A surrounding rock dynamic disturbance correlation prediction method and system based on TBM tunneling parameters

Abstract

The application provides a kind of surrounding rock dynamic disturbance correlation prediction method and system based on TBM driving parameter, relating to the technical field of tunnel engineering, including synchronously obtaining surrounding rock original vibration signal and TBM real-time driving parameter;The vibration signal is analyzed in time domain and frequency domain to extract characteristic parameters representing disturbance intensity;Align the time and match the spatial position of the two types of data, and construct a one-to-one corresponding multi-source sample data set;Through correlation analysis and significance test, a key parameter set with a dominant role in disturbance is selected from the driving parameters;Based on the key parameters and disturbance characteristic parameters, a quantitative prediction model is constructed to capture the nonlinear relationship between them.The application realizes real-time and quantitative prediction from construction parameters to surrounding rock dynamic disturbance intensity.

Description

Technical Field

[0001] This invention relates to the field of tunnel engineering technology, and more specifically, to a method and system for predicting the correlation of surrounding rock dynamic disturbances based on TBM tunneling parameters. Background Technology

[0002] As tunnel construction continues to extend deeper, deep rock masses generally exhibit typical engineering characteristics such as high ground stress, high rock mass strength, high energy storage, and strong disturbance, leading to frequent engineering disasters such as rock bursts. This has become a key engineering problem restricting the safe construction of deep-buried tunnels. Traditional drill-and-blast methods are difficult to meet the requirements of scale and efficiency in current tunnel construction, and full-face tunnel boring machine (TBM) excavation has become an important construction method for deep-buried tunnels.

[0003] In recent years, with the widespread application of TBMs in deep-buried tunnels, the dynamic disturbance problem during TBM construction has gradually attracted attention. TBM excavation activities, such as rock breaking and the interaction between the shield and support shoes and the surrounding rock, generate continuous vibrations that propagate through the surrounding rock, forming a dynamic stress wave field. In deep-buried, high-stress environments, the tangential stress increases significantly and the radial stress decreases markedly after excavation, resulting in a continuously increasing stress difference, often placing the surrounding rock in a near-critical state of failure. Under these conditions, the dynamic disturbances generated by TBM construction can couple with the geological conditions conducive to rockburst, thereby inducing surrounding rock fracturing and triggering rockburst disasters.

[0004] Field investigation results indicate that dynamic disturbance-triggered rockbursts are mostly of medium to strong intensity, posing a significant threat to personnel and equipment at tunnel construction sites. Existing research on the dynamic disturbance characteristics of TBMs primarily focuses on shallow-buried tunnels; a systematic study of the dynamic disturbance effects during TBM construction in deep-buried tunnels is lacking, and a quantitative relationship between TBM tunneling parameters and surrounding rock dynamic disturbance has not been established. This makes it difficult to obtain surrounding rock dynamic disturbance parameters in real time and to assess the immediate impact of dynamic disturbance factors on rockburst induction. Furthermore, traditional surrounding rock vibration monitoring relies mainly on post-event signal analysis, lacking immediacy and predictability, and thus failing to meet the engineering requirements for real-time dynamic disturbance rockburst early warning in deep-buried tunnels.

[0005] Therefore, existing technologies still have significant shortcomings in the analysis of dynamic disturbance characteristics, parameter correlation modeling, and real-time quantitative prediction during the construction of deep-buried TBM tunnels. There is an urgent need to establish a correlation prediction method that combines TBM tunneling parameters and dynamic disturbance effects to provide data support for early warning of rock mass fracture triggered by dynamic disturbances in deep-buried TBM tunnels. Summary of the Invention

[0006] The purpose of this invention is to provide a method and system for predicting the dynamic disturbance of surrounding rock based on TBM tunneling parameters, in order to improve the aforementioned problems. To achieve the above objective, the technical solution adopted by this invention is as follows:

[0007] Firstly, this application provides a method for predicting the correlation of surrounding rock dynamic disturbances based on TBM tunneling parameters, including:

[0008] Acquire the original vibration signals of the surrounding rock dynamic disturbance and the real-time tunneling parameters during TBM construction;

[0009] Feature extraction processing is performed based on the original vibration signal, wherein by analyzing the time domain and frequency domain characteristics of the signal, the surrounding rock dynamic disturbance characteristic parameters used to characterize the disturbance intensity are obtained.

[0010] Based on the real-time tunneling parameters and the surrounding rock dynamic disturbance feature parameters, a synchronous matching process is performed, wherein the real-time tunneling parameters and the surrounding rock dynamic disturbance feature parameters are time-aligned and spatially matched to obtain a one-to-one corresponding multi-source sample dataset.

[0011] Based on the multi-source sample dataset, correlation screening is performed. By calculating the correlation degree between each real-time tunneling parameter and the surrounding rock dynamic disturbance characteristic parameter, and performing a significance test, the set of key tunneling parameters that have a dominant control effect on the surrounding rock dynamic disturbance and their corresponding surrounding rock dynamic disturbance characteristic parameters are obtained.

[0012] Based on the set of key tunneling parameters and their corresponding surrounding rock dynamic disturbance characteristic parameters, a mapping relationship is constructed. Specifically, by capturing the nonlinear relationship between the set of key tunneling parameters and their corresponding surrounding rock dynamic disturbance characteristic parameters, a quantitative correlation prediction model for surrounding rock dynamic disturbance is constructed, and the intensity of surrounding rock dynamic disturbance is predicted according to the real-time key tunneling parameters.

[0013] Secondly, this application also provides a system for predicting the correlation of surrounding rock dynamic disturbance based on TBM tunneling parameters, including:

[0014] The acquisition unit is used to acquire the original vibration signal of the surrounding rock dynamic disturbance and the real-time tunneling parameters during TBM construction;

[0015] The processing unit is used to perform feature extraction processing based on the original vibration signal, wherein by analyzing the time domain and frequency domain characteristics of the signal, the surrounding rock dynamic disturbance characteristic parameters used to characterize the disturbance intensity are obtained.

[0016] The matching unit is used to perform synchronous matching processing based on the real-time tunneling parameters and the surrounding rock dynamic disturbance feature parameters. In this process, by performing time alignment and spatial location matching on the real-time tunneling parameters and the surrounding rock dynamic disturbance feature parameters, a one-to-one corresponding multi-source sample dataset is obtained.

[0017] The filtering unit is used to perform correlation filtering based on the multi-source sample dataset. By calculating the correlation degree between each real-time tunneling parameter and the surrounding rock dynamic disturbance characteristic parameter and performing a significance test, the set of key tunneling parameters that have a dominant control effect on the surrounding rock dynamic disturbance and their corresponding surrounding rock dynamic disturbance characteristic parameters are obtained.

[0018] The prediction unit is used to construct a mapping relationship based on the set of key tunneling parameters and their corresponding surrounding rock dynamic disturbance characteristic parameters. Specifically, by capturing the nonlinear relationship between the set of key tunneling parameters and their corresponding surrounding rock dynamic disturbance characteristic parameters, a quantitative correlation prediction model for surrounding rock dynamic disturbance is constructed, and the intensity of surrounding rock dynamic disturbance is predicted according to the real-time key tunneling parameters.

[0019] The beneficial effects of this invention are as follows:

[0020] This invention establishes a quantitative mapping relationship between TBM tunneling parameters and surrounding rock dynamic disturbance characteristic parameters, realizing the transformation of surrounding rock dynamic disturbance from ex-post monitoring to real-time prediction; and improves the accuracy and stability of surrounding rock dynamic disturbance prediction through vibration monitoring, spatiotemporal synchronization, correlation screening and machine learning modeling; this invention can also provide a basis for real-time early warning and construction parameter control of dynamic disturbance-triggered rockburst in deep buried high ground stress environment.

[0021] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing embodiments of the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description, claims, and drawings. Attached Figure Description

[0022] To more clearly illustrate the technical solutions of the embodiments of the present invention, 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 the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a schematic diagram of the process for predicting the correlation of surrounding rock dynamic disturbance based on TBM tunneling parameters as described in this embodiment of the invention.

[0024] Figure 2 This is a schematic diagram of the surrounding rock dynamic disturbance correlation prediction system based on TBM tunneling parameters as described in an embodiment of the present invention.

[0025] In the diagram: 701, acquisition unit; 702, processing unit; 703, matching unit; 704, filtering unit; 705, prediction unit. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0027] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this invention, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0028] Example 1:

[0029] This embodiment provides a method for predicting the dynamic disturbance of surrounding rock based on TBM tunneling parameters.

[0030] See Figure 1 The figure shows that the method includes steps S1, S2, S3, S4 and S5.

[0031] Step S1: Obtain the original vibration signal of the surrounding rock dynamic disturbance and the real-time tunneling parameters during TBM construction;

[0032] It is understandable that in the TBM construction scenario of deep-buried tunnels, the response of the surrounding rock to construction dynamic disturbances is implicit and difficult to quantify directly, while the TBM's operating parameters are explicit and can be continuously recorded. Therefore, this step uses high-precision vibration sensors deployed at the locations corresponding to the shield or support shoes in the surrounding rock to directly capture the original vibration waveforms induced by construction and propagating in the rock mass, thereby obtaining first-hand data on the surrounding rock response. Simultaneously, key construction parameters, including total thrust, cutterhead torque, rotational speed, penetration depth, tunneling rate, and shield and support shoe pressure, are acquired in real time through the TBM's onboard data acquisition system. Specifically, high-precision vibration sensors are deployed at different distances from the tunnel face within the TBM construction area. Preferably, vibration sensors are deployed at intervals along the tunnel axis, with intervals ranging from 3m to 10m; the deployment is appropriately denser near the shield and in the support shoe's action zone. Preferably, the vibration sensors are triaxial accelerometers, which collect the surrounding rock vibration responses along the tunnel axis, radial direction, and tangential direction, respectively. Vibration sensors are fixed to the surrounding rock surface using expansion bolts, anchors, or pre-embedded mounting components. When the surrounding rock surface is relatively fractured, shallow holes can be used for installation, supplemented with coupling materials for fixation, to improve sampling stability. All sensors are connected to a unified data acquisition terminal and synchronized with time via a unified timing module.

[0033] Step S2: Perform feature extraction processing based on the original vibration signal, wherein by analyzing the time domain and frequency domain characteristics of the signal, the surrounding rock dynamic disturbance characteristic parameters used to characterize the disturbance intensity are obtained;

[0034] Understandably, this step aims to "decode" the original signal, extracting parameters such as peak value and root mean square value that directly reflect vibration energy and amplitude through time-domain analysis, and identifying the main frequency band and spectral structure where vibration energy is concentrated through frequency-domain analysis. This process transforms the continuous, mixed vibration waveform into a series of discrete characteristic indicators with clear physical meaning, thereby elevating the "signal" into "information" that can be used for subsequent quantitative correlation analysis. In this step, step S2 includes steps S21, S22, and S23.

[0035] Step S21: Perform signal preprocessing based on the original vibration signal, wherein the preprocessed vibration signal is obtained by filtering out interference components introduced by the environment and equipment in the signal;

[0036] It is understood that this invention preprocesses the original vibration signal because it may be affected by electromagnetic interference, mechanical collisions, environmental noise, and baseline drift. The preprocessing steps in this invention include: processing the original signal using bandpass filtering. Preferably, the bandpass filter band is set according to the signal frequency distribution to remove excessively low-frequency drift and excessively high-frequency random noise. Baseline correction or trend term removal is then performed on the denoised vibration signal to make the signal fluctuate around the zero mean. Let the signal before detrending be... The trend item is Then the trend-following signal for:

[0037] ;

[0038] in, Obtained through linear fitting, polynomial fitting, or moving average.

[0039] Preprocessing also includes identifying and removing abrupt changes and outliers in the signal. Preferably, 3... σ Criterion, that is, for the mean of a signal sequence and standard deviation , Given a signal sequence, when a certain sample value satisfies:

[0040] ;

[0041] If it is an outlier, it will be replaced by nearest neighbor interpolation or local mean.

[0042] The preprocessing also includes dividing the continuous vibration signal into preset time windows, with each time window serving as a sample unit. The length of the time window is preferably 1 s to 60 s, more preferably 5 s to 20 s. A fixed window or a sliding window can be used. When using a sliding window, the window overlap rate is preferably 20% to 80%.

[0043] Step S22: Perform time-domain feature analysis processing on the preprocessed vibration signal, wherein, by calculating the peak value and overall level statistics of the signal in the time dimension, time-domain feature parameters including peak acceleration, peak vibration velocity and vibration amplitude are obtained.

[0044] Understandably, this step involves performing time-domain analysis on the preprocessed effective vibration signal to extract the characteristic parameters of the surrounding rock dynamic disturbance.

[0045] The formula for extracting peak acceleration is shown below:

[0046] Peak acceleration within a certain time window Defined as the maximum absolute value of the vibration acceleration signal:

[0047] ;

[0048] in, For the first time window Acceleration values ​​at each sampling point This represents the maximum value.

[0049] The formula for extracting peak vibration velocity is shown below:

[0050] Peak vibration velocity within a certain time window Defined as the maximum absolute value of the vibration velocity signal:

[0051] ;

[0052] in, For the first time window The velocity value of each sampling point.

[0053] The formula for extracting the vibration amplitude is shown below:

[0054] The vibration amplitude can be expressed as either peak-to-peak value or root-mean-square value. Root-mean-square value is preferred. The overall vibration level is characterized by the following formula:

[0055] ;

[0056] in, This represents the number of vibration signals.

[0057] Step S23: Perform frequency domain feature analysis processing on the preprocessed vibration signal, wherein the frequency domain feature parameters, including the dominant frequency and spectral distribution characteristics, are obtained by converting the signal to the frequency domain and identifying the frequency band where its energy is concentrated.

[0058] It is understandable that this step involves performing a Fast Fourier Transform on the preprocessed vibration signal to obtain the spectrum. The calculation formula is as follows:

[0059] ;

[0060] in, The time-domain sampling point number, For the complex exponential basis functions of the Fourier transform, This represents the vibration signal amplitude at the k-th sampling point within the time window.

[0061] Among them, the frequency with the largest spectral amplitude is taken as the dominant frequency. :

[0062] ;

[0063] The spectral distribution characteristics are represented by the frequency band energy proportion, for example, a target frequency band. The proportion of internal energy to total energy:

[0064] ;

[0065] in, Target frequency band The proportion of internal energy to total energy The highest effective frequency used in spectrum analysis is usually the Nyquist frequency, i.e. ,in The sampling frequency of the vibration signal. For frequency f The power spectral density or spectral energy value at that location. For frequency variables.

[0066] This step also involves analysis by calculating the vibration duration, where the vibration duration... Defined as the vibration amplitude or vibration energy exceeding a preset threshold within a time window. Cumulative time:

[0067] ;

[0068] Among them, when hour, ,otherwise ; This represents the sampling time interval.

[0069] In this embodiment, peak acceleration and peak vibration velocity are preferably used as the core dynamic disturbance output parameters.

[0070] Step S3: Perform synchronous matching processing based on the real-time tunneling parameters and the surrounding rock dynamic disturbance feature parameters. Specifically, by performing time alignment and spatial location matching on the real-time tunneling parameters and the surrounding rock dynamic disturbance feature parameters, a one-to-one corresponding multi-source sample dataset is obtained.

[0071] Understandably, this step ensures that the analyzed tunneling parameters and their actual induced surrounding rock vibration response belong to the same time period through time alignment, and that the analyzed vibration signal originates from the surrounding rock in the area directly affected by the current TBM construction location (such as the cutterhead or specific support shoe) through spatial location matching. This step is a crucial prerequisite for fusing two types of heterogeneous data into high-quality "sample pairs" that can be used for machine learning training, aiming to overcome the correlation fallacy caused by the "spatiotemporal asynchrony" of multi-source data in the field. In this step, step S3 includes steps S31, S32, and S33.

[0072] Step S31: Perform parameter statistical processing based on the real-time tunneling parameters, wherein statistical characteristic values ​​representing the tunneling status during the time period are obtained by performing statistical processing on continuously changing parameter values ​​within a preset time window;

[0073] It is understood that this step involves acquiring tunneling parameters in real time through the TBM's onboard data acquisition system. These tunneling parameters include at least: total thrust, cutterhead torque, cutterhead rotation speed, cutter penetration, tunneling rate, shield contact pressure, and support shoe rock pressure.

[0074] Among them, total thrust reflects the overall performance of the TBM propulsion system; cutterhead torque reflects the change in cutterhead cutting resistance; cutterhead rotation speed and penetration reflect the rock-breaking conditions of the cutterhead; and shield contact pressure and support shoe surrounding rock pressure reflect the interaction state between the TBM structure and the surrounding rock.

[0075] Step S32: Perform time reference alignment processing based on the statistical feature values ​​and the surrounding rock dynamic disturbance feature parameters, wherein time-synchronized sample pairs are obtained by mapping data from different acquisition systems to a unified analysis time window;

[0076] It is understood that this step unifies the time synchronization between the vibration monitoring system and the TBM airborne acquisition system, using the same timestamp as the synchronization benchmark. For data sources with different sampling frequencies, the raw sampling data is mapped to the same analysis time window. Preferably, the TBM parameters are statistically summarized using the mean, maximum, minimum, or standard deviation within the window.

[0077] Among them, let the first The sequence of tunneling parameters within a certain time window is as follows: Then its window mean for:

[0078] ;

[0079] in, For the first The window mean of the tunneling parameter sequence within a time window. Indicates that there is One tunneling parameter, Indicates the first One tunneling parameter.

[0080] Step S33: Perform spatial relationship matching processing on the time-synchronized sample pairs, wherein the relative positional relationship between vibration monitoring points and construction disturbance sources is established to obtain a multi-source sample dataset corresponding to the spatial location.

[0081] Understandably, this step establishes a correspondence between monitoring points and construction disturbance sources based on the relative spatial relationship between the vibration sensor's location at the cross-section and the locations of the TBM cutterhead, shield, or support shoe. Preferably, the relative distance of the vibration sensor from the tunnel face, its axial distance from the cutterhead center, or its relative position from the shield / support shoe's action zone can be used as spatial matching features to participate in sample construction. After this step, the statistical values ​​of tunneling parameters within each time window are used as input features, and the surrounding rock dynamic disturbance feature parameters extracted within the same time window are used as output labels to form a single sample, thereby forming a complete dataset.

[0082] Step S4: Based on the multi-source sample dataset, perform correlation screening processing. Calculate the correlation degree between each real-time tunneling parameter and the surrounding rock dynamic disturbance characteristic parameters, and perform a significance test to obtain the set of key tunneling parameters that have a dominant control effect on the surrounding rock dynamic disturbance and their corresponding surrounding rock dynamic disturbance characteristic parameters.

[0083] Understandably, this step employs a combination of correlation analysis and statistical testing to quantitatively identify the key parameter set that exhibits significant statistical correlation with surrounding rock vibration characteristic parameters (such as peak velocity and acceleration) from all available tunneling parameters. This process represents a dimensionality reduction and focus from "all parameters" to "master parameters." The idea is to eliminate irrelevant or weakly correlated variables based on the patterns revealed by the data itself, thereby extracting the core construction factors that truly drive the dynamic disturbance of the surrounding rock, laying the foundation for building a concise and robust prediction model. In this step, step S4 includes steps S41, S42, and S43.

[0084] Step S41: Perform data scaling unification processing on the multi-source sample dataset, wherein the normalized dataset is obtained by adjusting the tunneling parameters of different dimensions to the same numerical range.

[0085] It is understandable that this step involves normalizing the input data to reduce the impact of these dimensional differences on model training, given the different dimensions of various tunneling parameters. The min-max normalization method is preferred. Let the original value of a certain input parameter be... Its minimum and maximum values ​​are respectively and The normalization result for:

[0086] ;

[0087] Standardization can also be used, that is:

[0088] ;

[0089] in, For the first The original tunneling parameter values ​​corresponding to each sample; For the first Standardized tunneling parameter values ​​corresponding to each sample; This is the mean of the tunneling parameter in the sample set; This represents the standard deviation of the tunneling parameter in the sample set.

[0090] Step S42: Perform nonparametric correlation analysis on the normalized dataset, wherein the correlation coefficient reflecting the degree of monotonic correlation between the two is obtained by calculating the rank correlation between each tunneling parameter sequence and the disturbance characteristic parameter sequence.

[0091] It is understandable that Spearman rank correlation analysis is performed between each tunneling parameter and the output variable in order to screen the key tunneling parameters that have a significant impact on the dynamic disturbance of the surrounding rock.

[0092] Let the rank difference between the two variables be... The sample size is Then the Spearman rank correlation coefficient for:

[0093] ;

[0094] Specifically, a correlation coefficient matrix is ​​formed by calculating the correlation coefficients between each input parameter and the peak vibration velocity and peak acceleration. The larger the absolute value of the correlation coefficient, the stronger the correlation between the corresponding tunneling parameter and the dynamic disturbance of the surrounding rock.

[0095] Step S43: Perform statistical significance discrimination processing based on the correlation coefficient. In this process, by examining the probability that the correlation result is caused by accidental factors, select tunneling parameters with significant correlation that exceed a preset threshold to obtain a set of key tunneling parameters.

[0096] It is understandable that this step is to exclude accidental related parameters and to perform Spearman correlation analysis on the results. Significance test of the value. Preferably, the significance threshold is set as follows:

[0097] ;

[0098] When a certain tunneling parameter and output variable satisfy the following relationship... and At that time, it was determined to be a key tunneling parameter; among which It can be set according to project requirements, preferably 0.2 to 0.5.

[0099] After screening, a set of key input features is obtained. For example, in this embodiment, the key parameters obtained by screening may be several of the following: total thrust, cutterhead torque, tunneling rate, cutter penetration, and rock pressure of the support shoe.

[0100] Step S5: Based on the set of key tunneling parameters and their corresponding surrounding rock dynamic disturbance characteristic parameters, a mapping relationship is constructed. Specifically, by capturing the nonlinear relationship between the set of key tunneling parameters and their corresponding surrounding rock dynamic disturbance characteristic parameters, a quantitative correlation prediction model for surrounding rock dynamic disturbance is constructed, and the intensity of surrounding rock dynamic disturbance is predicted according to the real-time key tunneling parameters.

[0101] Understandably, this step, based on the set of key tunneling parameters selected in step S4 and their corresponding historical disturbance characteristic parameters, uses a data-driven approach to automatically learn and construct a high-order mapping relationship between the two. By training this predictive model capable of capturing complex nonlinear patterns, the aim is to achieve quantitative and high-precision prediction of the intensity of dynamic disturbances in the surrounding rock. This step represents a paradigm shift from "qualitative experience-based judgment" to "quantitative dynamic prediction." Its core value lies in its ability to instantly calculate the level of surrounding rock disturbance that may be induced by current construction activities based on real-time acquired key tunneling parameters, thereby providing direct evidence for forward-looking risk warnings. In this step, step S5 includes steps S51, S52, and S53.

[0102] Step S51: Based on the set of key tunneling parameters and their corresponding surrounding rock dynamic disturbance characteristic parameters, perform multiple prediction relationship construction processes, wherein multiple candidate prediction relationships are obtained by initializing multiple data fitting relationships with different structures.

[0103] It is understandable that this step uses the key tunneling parameters obtained from the screening as input variables and peak vibration velocity and peak acceleration as output variables to establish a regression prediction model.

[0104] The sample dataset is divided into training, validation, and test sets. Preferably, the training set accounts for 60%–80%, the validation set for 10%–20%, and the test set for 10%–20%. When the sample size is small, only training and test sets may be used, and cross-validation can be employed to select the model. The candidate models used include: Support Vector Machine regression, Random Forest regression, Decision Tree regression, Backpropagation (BP) neural network regression, and XGBoost regression.

[0105] Each candidate model is trained using the training set and its prediction performance is evaluated on the validation or test set.

[0106] If only the peak vibration velocity is predicted, the model is expressed as:

[0107] ;

[0108] in, This is the formula for predicting peak vibration velocity.

[0109] If only peak acceleration is predicted, the model is expressed as:

[0110] ;

[0111] in, This is the formula for predicting peak acceleration.

[0112] If dual-output prediction is performed simultaneously, the model is expressed as:

[0113] ;

[0114] in, This is the prediction formula for peak acceleration and peak vibration velocity.

[0115] Step S52: Evaluate and screen the prediction performance of the multiple candidate prediction relationships, wherein the initial prediction relationship with the best prediction performance is obtained by comparing the prediction error and goodness of fit of each relationship on independent data.

[0116] It is understandable that in order to compare the predictive capabilities of different models, this embodiment uses mean square error, root mean square error, mean absolute error, and coefficient of determination as evaluation indicators.

[0117] Mean square error for:

[0118] ;

[0119] in, The mean square error of the prediction result. For the first Measured values ​​of the characteristic parameters of surrounding rock dynamic disturbance corresponding to each test sample; For the first Predicted values ​​of the characteristic parameters of surrounding rock dynamic disturbance corresponding to each test sample; The total number of test samples used in model evaluation; This is the test sample number.

[0120] Root mean square error for:

[0121] ;

[0122] Mean Absolute Error for:

[0123] ;

[0124] Coefficient of determination for:

[0125] ;

[0126] in, The coefficient of determination for the prediction result. This represents the average value of the measured values ​​of the characteristic parameters of the surrounding rock dynamic disturbance in the test sample.

[0127] Preferably, with Minimum Minimum Minimum The maximum value is used as the criterion for determining the optimal model.

[0128] Step S53: Perform internal parameter optimization processing based on the initial prediction relationship, wherein the internal parameters of the initial prediction relationship are iteratively adjusted by the particle swarm optimization algorithm to minimize its prediction error, and the final optimized quantitative correlation prediction model of surrounding rock dynamic disturbance is obtained.

[0129] It is understandable that this step involves using the particle swarm optimization algorithm to optimize the hyperparameters of the best candidate model in order to further improve the model's prediction accuracy.

[0130] In this process, the hyperparameter combination of the model to be optimized is encoded into a particle position vector. Taking support vector machines as an example, the penalty coefficients can be... and kernel function parameters The encoding is:

[0131] ;

[0132] in, For the support vector machine regression model hyperparameter combination corresponding to the i-th particle;

[0133] Taking random forest as an example, parameters such as the number of trees and maximum depth can be encoded as:

[0134] ;

[0135] in, The number of random forest trees corresponding to the i-th particle; The maximum depth of a single decision tree corresponding to the i-th particle; This represents the minimum number of samples for the leaf node corresponding to the i-th particle.

[0136] The model prediction error is used as the fitness function. Preferably, the root mean square error on the validation set is used as the fitness function.

[0137] ;

[0138] in, Let be the fitness function, and let be the objective function to minimize the fitness.

[0139] The particle velocity and position are iteratively updated using the following formula:

[0140] ; ;

[0141] in, For the first The particle in the first The velocity vector at the next iteration For the first The particle in the first The position vector at the next iteration. For the first The particle in the first The position vector at the next iteration. For inertial weights, and As a learning factor, and A random number between 0 and 1 This represents the optimal position for an individual particle. This is the optimal position for the group.

[0142] Preferably, inertial weight The learning factor is set between 0.4 and 0.9. and Use a value of 1.0 to 2.5, a particle number of 10 to 100, and an iteration count of 20 to 200.

[0143] After optimization, the optimal hyperparameter combination is substituted into the corresponding model to obtain the final prediction model of surrounding rock dynamic disturbance.

[0144] After step S53, the system further includes statistically processing the real-time parameters according to the same time window as the training phase, normalizing them in the same way, inputting them into the trained prediction model, and outputting the predicted peak vibration velocity and peak acceleration values ​​within the current time window.

[0145] Then, based on historical measured data, field statistical patterns, surrounding rock grade, and rockburst tendency analysis results, a warning threshold for dynamic disturbance of the surrounding rock is set. A single-level threshold or multiple-level thresholds can be set.

[0146] For example, a security threshold can be set. Warning threshold and danger threshold ;

[0147] And satisfy:

[0148] ;

[0149] When the predicted value falls within different threshold ranges, different levels of early warning information are output.

[0150] Taking peak vibration velocity as an example, if the following conditions are met:

[0151] ;

[0152] It is then determined to be a low-risk state;

[0153] If the following conditions are met:

[0154] ;

[0155] Then it is determined to be in a state of attention;

[0156] If the following conditions are met:

[0157] ;

[0158] Then it is determined to be a warning state;

[0159] If the following conditions are met:

[0160] ;

[0161] This indicates a high-risk situation and triggers a dynamic disturbance-type rockburst warning.

[0162] Similarly, a threshold determination can be made for peak acceleration; a dual-parameter joint threshold determination method can also be used, that is, to consider whether the peak vibration velocity and peak acceleration exceed the limits at the same time.

[0163] After the system issues an early warning, it can prompt construction personnel to take risk control measures, including but not limited to: reducing the total thrust of the TBM, adjusting the cutterhead speed, reducing the cutter penetration, reducing the tunneling rate, optimizing the support shoe loading status, and strengthening local support and monitoring. These measures can reduce the intensity of dynamic disturbance during construction and decrease the risk of surrounding rock fracturing and rockburst triggering.

[0164] Example 2:

[0165] like Figure 2 As shown, this embodiment provides a system for predicting the correlation of surrounding rock dynamic disturbances based on TBM tunneling parameters. See [link to documentation]. Figure 2 The system includes an acquisition unit 701, a processing unit 702, a matching unit 703, a filtering unit 704, and a prediction unit 705.

[0166] The acquisition unit 701 is used to acquire the original vibration signal of the surrounding rock dynamic disturbance and the real-time tunneling parameters during TBM construction;

[0167] The processing unit 702 is used to perform feature extraction processing based on the original vibration signal, wherein by analyzing the time domain and frequency domain characteristics of the signal, the surrounding rock dynamic disturbance characteristic parameters used to characterize the disturbance intensity are obtained.

[0168] The matching unit 703 is used to perform synchronous matching processing based on the real-time tunneling parameters and the surrounding rock dynamic disturbance feature parameters. In this process, by performing time alignment and spatial location matching on the real-time tunneling parameters and the surrounding rock dynamic disturbance feature parameters, a one-to-one corresponding multi-source sample dataset is obtained.

[0169] The filtering unit 704 is used to perform correlation filtering processing based on the multi-source sample dataset. By calculating the correlation degree between each real-time tunneling parameter and the surrounding rock dynamic disturbance characteristic parameter and performing a significance test, the set of key tunneling parameters that have a dominant control effect on the surrounding rock dynamic disturbance and their corresponding surrounding rock dynamic disturbance characteristic parameters are obtained.

[0170] The prediction unit 705 is used to construct a mapping relationship based on the set of key tunneling parameters and their corresponding surrounding rock dynamic disturbance characteristic parameters. Specifically, by capturing the nonlinear relationship between the set of key tunneling parameters and their corresponding surrounding rock dynamic disturbance characteristic parameters, a quantitative correlation prediction model for surrounding rock dynamic disturbance is constructed, and the intensity of the surrounding rock dynamic disturbance corresponding to the key tunneling parameters is predicted based on the real-time key tunneling parameters.

[0171] It should be noted that the specific methods by which each module performs operations in the system described in the above embodiments have been described in detail in the embodiments related to the method, and will not be elaborated here.

[0172] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

[0173] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for predicting the correlation of surrounding rock dynamic disturbance based on TBM tunneling parameters, characterized in that, include: Acquire the original vibration signals of the surrounding rock dynamic disturbance and the real-time tunneling parameters during TBM construction; Feature extraction processing is performed based on the original vibration signal, wherein by analyzing the time domain and frequency domain characteristics of the signal, the surrounding rock dynamic disturbance characteristic parameters used to characterize the disturbance intensity are obtained. Based on the real-time tunneling parameters and the surrounding rock dynamic disturbance feature parameters, a synchronous matching process is performed, wherein the real-time tunneling parameters and the surrounding rock dynamic disturbance feature parameters are time-aligned and spatially matched to obtain a one-to-one corresponding multi-source sample dataset. Based on the multi-source sample dataset, correlation screening is performed. By calculating the correlation degree between each real-time tunneling parameter and the surrounding rock dynamic disturbance characteristic parameter, and performing a significance test, the set of key tunneling parameters that have a dominant control effect on the surrounding rock dynamic disturbance and their corresponding surrounding rock dynamic disturbance characteristic parameters are obtained. Based on the set of key tunneling parameters and their corresponding surrounding rock dynamic disturbance characteristic parameters, a mapping relationship is constructed. Specifically, by capturing the nonlinear relationship between the set of key tunneling parameters and their corresponding surrounding rock dynamic disturbance characteristic parameters, a quantitative correlation prediction model for surrounding rock dynamic disturbance is constructed, and the intensity of surrounding rock dynamic disturbance is predicted according to the real-time key tunneling parameters.

2. The method for predicting surrounding rock dynamic disturbance based on TBM tunneling parameters according to claim 1, characterized in that, Feature extraction processing is performed based on the original vibration signal, including analyzing the time-domain and frequency-domain characteristics of the signal, including: The original vibration signal is preprocessed, wherein interference components introduced by the environment and equipment are filtered out to obtain the preprocessed vibration signal. The preprocessed vibration signal is subjected to time-domain feature analysis, wherein the peak value and overall level statistics of the signal in the time dimension are calculated to obtain time-domain feature parameters including peak acceleration, peak vibration velocity and vibration amplitude. The preprocessed vibration signal is subjected to frequency domain feature analysis, wherein the signal is converted to the frequency domain and the frequency bands in which its energy is concentrated are identified, thereby obtaining frequency domain feature parameters including the dominant frequency and spectral distribution characteristics.

3. The method for predicting the dynamic disturbance of surrounding rock based on TBM tunneling parameters according to claim 1, characterized in that, Synchronous matching processing based on the real-time tunneling parameters and the surrounding rock dynamic disturbance characteristic parameters includes: Parameter statistical processing is performed based on the real-time tunneling parameters, wherein statistical characteristic values ​​representing the tunneling status during that period are obtained by statistically processing the continuously changing parameter values ​​within a preset time window; The statistical characteristic values ​​and the surrounding rock dynamic disturbance characteristic parameters are aligned with the time reference. Specifically, by mapping data from different acquisition systems to a unified analysis time window, time-synchronized sample pairs are obtained. Spatial relationship matching is performed on the time-synchronized sample pairs, wherein the relative positional relationship between vibration monitoring points and construction disturbance sources is established to obtain a multi-source sample dataset corresponding to the spatial location.

4. The method for predicting the dynamic disturbance of surrounding rock based on TBM tunneling parameters according to claim 1, characterized in that, Based on the multi-source sample dataset, correlation filtering is performed by calculating the correlation degree between each real-time tunneling parameter and the surrounding rock dynamic disturbance characteristic parameters, and then conducting a significance test, including: Data scaling is unified based on the multi-source sample dataset, wherein the normalized dataset is obtained by adjusting the tunneling parameters of different dimensions to the same numerical range. Nonparametric correlation analysis is performed on the normalized dataset, wherein the correlation coefficient reflecting the degree of monotonic correlation between the two is obtained by calculating the rank correlation between each tunneling parameter sequence and the disturbance characteristic parameter sequence. The statistical significance of the correlation coefficient is determined by examining the probability that the correlation result is caused by accidental factors. The tunneling parameters with significant correlation and exceeding the preset threshold are selected to obtain the key tunneling parameter set.

5. The method for predicting the dynamic disturbance of surrounding rock based on TBM tunneling parameters according to claim 1, characterized in that, Based on the set of key tunneling parameters and their corresponding surrounding rock dynamic disturbance characteristic parameters, a mapping relationship is constructed, including: Based on the set of key tunneling parameters and their corresponding surrounding rock dynamic disturbance characteristic parameters, multiple prediction relationship construction processes are carried out. Among them, multiple candidate prediction relationships are obtained by initializing multiple data fitting relationships with different structures. The prediction performance is evaluated and screened based on the multiple candidate prediction relationships. The initial prediction relationship with the best prediction performance is obtained by comparing the prediction error and goodness of fit of each relationship on independent data. The internal parameters of the initial prediction relationship are optimized by iteratively adjusting the internal parameters of the initial prediction relationship through a particle swarm optimization algorithm to minimize its prediction error, thereby obtaining the final optimized quantitative correlation prediction model for surrounding rock dynamic disturbance.

6. A system for predicting the correlation of surrounding rock dynamic disturbance based on TBM tunneling parameters, characterized in that, include: The acquisition unit is used to acquire the original vibration signal of the surrounding rock dynamic disturbance and the real-time tunneling parameters during TBM construction; The processing unit is used to perform feature extraction processing based on the original vibration signal, wherein by analyzing the time domain and frequency domain characteristics of the signal, the surrounding rock dynamic disturbance characteristic parameters used to characterize the disturbance intensity are obtained. The matching unit is used to perform synchronous matching processing based on the real-time tunneling parameters and the surrounding rock dynamic disturbance feature parameters. In this process, by performing time alignment and spatial location matching on the real-time tunneling parameters and the surrounding rock dynamic disturbance feature parameters, a one-to-one corresponding multi-source sample dataset is obtained. The filtering unit is used to perform correlation filtering based on the multi-source sample dataset. By calculating the correlation degree between each real-time tunneling parameter and the surrounding rock dynamic disturbance characteristic parameter and performing a significance test, the set of key tunneling parameters that have a dominant control effect on the surrounding rock dynamic disturbance and their corresponding surrounding rock dynamic disturbance characteristic parameters are obtained. The prediction unit is used to construct a mapping relationship based on the set of key tunneling parameters and their corresponding surrounding rock dynamic disturbance characteristic parameters. Specifically, by capturing the nonlinear relationship between the set of key tunneling parameters and their corresponding surrounding rock dynamic disturbance characteristic parameters, a quantitative correlation prediction model for surrounding rock dynamic disturbance is constructed, and the intensity of surrounding rock dynamic disturbance is predicted according to the real-time key tunneling parameters.

7. The surrounding rock dynamic disturbance correlation prediction system based on TBM tunneling parameters according to claim 6, characterized in that, The processing unit includes: The first processing subunit is used to perform signal preprocessing based on the original vibration signal, wherein the preprocessed vibration signal is obtained by filtering out interference components introduced by the environment and equipment in the signal. The second processing subunit is used to perform time-domain feature analysis processing on the preprocessed vibration signal, wherein time-domain feature parameters including peak acceleration, peak vibration velocity and vibration amplitude are obtained by calculating the peak value and overall level statistics of the signal in the time dimension. The third processing subunit is used to perform frequency domain feature analysis processing on the preprocessed vibration signal, wherein frequency domain feature parameters including the dominant frequency and spectral distribution characteristics are obtained by converting the signal to the frequency domain and identifying the frequency band where its energy is concentrated.

8. The surrounding rock dynamic disturbance correlation prediction system based on TBM tunneling parameters according to claim 6, characterized in that, The matching unit includes: The first matching subunit is used to perform parameter statistical processing based on the real-time tunneling parameters, wherein statistical characteristic values ​​representing the tunneling status during the time period are obtained by performing statistical processing on continuously changing parameter values ​​within a preset time window. The second matching subunit is used to perform time reference alignment processing based on the statistical feature values ​​and the surrounding rock dynamic disturbance feature parameters, wherein time-synchronized sample pairs are obtained by mapping data from different acquisition systems to a unified analysis time window. The third matching subunit is used to perform spatial relationship matching processing based on the time-synchronized sample pairs. Specifically, by establishing the relative positional relationship between vibration monitoring points and construction disturbance sources, a multi-source sample dataset corresponding to the spatial location is obtained.

9. The surrounding rock dynamic disturbance correlation prediction system based on TBM tunneling parameters according to claim 6, characterized in that, The filtering unit includes: The first screening subunit is used to perform data scale unification processing on the multi-source sample dataset, wherein the normalized dataset is obtained by adjusting the tunneling parameters of different dimensions to the same numerical range. The second screening subunit is used to perform nonparametric correlation analysis processing on the normalized dataset, wherein the correlation coefficient reflecting the degree of monotonic correlation between the two is obtained by calculating the rank correlation between each tunneling parameter sequence and the disturbance characteristic parameter sequence. The third screening subunit is used to perform statistical significance discrimination processing based on the correlation coefficient. Specifically, by examining the probability that the correlation result is caused by accidental factors, the tunneling parameters with significant correlation and exceeding the preset threshold are screened out to obtain the key tunneling parameter set.

10. The surrounding rock dynamic disturbance correlation prediction system based on TBM tunneling parameters according to claim 6, characterized in that, The prediction unit includes: The first prediction subunit is used to construct multiple prediction relationships based on the set of key tunneling parameters and their corresponding surrounding rock dynamic disturbance characteristic parameters. In this process, multiple candidate prediction relationships are obtained by initializing multiple data fitting relationships with different structures. The second prediction subunit is used to evaluate and filter the prediction performance based on the multiple candidate prediction relationships. The initial prediction relationship with the best prediction performance is obtained by comparing the prediction error and goodness of fit of each relationship on independent data. The third prediction subunit is used to perform internal parameter optimization processing based on the initial prediction relationship. Specifically, the internal parameters of the initial prediction relationship are iteratively adjusted using a particle swarm optimization algorithm to minimize its prediction error, thereby obtaining the final optimized quantitative correlation prediction model for surrounding rock dynamic disturbance.

Images