velocity in the vicinity of a high-steep slope according to claim 1, wherein Taking the logarithm of both sides of the Sadovsky empirical formula using the least squares method and performing equivalent transformation includes: Among them, the first sample parameter is The second sample parameter is αln R, the third sample parameter is ln P PV - lnK, where K is a preset coefficient between the sample measurement point and the sample explosion source, Q is the sample equivalent charge, R is the distance from the sample explosion source, α is the attenuation index between the sample measurement point and the sample explosion source, and PPV is the peak particle velocity of the sample.

4. The method for predicting the peak particle velocity in the vicinity of a high-steep slope according to claim 1, characterized in that, Before inputting the first variable parameter, the second variable parameter, the third variable parameter and the current blasting parameter into the preset CFOA-BP-ANN model, the method further includes: Using the preset fish swarm optimization algorithm CFOA to perform hyperparameter optimization on the BP neural network to obtain the initial CFOA-BP-ANN model; The hyperparameters include learning rate η, regularization parameter λ, number of layers L of the neural network, number of neurons j in each hidden layer, number of learning epochs Epoch, size of the mini-batch data Batch Size, optimizer, weight initialization method and type of neuron activation function; 5. The method for predicting the peak particle velocity in the adjacent area of a high-steep slope according to claim 4, characterized in that The preset fish swarm optimization algorithm CFOA includes an exploration stage and a development stage; The using the preset fish swarm optimization algorithm CFOA to perform hyperparameter optimization on the BP neural network to obtain the initial CFOA-BP-ANN model includes: In the exploration stage, simulate fishermen to establish advantages through autonomous search and disturbing the water area by the capture rate parameter, use group envelopment as an auxiliary search method, if the capture situation at the current location is the same as the capture rate parameter, then perform local search, otherwise follow search; In the development stage, use the Gaussian distribution to simulate the update process; 6. The method for predicting the peak particle velocity in the vicinity of a high and steep slope according to claim 5, characterized in that, The simulating fishermen to establish advantages through autonomous search and disturbing the water area by the capture rate parameter, using group envelopment as an auxiliary search method, if the capture situation at the current location is the same as the capture rate parameter, then perform local search, otherwise follow search includes: Where α is the capture rate parameter, CF is the current evaluation quantity, and MaxCF is the evaluation maximum value; Among them, are fishermen in groups of 3 or 4, c is the center point, and are the positions of the i-th fisherman in the j-th dimension in the group after the (T + 1)-th and T-th updates; r2 is the speed at which the fisherman approaches the center, and its value range is (0, 1); r3 is the offset of the movement, and its value range is (-1, 1).

7. The method for predicting the peak particle velocity in the vicinity of a high and steep slope according to claim 5, wherein The use of Gaussian distribution to simulate the update process includes: The distribution function, σ is the population variance, CF is the current evaluation quantity, MaxCF is the maximum evaluation value, Gbest is the global optimal position, r4 is a random number of 1, 2, 3, mean(Fisher) is the matrix of the means of each dimension of the fishermen center, is the position of the i-th fisherman at the (T + 1)-th update.

8. The method for predicting the peak particle velocity in the vicinity of a high and steep slope according to claim 1, characterized in that, Before inputting the first variable parameter, the second variable parameter, the third variable parameter, and the current blasting parameter into the preset CFOA-BP-ANN model, the method further includes: Performing standardization processing on the first variable parameter, the second variable parameter, the third variable parameter, and the current blasting parameter to obtain the standardized first variable parameter, the second variable parameter, the third variable parameter, and the current blasting parameter.

9. A device for predicting the peak particle velocity in the vicinity of a high and steep slope, characterized in that, It includes: A determination unit, which is used to determine the first variable parameter according to the attenuation index and equivalent charge amount between the current measuring point and the current blasting source, determine the second variable parameter according to the attenuation index and the distance from the blasting source between the current measuring point and the current blasting source, and determine the third variable parameter according to the preset coefficient between the current measuring point and the current blasting source and the particle peak velocity obtained at the current measuring point; An input unit, which is used to input the first variable parameter, the second variable parameter, the third variable parameter, and the current blasting parameter into the preset CFOA-BP-ANN model to obtain the predicted value of the particle peak velocity output by the preset CFOA-BP-ANN model; The preset CFOA-BP-ANN model is determined after training the initial CFOA-BP-ANN model according to the sample first parameter, the sample second parameter, the sample third parameter, the sample blasting parameter, and the sample predicted value; The sample first parameter, the sample second parameter, and the sample third parameter are determined after taking the logarithm of both sides of the Sadovskii empirical formula using the least squares method and performing equivalent transformation.

10. An electronic device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the method for predicting the particle peak velocity in the vicinity of a high-steep slope as described in any one of claims 1 to 8.

Images