A method and system for predicting cumulative damage of surrounding rock of deep-buried highway tunnel by blasting

Abstract

The present application relates to the field of tunnel engineering blasting technology, especially to a deep buried highway tunnel blasting surrounding rock cumulative damage prediction method and system, the method comprises the following steps: collecting the measured vibration time history signal of the buried depth test area through blasting vibration experiment and carrying out noise reduction processing to obtain the vibration signal; the blasting vibration and unloading vibration signal analysis is carried out to the vibration signal, and then the separation of blasting vibration signal and unloading vibration signal is realized; on the basis of signal separation result, the characteristic variable and related variable of surrounding rock cumulative damage are determined through surrounding rock vibration response and damage correlation analysis; the cumulative damage prediction model is constructed, and then the prediction of related variable is realized by using characteristic variable. The present application solves the problem that the existing technology has not constructed the surrounding rock cumulative damage prediction model based on the characteristics of blasting vibration and unloading vibration, and the prediction result of the model is more reliable, which has certain guiding significance for the control of deep buried tunnel excavation blasting surrounding rock damage and vibration effect.

Description

Technical Field

[0001] This invention relates to the field of tunnel engineering blasting technology, and in particular to a method and system for predicting cumulative damage to surrounding rock during blasting in deep-buried highway tunnels. Background Technology

[0002] Accurately predicting the damage to rock masses under blasting excavation disturbance is a key aspect of underground structure design and stability assessment. The degree of damage to the rock mass depends on factors such as the charge quantity, the distance to the blast source, and the properties of the rock mass. However, incorporating all these factors into the analysis is quite difficult because the problem involves numerous complex physicochemical and mechanical processes, such as explosive detonation, crack initiation, and propagation.

[0003] Currently, scholars both domestically and internationally have conducted numerous studies on surrounding rock damage prediction and have achieved some beneficial results. However, most of these studies involve simple regression analysis of blasting vibration PPV (peak velocity of blasting particles) and damage to formulate prediction formulas. A neural network prediction model for surrounding rock damage based on the characteristics of blasting vibration and unloading vibration has not yet been constructed. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a method and system for predicting cumulative damage to surrounding rock during blasting in deeply buried highway tunnels.

[0005] To achieve the above objectives, in a first aspect, the present invention provides a method for predicting cumulative damage to surrounding rock in deep-buried highway tunnel blasting. The method includes the following steps: acquiring measured vibration time-history signals from a test area at a burial depth through blasting vibration experiments, and performing noise reduction processing on the measured vibration time-history signals to obtain vibration signals; analyzing the vibration signals for blasting vibration and transient unloading vibration, thereby achieving separation of the blasting vibration signals and transient unloading vibration signals; based on the signal separation results, determining the characteristic variables and related variables of cumulative damage to the surrounding rock through correlation analysis of surrounding rock vibration response and damage; constructing a cumulative damage prediction model for predicting cumulative damage to surrounding rock in deep-buried highway tunnel blasting, and inputting the related variables into the cumulative damage prediction model to achieve prediction of the characteristic variables. This method solves the problem that existing technologies have not yet constructed a cumulative damage prediction model for surrounding rock based on the characteristics of blasting vibration and unloading vibration. The prediction results of this model are relatively reliable and have certain guiding significance for controlling the damage and vibration effects of surrounding rock in deep-buried tunnel excavation blasting.

[0006] Optionally, the step of acquiring the measured vibration time history signal of the burial depth test area through the blasting vibration experiment, and performing noise reduction processing on the measured vibration time history signal to obtain the vibration signal, includes the following steps: Measured vibration time history signals of the buried test area were collected through blasting vibration experiments. The measured vibration time history signal was decomposed into multiple modal function components using a variational mode decomposition algorithm; Multi-scale permutation entropy analysis is performed on the modal function components to identify and eliminate noise components, thereby obtaining the vibration signal.

[0007] Optionally, when using the variational mode decomposition algorithm, if the difference between the signal energy change rate of the mode number K+1 and the signal energy change rate of the mode number K exceeds a preset threshold, K is taken as the optimal number of modes; when using the variational mode decomposition algorithm, with the aim of maximizing the signal-to-noise ratio of the reconstructed signal, the particle swarm optimization algorithm is used to obtain the optimal penalty factor.

[0008] Optionally, the step of analyzing the vibration signal for blasting vibration and transient unloading vibration to separate the blasting vibration signal from the transient unloading vibration signal includes the following steps: Based on the Hilbert transform method, the instantaneous energy of the vibration signal is calculated to identify blasting vibration and transient unloading vibration. Based on the vibration identification results, a finite impulse response low-pass digital filter is used to filter the vibration signal, thereby separating the blasting vibration signal from the transient unloading vibration signal.

[0009] Optionally, the step of determining the characteristic variables and related variables of cumulative damage in the surrounding rock through correlation analysis between the surrounding rock vibration response and damage, based on the signal separation results, includes the following steps: Based on the signal separation results, the characteristic variable and alternative relevant variable of the cumulative damage to the surrounding rock are determined, wherein the characteristic variable is the range of damage to the surrounding rock; The candidate relevant variables were screened through correlation analysis to obtain the relevant variables.

[0010] Optionally, the step of screening the candidate relevant variables through correlation analysis to obtain the relevant variables includes the following steps: Collect candidate relevant variable data and corresponding feature variable data, and then calculate the correlation between the candidate relevant variables and the feature variables; Set a correlation threshold. When the correlation is greater than the correlation threshold, the corresponding candidate correlation variable is used as the correlation variable.

[0011] Optionally, the construction of a cumulative damage prediction model for predicting cumulative damage to the surrounding rock in deep-buried highway tunnel blasting, by inputting the relevant variables into the cumulative damage prediction model to predict the feature variables, includes the following steps: Construct a dataset for predicting cumulative damage to surrounding rock; The sparrow search algorithm was improved to obtain the improved sparrow search algorithm; An improved sparrow search algorithm is used to optimize the weights and thresholds of the BP neural network, and then the cumulative damage prediction model is constructed using the damage prediction dataset and the BP neural network. The relevant variables are input into the cumulative damage prediction model to predict the feature variables.

[0012] Optionally, the improvement of the sparrow search algorithm to obtain an improved sparrow search algorithm includes the following steps: The sparrow population was initialized using a dimensionally balanced hierarchical random grid method. The discoverer location update formula is improved by introducing a reference individual and a random perturbation factor.

[0013] Optionally, the initialization of the sparrow population using the dimensionally balanced hierarchical random grid method includes the following steps: Determine the search space dimension D and the population size N. For each of the search space dimensions, divide the interval [0,1] into N equally probable subintervals. For each dimension of the search space, a value is randomly selected from each of its sub-intervals to obtain D vectors of length N; The elements of the D vectors are randomly combined to form N D-dimensional sample sequences; The numerical values ​​in the sample sequence are linearly mapped to the actual search space, and the N mapped sample sequences are used as the initial population of the sparrow search algorithm.

[0014] Secondly, the present invention provides a system for predicting cumulative damage to blasting surrounding rock in deeply buried highway tunnels. The system includes: a data acquisition device, a data output device, a processor, and a storage device. The storage device includes a computer-readable storage medium storing a computer program. The computer program includes program instructions, which, when executed by the processor, cause the processor to implement the method for predicting cumulative damage to blasting surrounding rock in deeply buried highway tunnels provided by the present invention.

[0015] The present invention has at least the following beneficial effects: 1. By automatically judging the mode number K and penalty factor α of the variational mode decomposition algorithm, the problems of mode aliasing and over-decomposition during signal decomposition can be effectively solved. Combined with multi-scale permutation entropy analysis, noise signals can be accurately removed to achieve the purpose of noise reduction.

[0016] 2. The blasting vibration and transient unloading vibration signals were analyzed and identified, and the separation of the blasting vibration signal and transient unloading vibration signal was realized, laying the foundation for constructing a prediction model of cumulative damage of surrounding rock based on the characteristics of blasting vibration and unloading vibration.

[0017] 3. Taking into account the characteristics of blasting vibration and unloading vibration, an improved sparrow search algorithm was used to optimize the weights and thresholds of the BP neural network, thereby constructing a cumulative damage prediction model. This solves the problem that existing technologies have not yet constructed a cumulative damage prediction model for surrounding rock based on the characteristics of blasting vibration and unloading vibration.

[0018] 4. The sparrow population is initialized using a dimensionally balanced hierarchical random grid method, which makes the initial sparrow population highly uniform, controllable, and applicable. By modifying the finder position update formula and adaptively updating the safety threshold, the search capability of the sparrow search algorithm is increased, which is beneficial to improving the performance of the cumulative damage prediction model.

[0019] 5. A system adapted to this method is provided, which can improve the practicality of this method and facilitate its promotion. Attached Figure Description

[0020] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a flowchart illustrating a method for predicting cumulative damage to surrounding rock during blasting in a deep-buried highway tunnel, according to an embodiment of the present invention. Figure 2 This is a partial measured vibration time history signal at 200 m of the target tunnel in this embodiment of the invention; Figure 3 This is a partial measured vibration time history signal at 350 m of the target tunnel in this embodiment of the invention; Figure 4 This is a partial measured vibration time history signal at 500 m of the target tunnel in this embodiment of the invention; Figure 5 These are the five modal function components of a typical vibration signal according to an embodiment of the present invention; Figure 6 This is a comparison diagram of typical vibration signals and reconstructed signals in an embodiment of the present invention; Figure 7 The vibration waveforms and instantaneous energy curves at some measuring points and boreholes in this embodiment of the invention are shown below. Figure 8 The signal separation results of vibration waveforms and instantaneous energy curves at some measuring points and boreholes in this embodiment of the invention; Figure 9 The results of the correlation analysis between relevant variables and feature variables in this embodiment of the invention; Figure 10 This is a schematic diagram of the framework of a system for predicting cumulative damage to surrounding rock during blasting in a deep-buried highway tunnel, according to an embodiment of the present invention. Detailed Implementation

[0022] Specific embodiments of the present invention will now be described in detail. It should be noted that the embodiments described herein are for illustrative purposes only and are not intended to limit the invention. In the following description, numerous specific details are set forth in order to provide a thorough understanding of the invention. However, it will be apparent to those skilled in the art that these specific details are not necessary to practice the invention. In other instances, well-known circuits, software, or methods have not been specifically described to avoid obscuring the invention.

[0023] Throughout this specification, references to "an embodiment," "an embodiment," "an example," or "an example" mean that a particular feature, structure, or characteristic described in connection with that embodiment or example is included in at least one embodiment of the invention. Therefore, the phrases "in an embodiment," "in an embodiment," "an example," or "an example" appearing in various places throughout the specification do not necessarily refer to the same embodiment or example. Furthermore, specific features, structures, or characteristics can be combined in one or more embodiments or examples in any suitable combination and / or sub-combination. Moreover, those skilled in the art will understand that the illustrations provided herein are for illustrative purposes and are not necessarily drawn to scale.

[0024] It should be noted in advance that, in one alternative embodiment, except for independent descriptions, the same symbols or letters appearing in all formulas have the same meaning and value.

[0025] In one optional embodiment, please refer to Figure 1 This invention provides a method for predicting cumulative damage to surrounding rock during blasting in deep-buried highway tunnels, the method comprising the following steps: S1. The measured vibration time history signal of the buried test area is collected through the blasting vibration experiment, and the measured vibration time history signal is denoised to obtain the vibration signal.

[0026] Step S1 specifically includes the following steps: S11. Collect measured vibration time history signals of the buried depth test area through blasting vibration experiments.

[0027] Specifically, in this embodiment, three typical burial depth test areas at 200 m, 350 m, and 500 m of the target tunnel were drilled using the same hole arrangement, including slotting holes, widening holes, auxiliary holes 1-4, peripheral holes, and bottom plate holes. A digital electronic detonator was used for blasting vibration experiments. During the experiments, a TC-6850 blasting vibration meter was used to collect measured vibration time history signals at the measuring points in the burial depth test areas. This blasting vibration meter has a velocity range of 0~25 cm / s, a vibration frequency range of 0.01~300 Hz, and a sampling rate of 6.4 KSps. Before each blast, three monitoring points were arranged on the right arch waist of the upper step of the preceding tunnel. The distances of the first, second, and third monitoring points from the blasting face were 30 m, 38 m, and 46 m, respectively, and the height of the monitoring points from the ground was approximately 1.5 m. During installation, the X, Y, and Z axes of the blasting vibration meter were aligned with the longitudinal, transverse, and vertical directions of the tunnel, respectively. Meanwhile, to prevent flying rocks and shock waves from affecting the monitoring data, a metal protective cover was fixed around the blasting vibration meter. The measured vibration time history signals at 200 m, 350 m, and 500 m of the target tunnel are as follows: Figure 2 , Figure 3 and Figure 4 As shown.

[0028] Furthermore, when conducting blasting vibration analysis, there is currently no consensus on whether to describe the vibration velocity using one of the three velocity components (X, Y, and Z) or the sum of the three-dimensional velocity vectors. This can be determined through observation. Figure 2 , Figure 3 and Figure 4 It can be observed that among the vibration signals in the three directions, the vibration velocity in the Y direction is the largest, followed by the vibration velocity in the X direction, and the vibration velocity in the Z direction is the smallest. Therefore, this embodiment selects the vibration signal in the Y direction as the main research object.

[0029] S12. The measured vibration time history signal is decomposed into multiple modal function components using a variational mode decomposition algorithm.

[0030] Specifically, in this embodiment, due to the complexity of the tunnel construction environment and electromagnetic interference, the measured original vibration signal often carries a large amount of high-frequency noise. To obtain a more realistic vibration signal, noise reduction processing is necessary. Currently, commonly used noise reduction algorithms for blasting vibration signals include classical mode decomposition (DMD), wavelet-based algorithms, and improved versions of DM. While these algorithms can meet the noise removal requirements to a certain extent, they still suffer from problems such as mode aliasing, subjective decision-making, and endpoint effects. Variational mode decomposition (VMD), as a completely non-recursive adaptive signal processing method, overcomes the mode aliasing problem of classical DM and the poor adaptability of wavelet algorithms. Its essence is a variational problem-solving process based on Wiener filtering, Hibert transform, and frequency mixing. Therefore, this embodiment uses VMD to achieve signal noise reduction.

[0031] In Variational Mode Decomposition (VMD) algorithms, when the number of modes K is increased, if the newly decomposed modes mainly contain noise or redundant information, their energy contribution will decrease significantly, leading to a slower or even decreased increase in the rate of change of total energy. Based on this, when the difference between the energy change rate of a signal with K+1 modes and that with K modes exceeds a preset threshold, the signal is in an over-decomposition state. Therefore, it can be determined that the number of modes K at this point is optimal. In this embodiment, the preset threshold is set to 0.10, and the signal energy change rate satisfies the following relationship: in, Let T be the rate of change of signal energy, T be the signal acquisition duration, and t be time. As the starting point of the signal, For the signal, K is the number of modes. Let be the k-th modal component of the signal.

[0032] After determining the method for obtaining the number of modes K, the optimal penalty factor is obtained using a particle swarm optimization algorithm with the aim of maximizing the signal-to-noise ratio of the reconstructed signal. Specifically, the particle dimension is 1 (only the penalty factor is optimized), the particle position range is [100, 5000], the particle velocity range is [-500, 500], the number of particles is 20, and the maximum number of iterations is 50.

[0033] S13. Perform multi-scale permutation entropy analysis on the modal function components to identify and eliminate noise components, thereby obtaining the vibration signal.

[0034] Specifically, in this embodiment, after decomposing the measured vibration time history signal into multiple modal function components using VMD, each modal function component is first subjected to multi-scale coarse-graining processing to obtain a coarse-grained sequence. Then, the coarse-grained sequences are reconstructed sequentially to obtain a reconstructed sequence, and the probability of each reconstructed sequence is calculated by statistically analyzing the reconstructed sequences arranged in ascending order. Finally, the permutation entropy of each reconstructed sequence is calculated, and the calculated permutation entropy is normalized to obtain the normalized permutation entropy. The calculated normalized permutation entropy can reflect the complexity of the modal function components. The larger the normalized permutation entropy, the more likely the corresponding modal function component is to have certain special vibration characteristics or interference factors. In this embodiment, the modal function components corresponding to a normalized permutation entropy greater than 0.8 are regarded as noise components and discarded, and the vibration signal is reconstructed using the remaining modal function components.

[0035] Furthermore, this embodiment processes a typical vibration signal to verify the signal noise reduction effect of the proposed solution. Table 1 shows the rate of change of signal energy and the difference between the rate of change of signal energy for this typical vibration signal under different K values.

[0036] Table 1. Rate of change of signal energy and difference of rate of change of signal energy at different K values. In Table 1, The rate of change of signal energy when the number of modes is k+1. Let K be the rate of change of signal energy when the number of modes is k. As shown in Table 1, when K increases from 5 to 6, the rate of change of signal energy increases from 0.1181 to 0.3027. If the value exceeds 0.10, the signal is in an over-decomposition state. Therefore, it can be determined that the typical vibration signal is best decomposed into 5 modal components. At the same time, with the help of the particle swarm optimization algorithm, the optimal value of the penalty factor is determined to be 1700.

[0037] When K is 5 and the penalty factor is 1700, the first modal function component IMF1 to the fifth modal function component IMF5 obtained from the decomposition of this typical vibration signal are as follows: Figure 5 As shown in (a) to (e). From Figure 5 As can be seen, the curves of each modal function component are smooth, with clear time scale distinctions, and are arranged with frequencies increasing from low to high. This indicates that the signal decomposition method proposed in this scheme can effectively solve the problems of mode aliasing and over-decomposition during signal decomposition. Meanwhile, the center frequencies of IMF4 and IMF5 are relatively high, and their amplitudes are much lower than those of IMF1 to IMF3. Therefore, it can be inferred that IMF4 and IMF5 are highly likely to be mixed-in high-frequency noise signals.

[0038] Furthermore, for this typical vibration signal, the normalized permutation entropy of each modal function component and the cross-correlation coefficient R between each modal function component and the typical vibration signal were calculated, and the results are shown in Table 2.

[0039] Table 2. Normalized permutation entropy and cross-correlation coefficients of each modal function component. It is easy to see from Table 2 that the normalized permutation entropy of IMF4 and IMF5 both exceed 0.8, and the cross-correlation coefficients between IMF4 and IMF5 and the typical vibration signal are both below 0.2. Therefore, it can be determined that IMF4 and IMF5 are noise signals and need to be removed.

[0040] The original signal (this typical vibration signal) was reconstructed using IMF1~IMF3, and the reconstruction result is as follows. Figure 6 As shown. From Figure 6 As can be seen, the reconstructed signal waveform is consistent with the original signal waveform, localized noise is effectively eliminated, and there is no obvious "peak clipping" phenomenon. The reconstructed signal effectively retains the information of the original signal. Therefore, the signal denoising scheme in this paper can accurately remove noise signals to achieve the purpose of noise reduction.

[0041] S2. Perform blasting vibration and transient unloading vibration signal analysis on the vibration signal to achieve separation of blasting vibration signal and transient unloading vibration signal.

[0042] In the process of blasting excavation of deep-buried highway tunnels, the blasting load and the transient unloading load of ground stress are coupled, and their action trajectories partially overlap. Generally, the onset of the transient unloading of ground stress occurs in the latter half of the blasting load's action trajectories. Therefore, the vibration effects induced by the two loads are superimposed, without a clear time-domain boundary. This step achieves the identification and separation of blasting vibration signals and transient unloading vibration signals, laying the foundation for constructing a prediction model of cumulative damage to surrounding rock based on the characteristics of blasting vibration and unloading vibration. Step S2 specifically includes the following steps: S21. Based on the Hilbert transform method, the instantaneous energy of the vibration signal is calculated to identify blasting vibration and transient unloading vibration.

[0043] Specifically, in this embodiment, firstly, the modal function components of the vibration signal are subjected to Hilbert transform to obtain the instantaneous amplitude, instantaneous phase, and instantaneous frequency of each modal function component, thereby constructing the Hilbert spectrum of the vibration signal. Then, the instantaneous energy of the vibration signal is calculated based on the Hilbert spectrum of the vibration signal. The expression for the instantaneous energy is: Where IE is the instantaneous energy of the vibration signal. The instantaneous angular frequency, This is the Hilbert spectrum of the vibration signal.

[0044] Instantaneous energy calculations were performed on the vibration signals to analyze the instantaneous energy characteristics of vibrations induced by the combined effects of blasting load and transient unloading of ground stress. This embodiment uses the vibration waveforms and instantaneous energy corresponding to auxiliary holes 3 and 4 at measuring points AK75+019-2# in the 200m burial depth test area, and AK77+388-2# in the 500m burial depth test area, respectively. The corresponding vibration waveforms are as follows: Figure 7 As shown in (a1) to (d1), the instantaneous energy curves corresponding to the vibration waveform are as follows: Figure 7 As shown in (a2) to (d2). Among them, the burial depth of the tunnel face corresponding to measuring point AK77+388-2# is greater than that corresponding to measuring point AK75+019-2#.

[0045] from Figure 7 It is not difficult to see that for the surrounding rock vibration signal induced by blasting excavation of deep-buried tunnels, both auxiliary holes 3 and 4 contain two distinct peak groups in their instantaneous energy curves. The first peak group occurs within the first ten milliseconds after the borehole detonation, while the second peak group begins at the end of the first peak group and lasts for approximately 10 ms. The duration of the blasting load is in the range of several milliseconds to tens of milliseconds, while the starting time of the transient unloading load of ground stress is in the latter half of the blasting load's action. The start and end times of the two peak groups in the instantaneous energy curves basically correspond to the action processes of the blasting load and the transient unloading load of ground stress. Therefore, it can be concluded that the excitation source of the first peak group is the blasting load, corresponding to blasting vibration, and the excitation source of the second peak group is the transient unloading load, corresponding to transient unloading vibration.

[0046] S22. Based on the vibration identification results, a finite impulse response low-pass digital filter is used to filter the vibration signal, thereby separating the blasting vibration signal from the transient unloading vibration signal.

[0047] Specifically, in this embodiment, based on the identification of the two types of vibration in step S21, the vibration signal needs to be further divided into a blasting vibration signal and a transient unloading vibration signal. This embodiment uses a finite impulse response low-pass digital filter to filter the vibration signal, using the frequency at the boundary between the two dominant frequency bands in the vibration signal spectrum as the filtering threshold. The output signal is the transient unloading vibration signal, and the remaining signal is the blasting vibration signal.

[0048] Figure 7The blasting vibration signal and transient unloading vibration signal, along with the corresponding instantaneous energy curves shown in the image, can be referenced sequentially. Figure 8 Table 3 shows the peak particle velocity (PPV) and dominant frequency (DF) of the blasting vibration signal and the transient unloading vibration signal, from (a1) to (d2). The coupled vibration signal in Table 3 is the vibration signal that has not been split.

[0049] Table 3. Peak particle velocity (PPV) and dominant frequency (DF) for two vibration signals Combination Figure 8 As shown in Table 3, during the blasting excavation of deep-buried tunnels, the transient unloading of ground stress on the excavation surface induces vibration, which increases the overall vibration intensity of the surrounding rock, but slightly decreases the overall dominant vibration frequency (DF). By comparing the PPV values ​​of the blasting vibration signal, the transient unloading vibration signal, and the coupled vibration signal, it can be found that the sum of the PPV values ​​of the separated blasting vibration signal and the transient unloading vibration signal is greater than the PPV value of the coupled vibration signal. This is because the excitation sources of the blasting seismic wave and the transient unloading seismic wave are different, and the time and vibration frequency of the two seismic waves to reach the measuring point are different. This results in the PPV values ​​of the two vibrations not overlapping in the time domain, thus causing a peak misalignment phenomenon.

[0050] Furthermore, due to the greater burial depth of the tunnel face corresponding to measuring point AK77+388-2#, the transient unloading effect of ground stress during blasting excavation is stronger. This results in the PPV value of the transient unloading vibration signal at measuring point AK77+388-2# being significantly greater than that at measuring point AK75+019-2#, and the PPV value of the transient unloading vibration signal at measuring point AK77+388-2# even approaching or exceeding the PPV value of the blasting vibration signal at that point. Therefore, for deep-buried tunnel blasting excavation, transient unloading of ground stress is a crucial factor in generating surrounding rock vibration, and its importance increases with increasing burial depth or ground stress.

[0051] S3. Based on the signal separation results, the characteristic variables and related variables of the cumulative damage of the surrounding rock are determined by the correlation analysis between the vibration response and damage of the surrounding rock.

[0052] Step S3 specifically includes the following steps: S31. Determine the characteristic variable and alternative relevant variable of the cumulative damage of the surrounding rock based on the signal separation result, wherein the characteristic variable is the range of damage to the surrounding rock.

[0053] Specifically, in this embodiment, there is a good correlation between blasting vibration and surrounding rock damage. By establishing the correspondence between blasting vibration and surrounding rock damage, the damage caused by blasting to the surrounding rock can be predicted conveniently and quickly. Since the damage range can well describe the damage status of the surrounding rock, the damage range of the surrounding rock is defined as a characteristic variable.

[0054] Furthermore, the surrounding rock vibration can be characterized by the peak particle velocity (PPV) and dominant frequency (DF). Based on the analysis results of step S2, the surrounding rock vibration induced by blasting excavation of deep-buried tunnels includes both blasting vibration and transient unloading vibration. Therefore, the blasting vibration PPV, blasting vibration dominant frequency (DF), transient unloading vibration PPV, and transient unloading vibration dominant frequency (DF) should all be considered as relevant variables. Simultaneously, when using delayed blasting technology for tunnel excavation, the formation of the surrounding rock damage zone is a cumulative process. All types of borehole blasting contribute to the formation of the final damage zone, but due to the influence of the charge amount and the distance between the borehole and the excavation outline, auxiliary boreholes 3 and 4 have the greatest impact. Therefore, considering only the surrounding rock vibration induced by a single type of borehole blasting or the maximum surrounding rock vibration within the entire blasting cycle is insufficient; it is necessary to consider the blasting vibration PPV and dominant frequency (DF) and transient unloading vibration PPV and dominant frequency (DF) corresponding to auxiliary boreholes 3 and 4. In addition, the burial depth, the distance between measuring points, and the frequency of blasting also affect the damage effect on the surrounding rock. However, transient unloading vibration can already characterize the effect of burial depth on the damage of the surrounding rock, so the burial depth factor will not be considered again.

[0055] In summary, a total of 10 candidate relevant variables and 1 characteristic variable were identified. The candidate relevant variables include the distance between measuring points (m), blasting frequency (times), blasting vibration PPV of auxiliary hole 3, blasting vibration dominant frequency DF of auxiliary hole 3, transient unloading vibration PPV of auxiliary hole 3, transient unloading vibration dominant frequency DF of auxiliary hole 3, blasting vibration PPV of auxiliary hole 4, blasting vibration dominant frequency DF of auxiliary hole 4, transient unloading vibration PPV of auxiliary hole 4, and transient unloading vibration dominant frequency DF of auxiliary hole 4, which are represented by X1 to X10 respectively. The characteristic variable is the surrounding rock damage range (m), represented by X0.

[0056] S32. The candidate relevant variables are screened through correlation analysis to obtain the relevant variables.

[0057] Specifically, step S32 includes the following steps: S321. Collect candidate relevant variable data and corresponding feature variable data, and then calculate the correlation between the candidate relevant variables and the feature variables.

[0058] Specifically, in this embodiment, some of the collected data is shown in Table 4.

[0059] Table 4. Data on some alternative relevant variables and corresponding characteristic variables. Furthermore, grey relational analysis was used to obtain the correlation between each candidate relevant variable and the characteristic variable, and the results are as follows: Figure 9 As shown.

[0060] S324. Set a correlation threshold. When the correlation is greater than the correlation threshold, the corresponding candidate correlation variable is used as the correlation variable.

[0061] Specifically, in this embodiment, the correlation threshold is set to 0.6, then according to Figure 9 It can be seen that the correlation between the dominant frequency (DF) of blasting vibration in auxiliary hole 3 and auxiliary hole 4 and the damage range of the surrounding rock is less than 0.6, indicating that these two related variables have poor correlation with the characteristic variable. Therefore, these two related variables can be disregarded when predicting the cumulative damage of the surrounding rock. Thus, after processing with grey relational analysis, a total of 8 related variables were selected: measuring point distance, blasting frequency, blasting vibration PPV of auxiliary hole 3, transient unloading vibration PPV of auxiliary hole 3, dominant frequency (DF) of transient unloading vibration of auxiliary hole 3, blasting vibration PPV of auxiliary hole 4, transient unloading vibration PPV of auxiliary hole 4, and dominant frequency (DF) of transient unloading vibration of auxiliary hole 4.

[0062] S4. Construct a cumulative damage prediction model for predicting cumulative damage to the surrounding rock of blasting in deep-buried highway tunnels, and input the relevant variables into the cumulative damage prediction model to achieve prediction of the feature variables.

[0063] Step S4 specifically includes the following steps: S41. Construct a dataset for predicting cumulative damage to surrounding rock.

[0064] Specifically, in this embodiment, the data of the main frequency DF of blasting vibration of auxiliary hole 3 and the main frequency DF of blasting vibration of auxiliary hole 4 are removed from the candidate relevant variable data and corresponding feature variable data collected in step S321, and the remaining data are normalized using the max-min normalization method to obtain the surrounding rock cumulative damage prediction dataset.

[0065] S42. Improve the sparrow search algorithm to obtain the improved sparrow search algorithm.

[0066] Specifically, step S42 includes the following steps: S421. Initialize the sparrow population using the dimensionally balanced hierarchical random grid method.

[0067] Because chaotic mappings possess randomness and ergodicity, existing improved sparrow search algorithms typically employ chaotic mappings to initialize the population. However, the performance of chaotic mappings is highly dependent on the choice of parameters; different parameters can significantly alter the properties of the chaotic sequence, such as period length and distribution characteristics. If the parameters are inappropriately chosen, chaotic mappings may fail to generate a uniformly distributed initial population, affecting the algorithm's global search capability. Furthermore, high-dimensional chaotic mappings also have high computational complexity. Therefore, this embodiment employs the following steps to initialize the sparrow population.

[0068] Step S421 specifically includes the following steps: S4211. Determine the search space dimension D and the population size N. For each of the search space dimensions, divide the interval [0,1] into N equally probable sub-intervals.

[0069] Specifically, in this embodiment, the search space dimension D can be calculated using the following formula: Where I, H, and O represent the number of neurons in the input layer, hidden layer, and output layer of the BP neural network, respectively.

[0070] According to the BP neural network settings in step S43, the search space dimension D should be 81. Simultaneously, the population size N should be set to 20. S4212. For each dimension of the search space, a value is randomly selected from each of its sub-intervals to obtain D vectors of length N.

[0071] Specifically, in this embodiment, for any search space dimension j, a value is randomly selected from each of its sub-intervals to obtain a vector of length N, and then... To indicate, Let j be the vector corresponding to the dimension j of the search space. Let be the value selected from the m-th sub-interval of search space dimension j. Furthermore, we can obtain the vector corresponding to each search space dimension, resulting in D vectors of length N.

[0072] S4213. Randomly combine the elements of the D vectors to form N D-dimensional sample sequences.

[0073] Specifically, in this embodiment, for D vectors, a value is randomly selected from each vector to form a D-dimensional sample sequence, and then... To indicate, among which, For the i-th sample sequence, Let be a value randomly selected from the vector corresponding to dimension j of the search space. Repeating the above steps will yield N D-dimensional sample sequences.

[0074] S4214. Linearly map the values ​​in the sample sequence to the actual search space, and use the mapped N sample sequences as the initial population of the sparrow search algorithm.

[0075] Specifically, in this embodiment, the search space is set to [-10, 10], and the following relationship is used to linearly map the values ​​in each sample sequence to the actual search space: in, for The mapping value in the actual search space, This represents the position of the i-th sparrow in dimension j. This is the lower bound of the search space. This represents the upper limit of the search space.

[0076] The sparrow population initialization method in this embodiment divides the search space in each dimension into several equally probable sub-intervals and randomly selects a value within each sub-interval, ensuring the uniformity and diversity of the samples and providing a broader search space for the algorithm. Furthermore, this sparrow population initialization method is simple, requires no adjustment of complex parameters, and can stably generate high-quality initial populations under different problem scenarios, reducing performance fluctuations caused by improper parameter selection.

[0077] S422. Improve the discoverer location update formula by introducing a reference individual and a random perturbation factor.

[0078] Specifically, in this embodiment, the sparrow search algorithm is prone to getting stuck in a local optimum during its iterative process. Therefore, this embodiment increases the randomness and diversity of the search process by randomly selecting reference individuals and introducing random perturbation factors, allowing the discoverer to explore a wider solution space and reducing the possibility of getting stuck in local optima. The position update method is determined based on the individual's fitness value, enabling individuals with poor fitness to move more actively to potentially better areas, while individuals with better fitness can perform refined searches in relatively better areas, improving the efficiency and targeting of the search.

[0079] The improved discoverer location update formula is as follows: in, Let i be the position of the i-th sparrow in the (n+1)th generation in the j-th dimension. Let i be the position of the i-th sparrow in the nth generation in the j-th dimension. For random disturbance factors, , Let be the fitness value of the i-th sparrow. Let be the fitness values ​​of i reference sparrows, Q be a random number following a normal distribution, and L be a 1×d matrix with all elements equal to 1. The warning value is randomly generated. ST is the safety threshold, ST=0.8. Let i be the position of the reference individual of the i-th sparrow in the n-th generation in the j-th dimension. Let be the comparison reference position of the i-th sparrow in the nth generation in the j-th dimension. The comparison reference position can be the current optimal discoverer position.

[0080] S43. The weights and thresholds of the BP neural network are optimized using an improved sparrow search algorithm, and then the cumulative damage prediction model is constructed using the damage prediction dataset and the BP neural network.

[0081] Specifically, in this embodiment, firstly, the mean squared error of the BP neural network is used as the fitness function of the improved sparrow search algorithm, and then the improved sparrow search algorithm is used to optimize the weights and thresholds of the BP neural network. Then, the cumulative damage prediction dataset is divided into a training set, a validation set, and a test set in a 7:2:1 ratio to complete the training, validation, and testing of the BP neural network, ultimately obtaining the cumulative damage prediction model.

[0082] In this embodiment, the BP neural network has 8 nodes in the input layer, 8 nodes in the hidden layer, and 1 node in the output layer. The tansig activation function is used from the input layer to the hidden layer, and the logsig activation function is used from the hidden layer to the output layer.

[0083] S44. Input the relevant variables into the cumulative damage prediction model to predict the feature variables.

[0084] Specifically, in this embodiment, the relevant variables collected in real time are input into the cumulative damage prediction model, which enables the prediction of the corresponding feature variables.

[0085] To verify the performance of the cumulative damage prediction model, blasting tests were conducted in four blasting areas on the main tunnel construction face and the auxiliary construction face of the inclined shaft. The final test results and prediction results are shown in Table 5.

[0086] Table 5 Model Validation Results As can be seen from Table 5, in the nine blasting tests in the four construction areas, the maximum prediction error was 13.59% and the minimum prediction error was 0.89%, which proves that the overall prediction accuracy of this model is good and can improve the intelligence level of prediction of cumulative damage to the surrounding rock.

[0087] It should be noted that in some cases, the actions described in the specification can be performed in different orders and still achieve the desired results. In this embodiment, the order of steps is given only to make the embodiment clearer and easier to explain, and not to limit it.

[0088] In one optional embodiment, please refer to Figure 10 The present invention also provides a system for predicting cumulative damage to blasting surrounding rock in deep-buried highway tunnels. The system includes: a data acquisition device 1, a data output device 2, a processor 3, and a storage device 4. The storage device 4 includes a computer-readable storage medium storing a computer program. The computer program includes program instructions, which, when executed by the processor 3, cause the processor 3 to perform the contents described in steps S1 to S4.

[0089] The present invention has at least the following beneficial effects: 1. This method calculates the difference in the rate of change between the reconstructed signal and the original signal under different modal numbers, and uses the particle swarm optimization algorithm to obtain the maximum signal-to-noise ratio (SNR). This achieves automatic judgment of modal values ​​and penalty factors in variational mode decomposition algorithms, avoiding interference from subjective decisions. It can effectively solve the problems of mode aliasing and over-decomposition during signal decomposition. Combined with multi-scale permutation entropy analysis, it can accurately remove noise signals to achieve the purpose of noise reduction.

[0090] 2. By analyzing the blasting vibration and transient unloading vibration signals, the blasting vibration and transient unloading vibration signals were identified and separated. Then, by analyzing the correlation between the surrounding rock vibration response and damage, the characteristic variables and related variables of the surrounding rock cumulative damage were determined, laying the foundation for constructing a prediction model of the surrounding rock cumulative damage based on the characteristics of blasting vibration and unloading vibration.

[0091] 3. Taking into account the characteristics of blasting vibration and unloading vibration, an improved sparrow search algorithm is used to optimize the weights and thresholds of the BP neural network, thereby constructing a cumulative damage prediction model. This solves the problem that existing technologies have not yet constructed a cumulative damage prediction model for surrounding rock based on the characteristics of blasting vibration and unloading vibration, and improves the accuracy and intelligence of the prediction of cumulative damage in surrounding rock.

[0092] 4. The sparrow population is initialized using a dimensionally balanced hierarchical random grid method, which makes the initial sparrow population more uniform, controllable, and applicable. This can solve the problems of uneven distribution and parameter sensitivity that may occur when chaotic mapping is used to initialize the population. By modifying the discoverer position update formula and the adaptive update safety threshold, the global search capability of the sparrow search algorithm in the early stage and the local search capability in the later stage are increased, which is beneficial to improving the performance of the cumulative damage prediction model.

[0093] 5. A system adapted to this method is provided, which can improve the practicality of this method and facilitate its promotion.

[0094] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered within the scope of the claims and specification of the present invention.

Claims

1. A method for predicting cumulative damage to surrounding rock during blasting in deep-buried highway tunnels, characterized in that, Includes the following steps: The measured vibration time history signal of the buried test area was collected by blasting vibration experiment, and the measured vibration time history signal was denoised to obtain the vibration signal. The vibration signals are analyzed for blasting vibration and transient unloading vibration, thereby achieving the separation of blasting vibration signals and transient unloading vibration signals; Based on the signal separation results, characteristic variables and alternative relevant variables of the cumulative damage to the surrounding rock are determined, wherein the characteristic variable is the range of damage to the surrounding rock. The candidate relevant variables include the distance between measuring points, blasting frequency, blasting vibration PPV of auxiliary hole 3, blasting vibration main frequency DF of auxiliary hole 3, transient unloading vibration PPV of auxiliary hole 3, transient unloading vibration main frequency DF of auxiliary hole 3, blasting vibration PPV of auxiliary hole 4, blasting vibration main frequency DF of auxiliary hole 4, transient unloading vibration PPV of auxiliary hole 4, and transient unloading vibration main frequency DF of auxiliary hole 4; Collect candidate relevant variable data and corresponding feature variable data, and then calculate the correlation between the candidate relevant variables and the feature variables; Set a correlation threshold. When the correlation is greater than the correlation threshold, the corresponding candidate correlation variables are used as correlation variables. The relevant variables include the distance between measuring points, blasting frequency, blasting vibration PPV of auxiliary hole 3, transient unloading vibration PPV of auxiliary hole 3, transient unloading vibration main frequency DF of auxiliary hole 3, blasting vibration PPV of auxiliary hole 4, transient unloading vibration PPV of auxiliary hole 4, and transient unloading vibration main frequency DF of auxiliary hole 4. A cumulative damage prediction model is constructed for predicting the cumulative damage of blasting surrounding rock in deep-buried highway tunnels. The relevant variables are input into the cumulative damage prediction model to achieve the prediction of the feature variables.

2. The method for predicting cumulative damage to surrounding rock during blasting in a deep-buried highway tunnel according to claim 1, characterized in that, The process of acquiring measured vibration time history signals from the burial depth test area through blasting vibration experiments and then performing noise reduction processing on the measured vibration time history signals to obtain vibration signals includes the following steps: Measured vibration time history signals of the buried test area were collected through blasting vibration experiments. The measured vibration time history signal was decomposed into multiple modal function components using a variational mode decomposition algorithm; Multi-scale permutation entropy analysis is performed on the modal function components to identify and eliminate noise components, thereby obtaining the vibration signal.

3. The method for predicting cumulative damage to surrounding rock during blasting in a deep-buried highway tunnel according to claim 2, characterized in that: When using the variational mode decomposition algorithm, if the difference between the signal energy change rate of the mode number K+1 and the signal energy change rate of the mode number K exceeds a preset threshold, K is taken as the optimal number of modes. When using the variational mode decomposition algorithm, the optimal penalty factor is obtained by using the particle swarm optimization algorithm with the aim of maximizing the signal-to-noise ratio of the reconstructed signal.

4. The method for predicting cumulative damage to surrounding rock during blasting in deep-buried highway tunnels according to claim 1, characterized in that, The step of analyzing the vibration signals to separate the blasting vibration signals from the transient unloading vibration signals includes the following steps: Based on the Hilbert transform method, the instantaneous energy of the vibration signal is calculated to identify blasting vibration and transient unloading vibration. Based on the vibration identification results, a finite impulse response low-pass digital filter is used to filter the vibration signal, thereby separating the blasting vibration signal from the transient unloading vibration signal.

5. The method for predicting cumulative damage to surrounding rock during blasting in deep-buried highway tunnels according to claim 1, characterized in that, The construction of a cumulative damage prediction model for predicting cumulative damage to surrounding rock in deep-buried highway tunnel blasting involves inputting the relevant variables into the cumulative damage prediction model to predict the feature variables, and includes the following steps: Construct a dataset for predicting cumulative damage to surrounding rock; The sparrow search algorithm was improved to obtain the improved sparrow search algorithm; An improved sparrow search algorithm is used to optimize the weights and thresholds of the BP neural network, and then the cumulative damage prediction model is constructed using the damage prediction dataset and the BP neural network. The relevant variables are input into the cumulative damage prediction model to predict the feature variables.

6. The method for predicting cumulative damage to surrounding rock during blasting in a deep-buried highway tunnel according to claim 5, characterized in that, The improvement of the sparrow search algorithm, resulting in the improved sparrow search algorithm, includes the following steps: The sparrow population was initialized using a dimensionally balanced hierarchical random grid method. The discoverer location update formula is improved by introducing a reference individual and a random perturbation factor.

7. The method for predicting cumulative damage to surrounding rock during blasting in a deep-buried highway tunnel according to claim 6, characterized in that, The method of initializing the sparrow population using a dimensionally balanced hierarchical random grid includes the following steps: Determine the search space dimension D and the population size N. For each of the search space dimensions, divide the interval [0,1] into N equally probable subintervals. For each dimension of the search space, a value is randomly selected from each of its sub-intervals to obtain D vectors of length N; The elements of the D vectors are randomly combined to form N D-dimensional sample sequences; The numerical values ​​in the sample sequence are linearly mapped to the actual search space, and the N mapped sample sequences are used as the initial population of the sparrow search algorithm.

8. A system for predicting cumulative damage to surrounding rock during blasting in deep-buried highway tunnels, characterized in that, The aforementioned system for predicting cumulative damage to surrounding rock in blasting of deep-buried highway tunnels includes: a data acquisition device, a data output device, a processor, and a storage device. The storage device includes a computer-readable storage medium storing a computer program. The computer program includes program instructions, which, when executed by the processor, cause the processor to implement the method for predicting cumulative damage to surrounding rock in blasting of deep-buried highway tunnels as described in any one of claims 1-7.

Images