Underground powerhouse surrounding rock stability prediction method, system, equipment and medium

Abstract

The invention relates to the technical field of engineering construction, and provides an underground powerhouse surrounding rock stability prediction method, system, device and medium. Performing fractal dimension modeling on the time sequence monitoring data to obtain a detection fractal dimension function, and performing local search on a scaling factor, a correction coefficient and a scale factor in the detection fractal dimension function according to a hybrid leapfrog algorithm to obtain an optimal scaling factor, an optimal correction coefficient and an optimal scale factor; based on the optimal scaling factor, the optimal correction coefficient, the optimal scale factor and the detection fractal dimension function, constructing a target dimension feature extraction function, and extracting a target dimension feature value in the time sequence monitoring data according to the target dimension feature extraction function; and inputting the target dimension characteristic value and the surrounding rock sampling vector into a chaos neural network for surrounding rock deformation prediction, and obtaining a surrounding rock deformation prediction result. According to the invention, the timeliness and accuracy of surrounding rock stability monitoring and early warning can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of engineering construction, and particularly to a method, system, device and medium for predicting the stability of surrounding rocks in an underground power house. Background Art

[0002] In the field of geological exploration for water conservancy and hydropower projects, the assessment of surrounding rock stability has always been crucial in the design and construction processes. Traditional surrounding rock stability assessments mainly rely on geological exploration data, on-site test results, and engineers' empirical judgments; although these traditional methods have been widely used historically, their limitations have gradually emerged with the expansion of project scale and the increasing complexity of geological conditions.

[0003] Chinese Patent with publication number CN116258071A discloses a method for evaluating the effectiveness of surrounding rock monitoring indicators. The method includes: obtaining various monitoring indicator data of a certain section of a tunnel, dividing the surrounding rock deformation process into several damage stages according to the mutation of surrounding rock displacement; establishing a two-dimensional table for each section to store the section monitoring data; using one or more monitoring indicator columns in the two-dimensional table as data features and the damage stage column as data labels to input into a machine learning model, classifying the data features according to the data labels, obtaining a prediction accuracy matrix for surrounding rock damage determination, and obtaining the significance scores of each data feature to determine the effectiveness of the monitoring indicators. However, although the above solution uses numerical simulation to perform simulation operations on the surrounding rock deformation data of the surrounding rock displacement monitoring data outside the surrounding rock, due to the fact that the simple displacement outside the surrounding rock and the external deformation of the surrounding rock cannot determine the internal stress and deformation of the surrounding rock, the accuracy of surrounding rock stability monitoring is relatively low. Therefore, it is very necessary to provide a method and system for predicting the stability of surrounding rocks in an underground power house to improve the timeliness and accuracy of surrounding rock stability monitoring and early warning. Summary of the Invention

[0004] The purpose of the present invention is to overcome the above-mentioned defects and problems existing in the prior art, and provide a method, system, device and medium for predicting the stability of surrounding rocks in an underground power house. By performing acoustic emission monitoring and sampling vector recording on the surrounding rocks of the underground power house, multi-angle data collection of the surrounding rock state is realized, and the subtle changes of the surrounding rocks are captured by using time series data, so as to improve the timeliness and accuracy of surrounding rock stability monitoring and early warning.

[0005] In a first aspect, the present invention provides a method for predicting the stability of surrounding rocks in an underground power house, including:

[0006] Collecting time series monitoring data of the surrounding rocks of the underground power house, wherein the time series monitoring data includes surrounding rock acoustic emission signal data and surrounding rock sampling vectors;

[0007] Perform fractal dimension modeling on the timing monitoring data to obtain a detection fractal dimension function, and perform local search on the scaling factor, correction coefficient, and scale factor in the detection fractal dimension function according to the hybrid leapfrog algorithm to obtain the optimal scaling factor, optimal correction coefficient, and optimal scale factor;

[0008] Based on the optimal scaling factor, the optimal correction coefficient, the optimal scale factor, and the detection fractal dimension function, construct a target dimension feature extraction function, and extract the target dimension feature value in the timing monitoring data according to the target dimension feature extraction function;

[0009] Input the target dimension feature value and the surrounding rock sampling vector into a chaotic neural network for surrounding rock deformation prediction to obtain a surrounding rock deformation prediction result.

[0010] On the basis of the above technical solutions, preferably, the acquisition of the timing monitoring data of the surrounding rock of the underground powerhouse specifically includes:

[0011] Obtain the surrounding rock monitoring data, surrounding rock acoustic emission signal data corresponding to multiple measurement points within a preset interval distance range in the underground space corresponding to the underground powerhouse, and the supporting structure force monitoring data corresponding to multiple measurement points, integrate all the supporting structure force monitoring data and all the surrounding rock monitoring data, and construct multiple surrounding rock sampling vectors corresponding to each measurement point;

[0012] When collecting the surrounding rock monitoring data and the supporting structure force monitoring data, each corresponding measurement point simultaneously monitors the horizontal and vertical displacements of the surrounding rock, and simultaneously monitors the radial and tangential surrounding rock stresses of the surrounding rock.

[0013] Even more preferably, the performing fractal dimension modeling on the timing monitoring data to obtain a detection fractal dimension function specifically includes:

[0014] Segment the surrounding rock acoustic emission signal data on the time axis to obtain multiple segments of surrounding rock acoustic emission sequences, and perform amplitude normalization processing on the surrounding rock acoustic emission sequences to obtain a standard surrounding rock acoustic emission sequence;

[0015] Construct an initial detection fractal dimension function according to the standard surrounding rock acoustic emission sequence;

[0016] Calculate the local fractal values at different sampling scales through the initial detection fractal dimension function, and obtain the sampling logarithmic relationship between the sampling scale and the corresponding local fractal values according to linear fitting;

[0017] Based on the sampling logarithmic relationship and the initial detection fractal dimension function, construct a detection fractal dimension function.

[0018] More preferably, the local search for the scaling factor, correction coefficient, and scale factor in the detection fractal dimension function according to the hybrid leapfrog algorithm specifically includes:

[0019] Randomly generate multiple frog individuals and initialize them within the feasible region of the frog individuals, where each frog individual represents an optimization parameter combination including a scaling factor, a correction coefficient, and a scale factor;

[0020] Based on the detection fractal dimension function and the fractal reference value, construct a fractal reference function, and calculate the fitness of each frog individual according to the fractal reference function and the fractal reference value;

[0021] Sort the frog individuals in the population in ascending order of fitness and divide the population into multiple sub-cultural gene bodies, where each cultural gene body contains several frog individuals;

[0022] In each sub-cultural gene body, select the frog individual with the best fitness and the worst frog individual respectively, and update the position of the worst frog individual based on the position update function to perform local search;

[0023] When at least one round of local search has been performed in each cultural gene body and the maximum number of iterations is reached, recombine the population and sort it again in ascending order of fitness, update the best frog individual in the population, and record the position of the globally best frog individual.

[0024] More preferably, the input of the target dimension eigenvalue and the surrounding rock sampling vector into the chaotic neural network for surrounding rock deformation prediction to obtain the surrounding rock deformation prediction result specifically includes:

[0025] Perform normalization processing on the target dimension eigenvalue and the surrounding rock sampling vector to obtain a standard target dimension eigenvalue and a standard surrounding rock sampling vector;

[0026] Perform vector splicing on the standard target dimension eigenvalue and the standard surrounding rock sampling vector to obtain a surrounding rock feature vector, and input the surrounding rock feature vector into the chaotic neural network;

[0027] Based on the internal state function and output state function of the chaotic neural network, obtain the surrounding rock deformation prediction result.

[0028] More preferably, the expression of the target dimension feature extraction function is:

[0029]

[0030] d WS = T(δ; α * , β * , m * );

[0031] Among them, T( ) represents the detection fractal dimension function; δ represents the sampling scale; x n represents the amplitude of the nth sampling point after the surrounding rock acoustic emission signal data is segmented; N represents the minimum number of scale numbers required to completely cover the signal at the sampling scale δ; α represents the scaling factor corresponding to the cumulative length of the signal; β represents the correction coefficient corresponding to the mapping value of the signal change and the scale relationship; m represents the scale factor corresponding to the scale window used in the fractal dimension estimation; MSE represents the mean square error; K represents the total number of surrounding rock acoustic emission sequences in the surrounding rock acoustic emission signal data; T k represents the fractal dimension value under the kth segment of the surrounding rock acoustic emission sequence; represents the fractal reference value under the kth segment of the surrounding rock acoustic emission sequence; d WS represents the target dimension feature extraction function; α * represents the optimal scaling factor corresponding to the cumulative length of the signal; β * represents the optimal correction coefficient corresponding to the mapping value of the signal change and the scale relationship; m * represents the optimal scale factor corresponding to the scale window used in the fractal dimension estimation.

[0032] More preferably, the expression of the chaotic neural network is:

[0033] F(t) = [d WS (t), σ(t), ε(t), P(t),....];

[0034]

[0035] yi(t + 1) = f(x i (t + 1));

[0036] Among them, F(t) represents the surrounding rock feature vector input into the chaotic neural network at the tth moment; d WS (t) represents the target dimension feature value at the tth moment; ε(t) represents the strain value of the surrounding rock at the tth moment; σ(t) represents the compressive strength of the rock mass material monitored at the tth moment; P(t) represents the real-time surrounding rock pressure monitored at the tth moment; x i (t) represents the internal state function of the ith neuron at the tth moment; g( ) represents the folding-back mapping function; J represents the total number of quantization neurons; w ij represents the contribution connection weight of the jth quantization neuron to the state update of the ith neuron; H represents the dimension number of the surrounding rock sampling vector; v ih represents the contribution connection weight of the surrounding rock feature under the hth dimension to the state update of the ith neuron; F h (t) represents the surrounding rock feature under the hth dimension input into the chaotic neural network at the tth moment; θi It represents the bias parameter of the i-th neuron; c represents the perturbation intensity parameter; z(t) represents the chaotic sequence generated by the chaotic map; y i (t + 1) represents the output state function of the i-th neuron at the moment of t + 1; f( ) represents the non-linear activation function.

[0037] In a second aspect, the present invention also provides a prediction system for the stability of surrounding rock in an underground powerhouse. This system is applied to the method described above, and the system includes:

[0038] A time-series monitoring data acquisition module, which is used to acquire the time-series monitoring data of the surrounding rock of the underground powerhouse. Among them, the time-series monitoring data includes the acoustic emission signal data of the surrounding rock and the sampling vector of the surrounding rock;

[0039] A target dimension eigenvalue acquisition module, which is used to perform fractal dimension modeling on the time-series monitoring data to obtain a detection fractal dimension function, and perform local search on the scaling factor, correction coefficient, and scale factor in the detection fractal dimension function according to the hybrid leapfrog algorithm to obtain the optimal scaling factor, optimal correction coefficient, and optimal scale factor; based on the optimal scaling factor, the optimal correction coefficient, the optimal scale factor, and the detection fractal dimension function, construct a target dimension feature extraction function, and extract the target dimension eigenvalues in the time-series monitoring data according to the target dimension feature extraction function;

[0040] A surrounding rock deformation prediction module, which is used to input the target dimension eigenvalues and the sampling vector of the surrounding rock into a chaotic neural network for surrounding rock deformation prediction to obtain a surrounding rock deformation prediction result.

[0041] In a third aspect, the present invention also provides a prediction device for the stability of surrounding rock in an underground powerhouse, including a memory and a processor;

[0042] The memory is used to store computer program codes and transmit the computer program codes to the processor;

[0043] The processor is used to execute the method described above according to the instructions in the computer program codes.

[0044] In a fourth aspect, the present invention also provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the method described above is implemented.

[0045] Compared with the prior art, the beneficial effects of the present invention are:

[0046] 1. In a method, system, device and medium for predicting the stability of surrounding rock in an underground powerhouse of the present invention, by performing acoustic emission monitoring and sampling vector recording on the surrounding rock of the underground powerhouse, multi-angle data collection of the surrounding rock state is achieved, ensuring that the monitoring data has higher information content and integrity. Moreover, by using time-series data to capture subtle changes in the surrounding rock, the timeliness and accuracy of the monitoring and early warning of the stability of the surrounding rock are improved. The fractal dimension model is used to model the monitoring data, which can reveal the complex non-linear behavior during the deformation process of the surrounding rock. By detecting the fractal dimension function to capture the inherent structural characteristics of the data, and using the hybrid leapfrog algorithm for local search, for the adjustment of the scaling factor, correction coefficient and scale factor, an optimal parameter combination can be obtained, significantly improving the accuracy and robustness of the fractal model in capturing the deformation characteristics of the surrounding rock. At the same time, by using the chaotic neural network to input and process the target features and sampling vectors, the randomness, non-linearity and complexity problems existing in the system can be better handled, realizing the efficient prediction of the deformation process of the surrounding rock.

[0047] 2. In a method, system, device and medium for predicting the stability of surrounding rock in an underground powerhouse of the present invention, by presetting an interval distance in the underground space, the surrounding rock monitoring data, the acoustic emission signal data of the surrounding rock and the stress monitoring data of the support structure are collected simultaneously at multiple measuring points, ensuring the comprehensive coverage of the monitoring information and the integrity of the spatial distribution, which helps to capture the state changes of the surrounding rock and the support structure at different positions. Moreover, monitoring the displacement and stress state of the surrounding rock can completely describe the stress state and change trend of the surrounding rock and the support structure from the two perspectives of deformation and stress. By integrating all the stress monitoring data of the support structure and the surrounding rock monitoring data, the multi-source data of each measuring point can be fully utilized to realize the cross-validation of information and enhance the effectiveness of the signal. At the same time, multiple surrounding rock sampling vectors are constructed for each measuring point, enabling subsequent data analysis, feature extraction and model construction to better reflect the local surrounding rock characteristics of each measuring point, enhancing the ability of the model to perceive local abnormal changes. Using the surrounding rock displacement and stress data, not only can the minute changes of the surrounding rock in different directions be obtained in a timely manner, but also multi-dimensional input can be provided for the hybrid data model, significantly improving the accuracy of the stability prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0049] Figure 1 It is a flowchart of a method for predicting the stability of surrounding rock in an underground powerhouse provided by the present invention.

[0050] Figure 2 It is a fitting curve graph of the size of acoustic emission sources and the released energy in surrounding rocks provided by the present invention.

[0051] Figure 3 It is a structural block diagram of a prediction system for the stability of surrounding rocks of an underground powerhouse provided by the present invention.

[0052] Figure 4 It is a structural block diagram of a prediction device for the stability of surrounding rocks of an underground powerhouse provided by the present invention. Specific embodiments

[0053] Next, in combination with the embodiments of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0054] See Figure 1 , the present invention provides a method for predicting the stability of surrounding rocks of an underground powerhouse, and the steps of the method include S1 to S4.

[0055] Step S1, collect the time-series monitoring data of the surrounding rocks of the underground powerhouse, where the time-series monitoring data includes the acoustic emission signal data of the surrounding rocks and the sampling vectors of the surrounding rocks.

[0056] In this embodiment, acoustic emission sensors, displacement sensors, strain sensors, etc. can be used for time-series monitoring data. The acoustic emission sensors can be high-sensitivity piezoelectric sensors or capacitive sensors to capture tiny acoustic wave signals to effectively cover the high-frequency signals generated by the common propagation of rock mass microcracks. The displacement sensors and strain sensors can also be equipped with strain gauges, displacement gauges, etc. according to needs to record physical quantities such as the deformation and stress changes of the surrounding rocks.

[0057] The layout of key areas can be based on the geological structure and engineering risk points of the surrounding rocks of the underground powerhouse, and key areas that may undergo fission, damage or deformation are selected for layout. For example, parts prone to cracks and stress concentration should be preferentially detected. In order to improve the representativeness of the data, a sensor network or array layout method is usually adopted. In this way, not only can data from multiple angles and directions be obtained, but also the overall behavior of the surrounding rocks can be evaluated through spatial data correlation analysis. Connect all sensors to the data acquisition device through a standardized interface. A unified time reference (such as GPS synchronization or network clock synchronization) must be adopted between the sensors to ensure the time consistency of multi-sensor data, which is beneficial to subsequent data correlation analysis.

[0058] Furthermore, acoustic emission signals usually contain high-frequency components, so it is necessary to set a sufficiently high sampling frequency (for example, several kHz to tens of kHz) to capture signal details. The specific value needs to be determined according to the signal bandwidth and the expected monitoring target. For low-frequency signals such as displacement and strain, selecting a suitable sampling frequency according to the rate of change can ensure data continuity without generating redundant data. The relevant sensors and data acquisition systems should have high resolution (such as 16 bits or even higher) to capture subtle changes in surrounding rock deformation and avoid data quantification losses. For acoustic emission signals, a trigger threshold needs to be preset to ensure that data is recorded only when the signal intensity exceeds the ambient noise level, thereby improving data quality and saving storage space. During the initial setting, trial mining can be carried out according to the site noise level, and a reasonable threshold can be determined before it is officially enabled. A mode combining continuous sampling and event triggering can be set. When an abnormal signal exceeding the threshold is detected, the data in the period is automatically saved; at the same time, low-frequency sampling is also performed during the low-risk period for subsequent comparison reference. Ensuring that all sensors share the same time base is crucial for multi-dimensional analysis of time series data. GPS synchronization, NTP (Network Time Protocol) or local clock calibration can be used to prevent data alignment problems caused by clock drift.

[0059] In one example, the surrounding rock monitoring data, surrounding rock acoustic emission signal data and support structure stress monitoring data corresponding to multiple measuring points within a preset interval range in the underground space corresponding to the underground powerhouse are obtained, all support structure stress monitoring data and all surrounding rock monitoring data are integrated, and multiple surrounding rock sampling vectors corresponding to each measuring point are constructed. When collecting surrounding rock monitoring data and support structure stress monitoring data, each corresponding measuring point simultaneously monitors the horizontal and vertical displacements of the surrounding rock, and simultaneously monitors the radial and tangential surrounding rock stresses of the surrounding rock.

[0060] According to the geological profile, fault direction and joint distribution of the surrounding rock of the underground powerhouse, areas with possible weak surfaces and concentrated cracks are selected as key monitoring areas. Measuring points should be arranged near key parts such as supporting structures, roof plates, and arch columns, as they have a greater risk of concentrated force and displacement.

[0061] For relatively homogeneous geological regions, evenly spaced points can be used for layout. For example, a measurement point can be set every 5 - 10 meters to ensure that data collection covers the entire area. In potentially dangerous areas or where support structures are concentratedly arranged, the spacing can be reduced to 2 - 3 meters to ensure that finer changes can be captured. On the same horizontal plane, multiple positions are selected for monitoring according to the size of the surrounding rock, construction design, and construction technology requirements. For surrounding rock with obvious stratification or depth changes, multiple vertical positions can be selected for monitoring to obtain response data at different depths. A grid is constructed on the horizontal plane, and a measurement point is set at the center of each grid. The grid size can be designed as 5m×5m or more dense according to the site size and monitoring requirements. According to the volumetric distribution and structural characteristics of the surrounding rock, data is collected by superimposing vertical stratification on the horizontal plane. For example, measurement points are arranged on the roof, arch crown, and floor of an underground powerhouse respectively to achieve three-dimensional displacement and stress monitoring.

[0062] In this embodiment, by presetting an interval distance in the underground space, surrounding rock monitoring data, surrounding rock acoustic emission signal data, and support structure stress monitoring data are collected simultaneously at multiple measurement points, ensuring comprehensive coverage and spatial distribution integrity of monitoring information, which helps to capture the state changes of the surrounding rock and support structure at different positions. Moreover, monitoring the displacement and stress state of the surrounding rock can completely describe the stress state and change trend of the surrounding rock and support structure from two perspectives of deformation and stress. By integrating all support structure stress monitoring data and surrounding rock monitoring data, the multi-source data of each measurement point can be fully utilized to achieve cross-verification of information and enhance the effectiveness of the signal. At the same time, multiple surrounding rock sampling vectors are constructed for each measurement point, enabling subsequent data analysis, feature extraction, and model construction to better reflect the local surrounding rock characteristics of each measurement point, improving the model's ability to perceive local abnormal changes. Using the surrounding rock displacement and stress data, not only can the minute changes of the surrounding rock in different directions be obtained in a timely manner, but also multi-dimensional input can be provided for the hybrid data model, significantly improving the accuracy of stability prediction.

[0063] Furthermore, the displacement sensor adopts a high-precision digital displacement meter or a fiber optic sensor. Generally, it is recommended to select a device with a resolution at the 0.01 mm level or higher. Two sensors are installed at each measurement point to collect lateral (horizontal) and vertical displacement data respectively. Sufficient bonding area should be reserved during installation to ensure the close coupling between the sensor and the surrounding rock or support structure. For the monitoring of the surrounding rock stress, pressure gauges or strain gauge sensors are often selected, which are divided into radial and tangential stress measurements. The stress sensor should have a high linear response and repeatability accuracy under the peak stress. According to the geographical location of the measurement point, attention should be paid to the direction calibration when fixing the sensor to ensure that its measurement direction is consistent with the actual stress direction of the surrounding rock. Acoustic emission sensor: A piezoelectric high-sensitivity acoustic emission sensor is adopted, which has a wide frequency response range (usually above 10 kHz to several hundred kHz), can capture fast transient signals, and is fixed in areas prone to microdamage and crack propagation. A special fixing bracket and high-quality coupling agent are used to improve the signal transmission efficiency. A multi-channel high-speed acquisition system is configured to be able to receive signals from multiple sensors simultaneously. For displacement and stress monitoring, the sampling frequency can be set between 1 and 10 Hz; for acoustic emission monitoring, the sampling frequency is 10 kHz or even higher. The module needs to be equipped with a high-resolution data converter (such as a 16-bit or 24-bit ADC) to ensure that the accuracy of the collected data meets the monitoring requirements. A unified time synchronization scheme, such as GPS clock, NTP network clock synchronization or a dedicated synchronization module, is adopted to ensure that all data records have a unified timestamp. The on-site local area network (WLAN or wired network) is used to realize the real-time upload of data to ensure that the data can be centrally managed in the monitoring center. The data acquisition system is designed for anti-electromagnetic interference, and shielded cables and filters are used to ensure that high-quality data can still be collected in the low-interference and complex environment of the underground powerhouse.

[0064] In this embodiment, the preset measuring point scheme not only ensures wide-area monitoring in homogeneous geological regions but also enables high-density point layout in key areas prone to weakening or deformation (such as faults, concentrated fracture zones, and key parts of the support structure). In this way, both the overall surrounding rock state can be grasped and local subtle changes can be captured. By constructing a grid in the horizontal plane and laying points in layers vertically, multi-dimensional data acquisition of the underground space is achieved, which helps to accurately restore the deformation and stress conditions of the surrounding rock at different depths and in different directions, providing comprehensive information for subsequent evaluation. High-precision displacement sensors and stress sensors are used to effectively capture the small displacements of the surrounding rock in the horizontal and vertical directions and the stress changes in the radial and tangential directions. The sampled data not only has a high time resolution but also a high spatial resolution, ensuring that the data reflects the true state of the surrounding rock. All the stress monitoring data of the support structure and all the surrounding rock monitoring data are integrated, and through time-space data alignment, signal denoising, and fusion processing, a unified data format is formed, providing stable and high-quality data samples for model input. For each measuring point, a sampling vector is constructed based on multi-dimensional data such as displacement, stress, and acoustic emission collected, enabling subsequent data-driven analysis (such as fractal feature extraction, intelligent prediction, etc.) to more accurately reflect the actual engineering state and improve the robustness and accuracy of the prediction model.

[0065] As Figure 2 shown, Figure 2 Figure shows the fitting curve of the acoustic emission source size D and the released energy En in the surrounding rock, divided into two subfigures (a) and (b), corresponding to different fitting situations respectively. The horizontal axis D represents the size of the acoustic emission source, ranging from 1.1 to 1.9. The vertical axis En represents the energy released by the acoustic emission, normalized to [0,1]. The figure contains two curves, corresponding to the first group of samples (dashed line) and the second group of samples (solid line) respectively.

[0066] In Figure 2 subfigure (a), as the acoustic emission source size D increases, the released energy En shows a trend of rapid decline and then levels off. There are certain differences between the curves of the first group of samples and the second group of samples when D is small (about 1.1 to 1.3), but they gradually tend to be the same after D > 1.5, indicating that the energy release characteristics of the two groups of samples are similar in the range of larger sizes. In Figure 2Among them, (b) also shows the relationship between D and En, but the curve descends faster. The difference between the first group of samples and the second group of samples is relatively large when D is small (about 1.1 to 1.3), and the energy release of the first group of samples is slightly higher than that of the second group of samples. As D increases, the two curves gradually approach after D > 1.4 and finally tend to be consistent. Obviously, there is a non-linear relationship between the acoustic emission source size and the released energy En, and the energy release decreases rapidly as the size increases and tends to be stable within a large size range. There are certain differences in the energy release characteristics of the first group of samples and the second group of samples within the small size range, but they show consistency within the large size range, reflecting the acoustic emission characteristics of the surrounding rock under different conditions.

[0067] Step S2: Perform fractal dimension modeling on the time-series monitoring data to obtain the detection fractal dimension function, and perform local search on the scaling factor, correction coefficient, and scale factor in the detection fractal dimension function according to the hybrid leapfrog algorithm to obtain the optimal scaling factor, optimal correction coefficient, and optimal scale factor.

[0068] In this embodiment, the acoustic emission signal data of the surrounding rock is segmented on the time axis to obtain multiple segments of acoustic emission sequences of the surrounding rock, and the acoustic emission sequences of the surrounding rock are subjected to amplitude normalization processing to obtain the standard acoustic emission sequences of the surrounding rock; according to the standard acoustic emission sequences of the surrounding rock, an initial detection fractal dimension function is constructed; the local fractal values at different sampling scales are calculated through the initial detection fractal dimension function, and the sampling logarithmic relationship between the sampling scale and the corresponding local fractal values is obtained through linear fitting; based on the sampling logarithmic relationship and the initial detection fractal dimension function, a detection fractal dimension function is constructed.

[0069] Set one or more amplitude thresholds according to the actual monitoring data and signal noise level. Generally, when the signal amplitude exceeds the preset threshold, it marks the start of an acoustic emission event, and when the signal drops to near the background noise level, it marks the end of the event. Scan the entire time series. When it is detected that the signal rises from below the threshold to above the threshold, record this moment as the starting point of the acoustic emission event, and then find the point where the signal drops from the high amplitude state to below the threshold or remains below a certain threshold for a certain period of time, and record this moment as the end of the event. For two events with close time and possibly consecutive occurrences, a minimum event interval time can be preset to avoid wrongly splitting a single continuous event; at the same time, set a minimum duration to filter out short noise signals to ensure that the captured events are all valid. Or divide the entire monitoring time into fixed periods (such as every 1 second or every few seconds), and perform energy calculation or other statistical analysis on the signals within each window. The window length can be selected according to the duration of the acoustic emission event and the signal sampling frequency; calculate information such as power, mean, and standard deviation within each window, and combine the threshold to determine whether the window contains an acoustic emission event. If the conditions are met, the signal within the window can be used as an event segment, otherwise it is discarded or merged with adjacent windows. After extracting each segment, multiple acoustic emission sequences are formed; perform normalization processing on each segment sequence so that the amplitudes of different events fall within the same standard range, which is convenient for subsequent fractal dimension modeling. Normalization usually includes mapping the amplitude to the interval of 0-1 or performing standardization processing according to statistical quantities (such as mean and standard deviation). Each acoustic emission sequence needs to be accompanied by its start and end times, amplitude information, and the position of the window in the entire collected data, which is convenient for subsequent time series window reconstruction and feature extraction.

[0070] In this embodiment, the expression of the target dimension feature extraction function is:

[0071]

[0072] d WS = T(δ; α * , β * , m * );

[0073] Wherein, T() represents the detection fractal dimension function; δ represents the sampling scale, which is an integer multiple of the minimum sampling interval; x n represents the amplitude of the nth sampling point after the surrounding rock acoustic emission signal data is segmented; N represents the minimum number of scales required to completely cover the signal at the sampling scale δ; α represents the scaling factor corresponding to the cumulative length of the signal; β represents the correction coefficient corresponding to the mapping value of the signal change and scale relationship; m represents the scale factor corresponding to the scale window adopted in the fractal dimension estimation; MSE represents the mean square error; K represents the total number of surrounding rock acoustic emission sequences in the surrounding rock acoustic emission signal data; T kIt represents the fractal dimension value under the acoustic emission sequence of the k-th section of surrounding rock; It represents the fractal reference value under the acoustic emission sequence of the k-th section of surrounding rock; d WS It represents the target dimension feature extraction function; α * It represents the optimal scaling factor corresponding to the signal cumulative length; β * It represents the optimal correction coefficient corresponding to the mapping value of the signal change and scale relationship; m * It represents the optimal scale factor corresponding to the scale window adopted in the fractal dimension estimation.

[0074] For the standard acoustic emission sequence x of the surrounding rock n , calculate the signal increment |x n+δ - x n | at the sampling scale δ, multiply the signal increment by the scaling factor α, and perform a power adjustment β on the signal increment |x n+δ - x n |, sum over all sampling points n to obtain the cumulative value Perform a logarithmic transformation on the cumulative value and the scale factor, and calculate the numerator and denominator respectively: Divide the numerator by the denominator to obtain the initial detection fractal dimension function.

[0075] It can be understood that since T(δ; α, β, m) is a fractal dimension function used to describe the fractal characteristics of the acoustic emission signal data of the surrounding rock at different sampling scales δ, it calculates the fractal dimension by performing a logarithmic transformation on the cumulative value of the signal increment |x n+δ - x n | and combining the logarithmic relationship of the sampling scale. Optimize the parameters α, β, and m by minimizing the mean square error (MSE) to minimize the error between the fitted fractal dimension function T(δ; α, β, m) and the fractal reference value .

[0076] In this step, steps S21 to S25 are also included.

[0077] Step S21: Randomly generate multiple frog individuals and initialize them within the feasible region of the frog individuals. Among them, each frog individual represents an optimized parameter combination including a scaling factor, a correction coefficient, and a scale factor.

[0078] In this step, each frog individual represents an optimized parameter combination including a scaling factor α, a correction coefficient β, and a scale factor m. The scaling factor α can be defined as a positive number, and its usual value range is [α min , α max , for example, [0.1, 10]; the correction coefficient β is defined as a positive number, and its usual value range is [β min , β max, for example, [0.5, 2]; define the scaling factor m as a positive number, and its general value range is [m min , m max , for example, [1, 5]. Set the size of the frog population to q (i.e., generate q frog individuals). For each frog individual, randomly generate its parameter combination (α, β, m). The scaling factor α randomly generates a value within the range of [α min , α max . That is, α = α min +(α min -α min )rand(0, 1). Similarly, the correction coefficient β and the scaling factor m can be randomly generated in the same way.

[0079] The position of each frog individual is represented by its parameter combination (α, β, m). Take the generated (α, β, m) as the initial position of the frog individual and store it in the population. Repeat the above random generation process q times to obtain a population containing q frog individuals, that is, population = {(α1, β1, m1), (α2, β2, m2), …, (α q , β q , m a )}. Ensure that the parameters (α, β, m) of each frog individual are within the feasible region. If a certain parameter exceeds the range, it can be regenerated or truncated to the boundary value. For example, if α > α max , then let α = α max .

[0080] Step S22: Based on the detection fractal dimension function and the fractal reference value, construct a fractal reference function, and calculate the fitness of each frog individual according to the fractal reference function and the fractal reference value.

[0081] In this step, calculate the fitness of each frog individual through the fractal reference function .

[0082] Step S23: Sort the frog individuals in the population in ascending order of fitness, and divide the population into multiple sub-cultural gene bodies, where each cultural gene body contains several frog individuals.

[0083] In this step, set the number M of sub-cultural gene bodies. Usually, M is a factor of the population size q. The number of frog individuals contained in each sub-cultural gene body is q / M. If q cannot be divided evenly by M, the remaining individuals can be assigned to the first few gene bodies.

[0084] According to the sorted order of frog individuals, the population is sequentially assigned to M sub-culture genomes. For example: Sub-culture genome 1 contains the 1st to the q / M-th sorted frog individuals; Sub-culture genome 2 contains the (q / M + 1)-th to the 2q / M-th sorted frog individuals. And so on until all frog individuals are assigned.

[0085] Step S24: In each sub-culture genome, select the frog individual with the best fitness and the worst fitness respectively, and update the position of the worst frog individual based on the position update function to perform local search.

[0086] In this step, each sub-culture genome has been sorted in descending order of fitness. The frog individual with the best fitness represents the individual with the smallest fitness in the sub-culture genome, denoted as Best, and its position can be expressed as BestPosition = (α best , β best , m best ). The frog individual with the worst fitness represents the individual with the largest fitness in the sub-culture genome, denoted as Worst, and its position can be expressed as WorstPosition = (α worst , β worst , m worst ). The position update formula can be expressed as NewPosition = WorstPosition + r(BestPosition - WorstPosition), where r is a random number, usually uniformly distributed in the range [0, 1], and is used to control the update step size.

[0087] Step S25: When at least one round of local search has been performed in each culture genome and the maximum number of iterations is reached, recombine the population, sort it again in ascending order of fitness, update the frog individual with the best fitness in the population, and record the position of the globally best frog individual.

[0088] In this embodiment, by randomly generating multiple frog individuals and initializing them within the feasible region, the diversity of the population and the global search ability are ensured, providing a good starting point for subsequent optimization; calculating the fitness based on the detection fractal dimension function and the fractal reference value can accurately evaluate the quality of individuals and improve the optimization accuracy; through fitness increasing sorting and dividing the population into multiple sub-cultural gene bodies, hierarchical optimization and parallel search are realized, which not only maintain the population diversity but also improve the combination efficiency of global and local searches. In local search, based on the position update function, the worst frog individual is optimized to make it approach the optimal individual. At the same time, a random step size and a boundary check mechanism are used to avoid falling into local optima, and a global search supplement mechanism is combined to enhance the robustness of the algorithm. Finally, by recombining the population and updating the global optimum, the global optimum solution is gradually approximated, the population structure is dynamically adjusted, and the local and global search capabilities are balanced. The overall technical effect is the efficient combination of global search and local search, the dynamic balance of diversity and convergence, strong adaptability, high optimization efficiency, and strong robustness, which is applicable to complex non-linear, multi-peak, high-dimensional, and dynamic optimization problems, and can quickly find high-quality solutions and gradually approach the global optimum solution.

[0089] Step S3: Based on the optimal scaling factor, the optimal correction coefficient, the optimal scale factor, and the detection fractal dimension function, construct an objective dimension feature extraction function, and extract the objective dimension feature values in the time series monitoring data according to the objective dimension feature extraction function.

[0090] In this step, based on the detection fractal dimension function T(δ; α, β, m) and the optimized parameters, the function is written as d WS = T(δ; α * , β * , m * ), where δ represents the sampling scale, and the parameters in the function can better reflect the fractal characteristics or complexity of the time series data signal after optimization.

[0091] Furthermore, the objective dimension feature extraction function d WS fuses the parameters α * , β * and m * and the detection fractal dimension function can be further constructed into a function of the following form: d WS (δ) = α * Δln(|x n+δ - x n |) + β * × δ m *. Since time series monitoring data often has multi-scale characteristics, that is, the data may exhibit different variation laws at different scales. To extract effective fractal features, it is first necessary to preprocess the original data, such as calculating the local difference or local fluctuation of the data at different sampling scales or time delays. In this way, a quantity describing the local change of the data will be generated at each scale, and thus multiple sets of d WS (δ) values can be obtained on the same data set. For each sampling scale δ, the corresponding d WS (δ) value is obtained through the above target dimension feature extraction function. If a unified target dimension feature needs to be obtained, the d WS (δ) values at multiple scales can be statistically summarized. For example, calculate the average value of d WS (δ) at different scales as a global feature description.

[0092] After processing the data obtained in the above steps, one or more target dimension feature values are output. These values reflect the fractal properties of the time series monitoring data at multiple scales and the complexity of the signal, providing a quantitative basis for subsequent data monitoring, anomaly detection, prediction and other tasks.

[0093] Step S4, input the target dimension feature value and the surrounding rock sampling vector into the chaotic neural network for surrounding rock deformation prediction to obtain the surrounding rock deformation prediction result.

[0094] In this step, steps S41 to S43 are also included.

[0095] Step S41, normalize the target dimension feature value and the surrounding rock sampling vector to obtain the standard target dimension feature value and the standard surrounding rock sampling vector.

[0096] In this step, calculate the minimum value, maximum value, mean value and standard deviation of the target dimension feature value vector and the surrounding rock sampling vector respectively, and then normalize them feature by feature. This can ensure that the numerical values of the two sets of data are at a similar scale in their respective dimensions. If the distribution characteristics of the two sets of data differ greatly, an appropriate normalization method can be selected according to the specific situation to retain as much data information as possible.

[0097] Step S42, splice the standard target dimension feature value and the standard surrounding rock sampling vector to obtain the surrounding rock feature vector, and input the surrounding rock feature vector into the chaotic neural network.

[0098] The expression of the chaotic neural network is:

[0099] F(t) = [d WS (t), σ(t), ε(t), P(t),....];

[0100]

[0101] y i (t + 1) = f(x i (t + 1));

[0102] Wherein, F(t) represents the surrounding rock characteristic vector input into the chaotic neural network at the t-th moment; d WS (t) represents the target dimension eigenvalue at the t-th moment; ε(t) represents the strain value of the surrounding rock at the t-th moment; σ(t) represents the compressive strength of the rock mass material monitored at the t-th moment; P(t) represents the real-time surrounding rock pressure monitored at the t-th moment; x i (t) represents the internal state function of the i-th neuron at the t-th moment; g( ) represents the folding-back mapping function; J represents the total number of quantized neurons; w ij represents the contribution connection weight of the j-th quantized neuron to the state update of the i-th neuron; H represents the dimension number of the surrounding rock sampling vector; v ih represents the contribution connection weight of the surrounding rock characteristic under the h-th dimension to the state update of the i-th neuron; F h (t) represents the surrounding rock characteristic under the h-th dimension in the chaotic neural network input at the t-th moment; θ i represents the bias parameter of the i-th neuron; c represents the perturbation intensity parameter; z(t) represents the chaotic sequence generated by the chaotic mapping; y i (t + 1) represents the output state function of the i-th neuron at the (t + 1)-th moment; f( ) represents the non-linear activation function.

[0103] Step S43, based on the internal state function and output state function of the chaotic neural network, obtain the prediction result of the surrounding rock deformation.

[0104] In this embodiment, by respectively normalizing the target dimension eigenvalue and the surrounding rock sampling vector, the dimension and numerical range differences between different data are eliminated, making the two groups of data on the same numerical scale. This not only improves the consistency and comparability of the data, but also provides a reliable basis for subsequent data fusion, reducing the problems caused by numerical instability during the neural network training process. The normalized standard target dimension eigenvalue and the standard surrounding rock sampling vector form a unified surrounding rock characteristic vector through vector splicing. This fusion method can fully retain the key information in the two groups of data, and at the same time construct a comprehensive description containing multi-dimensional characteristics, providing richer and more comprehensive input information for the chaotic neural network, which is conducive to capturing the complex non-linear characteristics in the surrounding rock deformation process. Based on the constructed surrounding rock characteristic vector, the chaotic neural network uses its internal state function and output state function for dynamic modeling and prediction, and can effectively learn and express the complex dynamic behavior of the system.

[0105] In one example, assume that the output layer of the chaotic neural network has only one neuron, and the output value is denoted as y(t), with its numerical range between [0, 1], representing the comprehensive index of "surrounding rock instability degree". When y(t) is closer to 1, it means that the model judges that the surrounding rock is closer to the unstable state; when y(t) is closer to 0, it means that the model judges that the surrounding rock is closer to the stable state. According to past engineering monitoring data and on-site experience, for the convenience of making quick decisions, two thresholds T1 and T2 (where 0 ≤ T1 < T2 < 1) can be set to divide the output value into three risk levels. For example: Stable area: 0 ≤ y(t) < T1; Critical area: T1 ≤ y(t) < T2; Unstable area: T2 ≤ y(t) ≤ 1.

[0106] The specific thresholds can be set based on historical surrounding rock rupture and instability data through statistical analysis or expert experience. For example, T1 = 0.4 and T2 = 0.7. If the network output y(t) < 0.4, it is considered that the surrounding rock is currently in a safe or relatively stable interval; if the network output y(t) is between 0.4 and 0.7, it may mean that certain degree of micro-crack expansion has occurred inside the surrounding rock, and monitoring and early warning need to be strengthened; if the network output y(t) ≥ 0.7, it indicates that the risk of surrounding rock instability increases significantly, and safety measures such as support, pressure relief or evacuation should be taken immediately.

[0107] Furthermore, in actual engineering, the state of the surrounding rock evolves continuously over time, and a single output only represents the situation at a certain moment. For a more stable and reliable judgment, the following method can be used to conduct trend analysis on y(t): Take the average value or weighted average of y(t) within a time window (such as the last 5 or 10 times of monitoring) to smooth out the influence of noise; If the output y(t) exceeds a certain threshold in consecutive multiple samplings, then trigger early warning or alarm; Evaluate the growth (or decline) rate of y(t) in the time dimension. If there is a rapid increase in a short period of time, it indicates that the condition of the surrounding rock has deteriorated and should be highly concerned.

[0108] In this embodiment, by performing acoustic emission monitoring and sampling vector recording on the surrounding rock of the underground powerhouse, multi-angle data collection of the surrounding rock state is achieved, ensuring that the monitoring data has higher information content and integrity. Moreover, by using time-series data to capture subtle changes in the surrounding rock, the timeliness and accuracy of monitoring and early warning are improved. The fractal dimension model is used to model the monitoring data, which can reveal the complex non-linear behavior during the deformation process of the surrounding rock. By detecting the fractal dimension function, the inherent structural characteristics of the data are captured. The hybrid leapfrog algorithm is used for local search. For the adjustment of the scaling factor, correction coefficient, and scale factor, the optimal parameter combination can be obtained, significantly improving the accuracy and robustness of the fractal model in capturing the deformation characteristics of the surrounding rock. At the same time, by using the chaotic neural network to input and process the target features and sampling vectors, the randomness, non-linearity, and complexity problems existing in the system can be better handled, realizing the efficient prediction of the deformation process of the surrounding rock.

[0109] See Figure 3 , the present invention also provides a prediction system for the stability of the surrounding rock of an underground powerhouse, including:

[0110] A time-series monitoring data acquisition module, configured to acquire time-series monitoring data of the surrounding rock of the underground powerhouse, where the time-series monitoring data includes surrounding rock acoustic emission signal data and surrounding rock sampling vectors;

[0111] A target dimension eigenvalue acquisition module, configured to perform fractal dimension modeling on the time-series monitoring data to obtain a detection fractal dimension function, and perform local search on the scaling factor, correction coefficient, and scale factor in the detection fractal dimension function according to the hybrid leapfrog algorithm to obtain an optimal scaling factor, an optimal correction coefficient, and an optimal scale factor; based on the optimal scaling factor, the optimal correction coefficient, the optimal scale factor, and the detection fractal dimension function, construct a target dimension feature extraction function, and extract the target dimension eigenvalues in the time-series monitoring data according to the target dimension feature extraction function;

[0112] A surrounding rock deformation prediction module, configured to input the target dimension eigenvalues and the surrounding rock sampling vectors into a chaotic neural network for surrounding rock deformation prediction to obtain a surrounding rock deformation prediction result.

[0113] In one example, the time-series monitoring data acquisition module is configured to obtain the surrounding rock monitoring data, the surrounding rock acoustic emission signal data, and the supporting structure force monitoring data corresponding to multiple measuring points within a preset interval distance range in the underground space corresponding to the underground powerhouse, integrate all the supporting structure force monitoring data and all the surrounding rock monitoring data, and construct multiple surrounding rock sampling vectors corresponding to each measuring point.

[0114] In one example, when the time-series monitoring data acquisition module is used to acquire surrounding rock monitoring data and support structure stress monitoring data, each corresponding measuring point simultaneously monitors the horizontal and vertical displacements of the surrounding rock, and simultaneously monitors the radial and tangential surrounding rock stresses of the surrounding rock.

[0115] In one example, the target dimension eigenvalue acquisition module is used to segment the surrounding rock acoustic emission signal data on the time axis, obtain multiple segments of surrounding rock acoustic emission sequences, and perform amplitude normalization processing on the surrounding rock acoustic emission sequences to obtain standard surrounding rock acoustic emission sequences; according to the standard surrounding rock acoustic emission sequences, construct an initial detection fractal dimension function; calculate the local fractal values at different sampling scales through the initial detection fractal dimension function, and obtain the sampling logarithmic relationship between the sampling scale and the corresponding local fractal values through linear fitting; based on the sampling logarithmic relationship and the initial detection fractal dimension function, construct a detection fractal dimension function.

[0116] In one example, the target dimension eigenvalue acquisition module is used to randomly generate multiple frog individuals and initialize them within the feasible region of the frog individuals, where each frog individual represents an optimization parameter combination including a scaling factor, a correction coefficient, and a scale factor; based on the detection fractal dimension function and the fractal reference value, construct a fractal reference function, and calculate the fitness of each frog individual according to the fractal reference function and the fractal reference value; sort the frog individuals in the population in ascending order of fitness, and divide the population into multiple sub-cultural gene bodies, where each cultural gene body contains several frog individuals; in each sub-cultural gene body, respectively select the frog individual with the optimal fitness and the frog individual with the worst fitness, and perform position update on the frog individual with the worst fitness based on the position update function to perform local search; when at least one round of local search has been performed in each cultural gene body and the maximum number of iterations is reached, recombine the population and sort it again in ascending order of fitness, update the optimal frog individual in the population, and record the position of the global optimal frog individual.

[0117] In one example, the surrounding rock deformation prediction module is used to perform normalization processing on the target dimension eigenvalue and the surrounding rock sampling vector to obtain a standard target dimension eigenvalue and a standard surrounding rock sampling vector; splice the standard target dimension eigenvalue and the standard surrounding rock sampling vector to obtain a surrounding rock feature vector, and input the surrounding rock feature vector into a chaotic neural network; based on the internal state function and the output state function of the chaotic neural network, obtain the surrounding rock deformation prediction result.

[0118] In one example, the expression of the target dimension feature extraction function is:

[0119]

[0120] d WS = T(δ; α * , β * , m* )

[0121] Among them, T() represents the detection fractal dimension function; δ represents the sampling scale, which is an integer multiple of the minimum sampling interval; x n represents the amplitude of the nth sampling point after the surrounding rock acoustic emission signal data is segmented; N represents the minimum number of scales required to completely cover the signal at the sampling scale δ; α represents the scaling factor corresponding to the cumulative length of the signal; β represents the correction coefficient corresponding to the mapping value of the signal change and scale relationship; m represents the scale factor corresponding to the scale window used in the fractal dimension estimation; MSE represents the mean square error; K represents the total number of surrounding rock acoustic emission sequences in the surrounding rock acoustic emission signal data; T k represents the fractal dimension value under the kth surrounding rock acoustic emission sequence; represents the fractal reference value under the kth surrounding rock acoustic emission sequence; d WS represents the target dimension feature extraction function; α * represents the optimal scaling factor corresponding to the cumulative length of the signal; β * represents the optimal correction coefficient corresponding to the mapping value of the signal change and scale relationship; m * represents the optimal scale factor corresponding to the scale window used in the fractal dimension estimation.

[0122] In an example, the expression of the chaotic neural network is:

[0123] F(t) = [d WS (t), σ(t), ε(t), P(t),....];

[0124]

[0125] y i (t + 1) = f(x i (t + 1));

[0126] Among them, F(t) represents the surrounding rock feature vector input into the chaotic neural network at the t-th moment; d WS (t) represents the target dimension feature value at the t-th moment; ε(t) represents the strain value of the surrounding rock at the t-th moment; σ(t) represents the compressive strength of the rock mass material monitored at the t-th moment; P(t) represents the real-time surrounding rock pressure monitored at the t-th moment; x i (t) represents the internal state function of the i-th neuron at the t-th moment; g( ) represents the folding-back mapping function; J represents the total number of quantization neurons; w ij represents the contribution connection weight of the j-th quantization neuron to the state update of the i-th neuron; H represents the dimension number of the surrounding rock sampling vector; v ih represents the contribution connection weight of the surrounding rock feature under the h-th dimension to the state update of the i-th neuron; F h(t) represents the surrounding rock characteristics in the h-th dimension input into the chaotic neural network at the t-th moment; θ i represents the bias parameter of the i-th neuron; c represents the perturbation intensity parameter; z(t) represents the chaotic sequence generated by the chaotic mapping; y i (t + 1) represents the output state function of the i-th neuron at the (t + 1)-th moment; f() represents the non-linear activation function.

[0127] See Figure 4 , the present invention also provides an underground powerhouse surrounding rock stability prediction device, including a memory and a processor;

[0128] The memory is used to store computer program codes and transmit the computer program codes to the processor;

[0129] The processor is used to execute the above-mentioned method for predicting the stability of the surrounding rock of the underground powerhouse according to the instructions in the computer program codes.

[0130] The present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the above-mentioned method for predicting the stability of the surrounding rock of the underground powerhouse is realized.

[0131] Generally speaking, the computer instructions for implementing the method of the present invention can be carried by any combination of one or more computer-readable storage media. The non-temporary computer-readable storage media can include any computer-readable medium except the signal propagating temporarily itself.

[0132] The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples (non-exhaustive list) of the computer-readable storage medium include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EKROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present invention, the computer-readable storage medium can be any tangible medium that contains or stores a program, and this program can be used by or in combination with an instruction execution system, apparatus, or device.

[0133] Computer program code for performing the operations of the present invention may be written in one or more programming languages or combinations thereof. The programming languages include object-oriented programming languages such as Java, Smalltalk, C++, and also include conventional procedural programming languages such as the "C" language or similar programming languages. In particular, the Python language suitable for neural network computing and platform frameworks based on TensorFlow, PyTorch, etc. can be used. The program code can be executed entirely on the user's computer, partially on the user's computer, executed as an independent software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or connected to an external computer (e.g., connected through the Internet using an Internet service provider).

[0134] For the above-mentioned devices and non-transitory computer-readable storage media, reference may be made to the specific description of a method for predicting the stability of surrounding rock in an underground powerhouse and its beneficial effects, which will not be elaborated herein.

[0135] Although the embodiments of the present invention have been shown and described above, it should be understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those of ordinary skill in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present invention.

Claims

1. A method for predicting the stability of surrounding rock in an underground powerhouse, characterized in that, Including: Collecting time-series monitoring data of the surrounding rock of an underground powerhouse, where the time-series monitoring data includes surrounding rock acoustic emission signal data and surrounding rock sampling vectors; Performing fractal dimension modeling on the time-series monitoring data to obtain a detection fractal dimension function, and locally searching for the scaling factor, correction coefficient, and scale factor in the detection fractal dimension function according to the hybrid leapfrog algorithm to obtain the optimal scaling factor, optimal correction coefficient, and optimal scale factor; Based on the optimal scaling factor, the optimal correction coefficient, the optimal scale factor, and the detection fractal dimension function, constructing a target dimension feature extraction function, and extracting target dimension feature values from the time-series monitoring data according to the target dimension feature extraction function; Inputting the target dimension feature values and the surrounding rock sampling vectors into a chaotic neural network for surrounding rock deformation prediction to obtain a surrounding rock deformation prediction result.

2. The method for predicting the stability of surrounding rock in an underground powerhouse according to claim 1, wherein The step of collecting time-series monitoring data of the surrounding rock of the underground powerhouse specifically includes: Obtaining the surrounding rock monitoring data, surrounding rock acoustic emission signal data corresponding to multiple measurement points within a preset interval distance range in the underground space corresponding to the underground powerhouse, and the supporting structure force monitoring data corresponding to multiple measurement points, integrating all the supporting structure force monitoring data and all the surrounding rock monitoring data, and constructing multiple surrounding rock sampling vectors corresponding to each measurement point; When collecting the surrounding rock monitoring data and the supporting structure force monitoring data, each corresponding measurement point simultaneously monitors the horizontal and vertical displacements of the surrounding rock and the radial and tangential surrounding rock stresses of the surrounding rock.

3. A method for predicting the stability of surrounding rock in an underground power house according to claim 1, characterized in that, The step of performing fractal dimension modeling on the time-series monitoring data to obtain a detection fractal dimension function specifically includes: Segmenting the surrounding rock acoustic emission signal data on the time axis to obtain multiple segments of surrounding rock acoustic emission sequences, and performing amplitude normalization processing on the surrounding rock acoustic emission sequences to obtain standard surrounding rock acoustic emission sequences; Constructing an initial detection fractal dimension function according to the standard surrounding rock acoustic emission sequences; Calculating the local fractal values at different sampling scales through the initial detection fractal dimension function, and obtaining the sampling logarithmic relationship between the sampling scale and the corresponding local fractal values according to linear fitting; Based on the sampling logarithmic relationship and the initial detection fractal dimension function, constructing a detection fractal dimension function.

4. A method for predicting the stability of surrounding rock in an underground powerhouse according to claim 3, characterized in that, The step of locally searching for the scaling factor, correction coefficient, and scale factor in the detection fractal dimension function according to the hybrid leapfrog algorithm specifically includes: Randomly generating multiple frog individuals and initializing them within the feasible domain of the frog individuals, where each frog individual represents an optimization parameter combination including a scaling factor, a correction coefficient, and a scale factor; Based on the detection fractal dimension function and the fractal reference value, constructing a fractal reference function, and calculating the fitness of each frog individual according to the fractal reference function and the fractal reference value; Sorting the frog individuals in the population in ascending order of fitness, and dividing the population into multiple sub-cultural gene bodies, where each cultural gene body contains several frog individuals; Selecting the frog individual with the optimal fitness and the frog individual with the worst fitness in each sub-cultural gene body respectively, and updating the position of the worst frog individual based on the position update function to perform local search; When at least one round of local search has been performed on each meme and the maximum number of iterations has been reached, the population is recombined and sorted again in ascending order according to fitness. The optimal frog individual in the population is updated, and the position of the globally optimal frog individual is recorded.

5. A method for predicting the stability of surrounding rock in an underground powerhouse according to claim 1, characterized in that, The method of inputting the target dimension eigenvalue and the surrounding rock sampling vector into the chaotic neural network for surrounding rock deformation prediction to obtain a surrounding rock deformation prediction result specifically includes: Performing normalization processing on the target dimension eigenvalue and the surrounding rock sampling vector to obtain a standard target dimension eigenvalue and a standard surrounding rock sampling vector; Performing vector splicing on the standard target dimension eigenvalue and the standard surrounding rock sampling vector to obtain a surrounding rock feature vector, and inputting the surrounding rock feature vector into the chaotic neural network; Based on the internal state function and output state function of the chaotic neural network, obtaining a surrounding rock deformation prediction result.

6. A method for predicting the stability of surrounding rock in an underground powerhouse according to claim 1, characterized in that, The expression of the target dimension feature extraction function is: d WS = T(δ; α * , β * , m * ); Among them, T() represents the function for detecting the fractal dimension; δ represents the sampling scale; x n represents the amplitude of the nth sampling point after the surrounding rock acoustic emission signal data is segmented; N represents the minimum number of scale units required to completely cover the signal at the sampling scale δ; α represents the scaling factor corresponding to the cumulative length of the signal; β represents the correction coefficient corresponding to the mapping value of the relationship between the signal change and the scale; m represents the scale factor corresponding to the scale window used in the fractal dimension estimation; MSE represents the mean square error; K represents the total number of surrounding rock acoustic emission sequences in the surrounding rock acoustic emission signal data; T k represents the fractal dimension value under the kth surrounding rock acoustic emission sequence; represents the fractal reference value under the kth surrounding rock acoustic emission sequence; d WS represents the target dimension feature extraction function; α * represents the optimal scaling factor corresponding to the cumulative length of the signal; β * represents the optimal correction coefficient corresponding to the mapping value of the relationship between the signal change and the scale; m * represents the optimal scale factor corresponding to the scale window used in the fractal dimension estimation.

7. A method for predicting the stability of surrounding rock in an underground powerhouse according to claim 5, characterized in that, The expression of the chaotic neural network is: F(t) = [d WS (t), σ(t), ε(t), P(t),....]; y i (t + 1) = f(x i (t + 1)); Among them, F(t) represents the surrounding rock characteristic vector input into the chaotic neural network at the t-th moment; d WS (t) represents the target dimension eigenvalue at the t-th moment; ε(t) represents the strain value of the surrounding rock at the t-th moment; σ(t) represents the compressive strength of the rock mass material monitored at the t-th moment; P(t) represents the real-time surrounding rock pressure monitored at the t-th moment; x i (t) represents the internal state function of the i-th neuron at the t-th moment; g() represents the folding-back mapping function; J represents the total number of quantization neurons; w ij represents the contribution connection weight of the j-th quantization neuron to the state update of the i-th neuron; H represents the dimension number of the surrounding rock sampling vector; v ih represents the contribution connection weight of the surrounding rock characteristic under the h-th dimension to the state update of the i-th neuron; F h (t) represents the surrounding rock characteristic under the h-th dimension in the chaotic neural network input at the t-th moment; θ i represents the bias parameter of the i-th neuron; c represents the perturbation intensity parameter; z(t) represents the chaotic sequence generated by the chaotic mapping; y i (t + 1) represents the output state function of the i-th neuron at the (t + 1)-th moment; f() represents the non-linear activation function.

8. A prediction system for the stability of surrounding rocks in an underground powerhouse, characterized in that, The system is applied to the method described in any one of claims 1-7. The system includes: A time-series monitoring data acquisition module for acquiring time-series monitoring data of the surrounding rock of the underground powerhouse, where the time-series monitoring data includes surrounding rock acoustic emission signal data and surrounding rock sampling vectors; A target dimension eigenvalue acquisition module for performing fractal dimension modeling on the time-series monitoring data to obtain a detection fractal dimension function, and performing local search on the scaling factor, correction coefficient, and scale factor in the detection fractal dimension function according to the hybrid leapfrog algorithm to obtain an optimal scaling factor, an optimal correction coefficient, and an optimal scale factor; constructing a target dimension feature extraction function based on the optimal scaling factor, the optimal correction coefficient, the optimal scale factor, and the detection fractal dimension function, and extracting the target dimension eigenvalue in the time-series monitoring data according to the target dimension feature extraction function; A surrounding rock deformation prediction module for inputting the target dimension eigenvalue and the surrounding rock sampling vector into the chaotic neural network for surrounding rock deformation prediction to obtain a surrounding rock deformation prediction result.

9. An underground powerhouse surrounding rock stability prediction device, characterized in that it includes a memory and a processor; The memory is used to store computer program code and transmit the computer program code to the processor; The processor is used to execute the method described in any one of claims 1 to 7 according to the instructions in the computer program code.

10. A computer-readable storage medium, characterized in that, A computer program is stored on the computer-readable storage medium, and when the computer program is executed by the processor, the method described in any one of claims 1 to 7 is implemented.

Images