Prediction method and system for rock stratum fracture disaster in tunnel engineering

Abstract

The invention provides a prediction method and system for a rock stratum fracture disaster in tunnel engineering, and belongs to the field of tunnel engineering. Comprising the steps of multi-modal monitoring data acquisition and synchronization, data preprocessing and feature calculation, multi-modal information entropy evolution analysis, catastrophe point detection and precursor identification, risk assessment and early warning output and the like. Multiple types of sensors are arranged at key positions of a tunnel, an edge computing gateway is accessed through an industrial-grade field bus, and space-time reference unification of multi-source heterogeneous data is achieved; carrying out preprocessing and feature calculation on the collected original data; performing sliding window probability distribution modeling on each modal feature, calculating an information theory index, and analyzing non-linear relevance among the multi-modal features; judging whether a catastrophe precursor event exists or not by adopting multiple methods; and finally, based on a catastrophe point detection result, performing risk assessment on the tunnel region by using a multi-dimensional risk quantification model, and performing early warning output on high-risk and emergency states.

Description

Technical Field

[0001] The present invention belongs to the field of tunnel engineering, and more specifically relates to a method and system for predicting rock burst disasters in tunnel engineering. Background Technique

[0002] Tunnel engineering runs through various rock and soil masses. Due to the influence of various factors such as rock properties, working conditions, and construction techniques, it is prone to tunnel rock burst disasters, such as rock bursts, collapses, water inrush, geological disasters, etc., which have a serious impact on the safety and economic benefits of tunnel engineering construction.

[0003] Traditional methods for predicting tunnel rock burst disasters usually estimate based on geological exploration reports and on-site experience. Limited by complex geological structures, large changes in working conditions, and human factors, their prediction accuracy is not ideal. Once a burst disaster occurs, it often causes serious consequences such as casualties, equipment losses, and construction delays. Therefore, developing a method that can accurately and timely predict tunnel rock burst disasters has important practical value and broad market prospects.

[0004] In recent years, although there have been certain improvements in the monitoring and prediction technologies for tunnel engineering through the use of sensor devices and data analysis methods. However, existing prediction technologies for tunnel engineering rock burst disasters are mostly based on a single monitoring modality, such as only examining acoustic emission or microseismicity, etc. Such prediction methods often cannot comprehensively and accurately reflect the true state of tunnel rock formations, and are prone to false alarms and missed alarms. At the same time, most prediction systems can only provide prediction results, and cannot give specific early warning and protection measures, and the recognition mechanism for precursor events is also relatively single, unable to meet the disaster prediction requirements under complex geological conditions.

[0005] In the prior art, CN110671153A - Monitoring and Early Warning System for Water Inrush and Outburst Disasters in Tunnels and Underground Projects pays attention to tunnel construction geological disasters, covers various disasters caused by bad geology, and collects tunnel general situation, geological exploration data, and advanced prediction data.

[0006] CN116092269A - A Method, Device and Electronic Equipment for Warning of Rock Mass Disasters in Tunnel Engineering mainly aims at geological disasters in the construction of rock and soil engineering, and obtains relevant data of the construction area through a multi-source monitoring system.

[0007] CN116562656A - Intelligent Decision-making Method and Auxiliary Platform for Warning and Prevention and Control of Tunnel Construction Geological Disasters aims at water inrush disasters in tunnels. Based on geological information near the tunnel axis, geological disclosure conditions of the excavated tunnel sections, and information obtained from advanced geological forecasts, it monitors high-risk sections of water inrush.

[0008] CN117711140A - A method and system for timed early warning of water inrush disasters in tunnels based on multi - source data fusion do not mention the type of disasters, but it is speculated from the content that it is centered around a certain type of engineering disaster, and data is collected through a variety of monitoring means.

[0009] In addition, existing tunnel rock stratum disaster prediction technologies often neglect the correlation analysis between various monitoring data and do not conduct fusion analysis on multi - modal data, which to a certain extent reduces the prediction accuracy. The acquisition, synchronization, processing, and analysis of multi - modal data are difficult, and existing monitoring systems cannot effectively handle them.

[0010] Therefore, developing a system and method that can comprehensively process multi - modal monitoring data, accurately predict, and timely warn of rock stratum rupture disasters in tunnel engineering is of great significance for improving the safety of tunnel engineering and ensuring the safety of personnel's lives and property. Summary of the Invention

[0011] The present invention aims to solve the problems in the prior art that the prediction method of rock stratum rupture disasters in tunnel engineering cannot comprehensively and accurately reflect the true state of tunnel rock strata, there are false alarms and missed alarms, and there are deficiencies in early warning and protection measures. In addition, in the prior art, the acquisition, synchronization, processing, and analysis of multi - modal data are difficult, lacking effective methods for comprehensive processing, which reduces the prediction accuracy. The present invention proposes a system and method that can comprehensively process multi - modal monitoring data, accurately predict, and timely warn of rock stratum rupture disasters in tunnel engineering, improving the safety of tunnel engineering and ensuring the safety of personnel's lives and property.

[0012] To achieve the above - mentioned purpose, the present invention is implemented by adopting the following technical solutions: The method includes: Multi - modal monitoring data acquisition and synchronization: Deploy multi - type sensors at key parts of the tunnel, including acoustic emission sensors, microseismic sensors, stress sensors, displacement sensors, and ground temperature sensors. According to the geological exploration report and numerical simulation results, deploy a multi - modal sensor array in high - risk areas. All sensors are connected to the edge computing gateway through an industrial - grade fieldbus. The precise time protocol is used to achieve clock synchronization. The main clock source is used for timing and combined with the optical fiber transmission delay compensation algorithm to eliminate the link delay difference. The data acquisition terminal is configured with a multi - channel synchronous sampling card, and the time axes of each modal data are aligned through a hardware trigger signal. The original data stream is pre - processed and encapsulated into data packets with a unified timestamp, and uploaded to the central server through a dedicated network to achieve the spatio - temporal benchmark unification of multi - source heterogeneous data; Data pre - processing and feature calculation: Perform pre - processing operations such as denoising, standardization, and drift compensation on the collected various original monitoring data. Extract features such as acoustic emission energy, microseismic event number, stress volatility, displacement accumulation, and ground temperature gradient through the sliding window statistical method, and generate a feature time - series matrix; Multi-modal information entropy evolution analysis: Model the probability distribution of each modal feature using a sliding window, calculate information theory metrics such as Shannon entropy, joint entropy, and mutual information, extract the entropy change gradient through the entropy evolution curve and smoothing process, and analyze the non-linear correlation between multi-modal features using methods such as multi-scale entropy correlation matrix and maximum information coefficient; Catastrophe point detection and precursor identification: Based on methods such as single-modal mutation detection, multi-modal collaborative criterion, joint entropy extreme value detection, and causal test, determine whether there are catastrophe precursor events, and output the spatio-temporal markers of catastrophe points, precursor confidence levels, and dominant modal features; Risk assessment and early warning output: Based on the catastrophe point detection results, combined with parameters such as precursor event confidence level, entropy change gradient, and historical case similarity, use a multi-dimensional risk quantification model to evaluate the risk level of the tunnel area. The risk level is divided into five levels. The system automatically associates the three-dimensional geological model with the BIM model to achieve spatial visualization positioning of risk events, and synchronously triggers multi-channel early warnings in high-risk and emergency states, including information push, acoustic and optical alarms, automatic generation of protection and support plans, and a feedback learning mechanism to dynamically optimize model parameters.

[0013] In one solution, the layout of the multi-modal sensor array includes: installing acoustic emission sensors at intervals of 10 meters on the tunnel crown and side walls, arranging microseismic sensors using three-component geophones in a grid pattern with a spacing of no more than 20 meters, using fiber Bragg grating sensors for stress sensors to form circumferential and longitudinal cross-measurement points, using laser rangefinders and total stations for displacement sensors to form a three-dimensional monitoring network, and using a distributed fiber optic temperature measurement system for ground temperature sensors to be arranged in the tunnel surrounding rock to achieve all-round and multi-dimensional monitoring of high-risk areas.

[0014] In one solution, the data synchronization is connected to the edge computing gateway using an industrial field bus, and the master-slave clock synchronization is carried out using the Precision Time Protocol. The master clock source uses a satellite time synchronization module deployed at the tunnel entrance. Each slave node eliminates the time delay through the fiber optic link delay compensation algorithm. All monitoring data acquisition terminals use multi-channel synchronous sampling cards to ensure the time consistency of various data acquisitions through hardware triggering. Finally, the data is encapsulated into data packets with a unified timestamp for remote transmission and storage.

[0015] In one solution, in the risk assessment and early warning output step, after the system detects high-risk and emergency states, it can automatically push structured alarm information including risk coordinates, dominant modal features, and recommended disposal measures to the monitoring center, activate the on-site acoustic and optical alarm device through the Internet of Things relay device, and at the same time, for areas with a risk of collapse chain reaction, automatically generate a protection and support plan and send it to the construction robot queue. The system also has a feedback learning mechanism to automatically optimize model parameters according to the subsequent microseismic event frequency and energy release situation.

[0016] In one solution, in the multi-modal information entropy evolution analysis step, a sliding window is used to model the probability distribution of each modal feature, information theory metrics including Shannon entropy, joint entropy, and mutual information are calculated, the entropy curve is smoothed by cubic spline interpolation, the coupling relationship between each monitoring mode is analyzed using the entropy change gradient and the maximum information coefficient, and this is used as an important basis for identifying precursor events of catastrophes.

[0017] In one solution, in the data preprocessing and feature calculation step, wavelet denoising and band-pass filtering are respectively used for acoustic emission and microseismic signals for noise suppression, and multi-order smoothing algorithms and temperature drift compensation techniques are used for stress and displacement data to achieve high-precision preprocessing of various monitoring data.

[0018] In one solution, in the catastrophe point detection and precursor identification step, a method combining multi-modal mutation detection and comparison with historical cases is adopted. Only when anomalies occur simultaneously in multiple monitoring modes and are highly similar to historical disaster samples is it determined as an effective catastrophe point, significantly reducing the false alarm and missed alarm probabilities.

[0019] In one solution, in the risk assessment and warning output step, the system can automatically spatially associate risk events with the tunnel three-dimensional geological model and the BIM model, and through color grading and dynamic annotation, display the risk level and precursor location on the visualization platform in real time, facilitating managers to promptly grasp the risk distribution and take targeted measures.

[0020] On the other hand, a prediction system for rock burst disasters in tunnel engineering, the system is applicable to the method, and the system includes: a multi-modal monitoring data acquisition and synchronization module for deploying multiple types of sensors at key parts of the tunnel to achieve the spatio-temporal reference unification of multi-source heterogeneous data; a data preprocessing and feature calculation module for preprocessing various types of original monitoring data collected and generating a feature time series matrix; a multi-modal information entropy evolution analysis module for analyzing the non-linear correlation between multi-modal features; a catastrophe point detection and precursor identification module for determining whether there are precursor events of catastrophes; a risk assessment and warning output module for assessing the risk level of the tunnel area and giving a warning output after detecting high risks and emergency states.

[0021] Advantages of the present invention: 1. The present invention combines links such as multi-modal monitoring data acquisition and synchronization, data preprocessing and feature calculation, multi-modal information entropy evolution analysis, catastrophe point detection and precursor identification, risk assessment and warning output, comprehensively processes multi-modal monitoring data, accurately predicts and timely warns of rock burst disasters in tunnel engineering, greatly improving the warning accuracy and timeliness of rock burst disasters in tunnel engineering.

[0022] 2. The present invention realizes the spatio-temporal benchmark unification of multi-source heterogeneous data, adopts the Precision Time Protocol for master-slave clock synchronization, and combines the optical fiber transmission delay compensation algorithm to eliminate the link delay difference, ensuring the alignment of the time axes of various modal data and improving the accuracy and efficiency of data processing.

[0023] 3. In the process of disaster point detection and precursor identification of the present invention, a variety of methods such as single-modal mutation detection, multi-modal collaborative criterion, joint entropy extreme value detection, and causal test are combined to determine whether there are disaster precursor events, significantly reducing the false alarm and missed alarm probabilities.

[0024] 4. Compared with the prior art, the prediction method and system for rock burst disasters in tunnel engineering proposed by the present invention have significant distinguishing technical features and outstanding creativity in terms of monitoring object, data acquisition, processing and analysis, early warning decision-making, etc. In terms of the monitoring object, the present invention focuses on rock burst disasters, collects multi-modal data through multiple types of sensors and realizes spatio-temporal benchmark unification, while the prior art respectively focuses on different types of disasters, and the data acquisition synchronization and heterogeneous processing are relatively weak; in terms of data processing, the present invention adopts sliding window probability distribution modeling and information theory indicators to analyze the non-linear correlation of multi-modal features, and the preprocessing of different modal data is more targeted, and the prior art methods have their own focuses but lack the uniqueness of the present invention; in terms of early warning decision-making, the present invention conducts a five-level risk assessment based on a multi-dimensional risk quantification model, combines multiple methods to detect disaster points to reduce false alarms and missed alarms, triggers multi-channel early warnings when the risk is high and has a feedback learning mechanism, and the prior art is inferior to the present invention in terms of the comprehensiveness and accuracy of the early warning mechanism. The present invention provides a better solution for the safety guarantee of tunnel engineering with multi-modal monitoring and data fusion innovation, accurate disaster point detection and precursor identification, and comprehensive and efficient risk assessment and early warning. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 is the flowchart of the method of the present invention; Figure 2 is the system block diagram of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0026] In order to facilitate the understanding of the present invention, the present invention will be described more comprehensively with reference to the relevant drawings. The typical embodiments of the present invention are given in the drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. On the contrary, these embodiments are provided to make the disclosure of the present invention more thorough and comprehensive.

[0027] Unless otherwise defined, all technical and scientific terms used in this invention have the same meanings as those understood by those skilled in the technical field to which this invention belongs. The terms used in the description of this invention in this specification are only for the purpose of describing specific embodiments and are not intended to limit this invention. For the convenience of understanding this invention, the following will describe this invention more comprehensively with reference to the relevant drawings. The drawings show typical embodiments of this invention. However, this invention can be implemented in many different forms and is not limited to the embodiments described herein. On the contrary, the purpose of providing these embodiments is to make the disclosure of this invention more thorough and comprehensive.

[0028] As Figure 1 shown, a method for predicting rock mass rupture disasters in tunnel engineering specifically includes: Step 1. Multi-modal monitoring data collection and synchronization: Deploy multiple types of sensors (such as acoustic emission, microseismic, stress, displacement, ground temperature, etc.) at key parts of the tunnel to collect multi-modal monitoring data; unify the timestamps of all data to achieve the temporal synchronization of multi-source data.

[0029] The detailed implementation process of multi-modal monitoring data collection and synchronization needs to be carried out from three levels: sensor network construction, data synchronization mechanism, and hardware integration. First, according to the geological exploration report and numerical simulation results, deploy a multi-modal sensor array in high-risk areas of the tunnel (such as fault zones, high-stress areas, and intersections of weak surrounding rocks): install acoustic emission sensors (select broadband piezoelectric ceramic probes with a frequency range of 1 - 100 kHz) every 10 meters at the crown and side walls, deploy three-component geophones (sensitivity > 30 V / m / s) in a grid pattern (spacing ≤ 20 meters) for microseismic monitoring, embed fiber Bragg grating sensors (accuracy ±0.5%FS) in the initial support layer to form circumferential-longitudinal cross-measurement points for stress monitoring, use a laser rangefinder (range ±50 mm, resolution 0.01 mm) and a total station to form a three-dimensional monitoring network for displacement monitoring, and arrange a distributed fiber optic temperature measurement system (spatial resolution 1 m, temperature accuracy ±0.5°C) for ground temperature monitoring.

[0030] All sensors are connected to the edge computing gateway through an industrial fieldbus (such as CAN or Modbus), and the IEEE1588 Precision Time Protocol (PTP) is used to achieve microsecond-level clock synchronization. By deploying a GPS / Beidou dual-mode timing module at the tunnel entrance as the main clock source, each sub-node eliminates the transmission link delay difference through a fiber optic transmission delay compensation algorithm (compensation formula: , where T1 is the send timestamp, T2 is the receive timestamp, T3 is the response send timestamp, and T4 is the response receive timestamp), To delay the supplementary time. The data acquisition terminal is configured with a multi-channel synchronous sampling card (such as NI PXIe-6368, sampling rate ≥ 1MS / s). Through the hardware trigger signal, the time axes of various modal data are ensured to be aligned. The original data stream is encapsulated into HDF5 format data packets with a unified timestamp (format: UTC + geological time scale, accuracy ±1ms) after preprocessing, and finally uploaded to the central server through the dedicated 5G slice network for tunnels, realizing the spatio-temporal benchmark unification of multi-source heterogeneous data and providing a strictly synchronized multi-modal data stream for subsequent feature extraction.

[0031] Step 2. Data preprocessing and feature calculation; denoise and standardize various types of monitoring data. Using the sliding window method, the characteristic parameters of each mode are statistically calculated in time periods (such as acoustic emission energy, number of microseismic events, stress volatility, etc.).

[0032] The implementation process of data preprocessing and feature calculation needs to be customized according to the characteristics of different monitoring modes. For acoustic emission signals, first use the wavelet threshold denoising algorithm to eliminate equipment noise: select the Daubechies 8 wavelet basis for 5-layer decomposition, and perform soft threshold processing on each scale coefficient through an improved SUREShrink threshold function (formula: , where is the noise standard deviation, N is the signal length, is the threshold) to obtain the denoised waveform after reconstruction; microseismic data realizes event detection through band-pass filtering (1 - 200Hz) combined with the STA / LTA (Short-Term Average / Long-Term Average) algorithm (trigger condition: lasting 0.5 seconds, short-term average, long-term average). Stress monitoring data is smoothed using the Kalman filter, and the state equation is , the state vector at time t, the state transition matrix, the process noise. The observation equation is , the observation vector at time t, is the observation matrix, is the observation noise. Among them, the covariance matrices of the process noise w and the observation noise v are obtained through historical data training. Displacement data needs to compensate for temperature drift, and a temperature-displacement compensation model is established, where is the coefficient of thermal expansion, determined through laboratory calibration; the current temperature, the reference temperature.

[0033] The standardization process uses an improved Z-score method: , where is the median value, Original data, Data after standardization, is the median absolute deviation. The sliding window method adopts an overlapping design, and the window length w is dynamically adjusted according to the sampling rate (for example, when the 10-second window corresponds to a sampling rate of 100 Hz, it is 1000 sample points), and the sliding step size is set to . In terms of feature calculation: The acoustic emission energy is characterized by the root mean square of the signal within the window characterize, Acoustic emission energy, Number of sample points within the window, Signal value of the i-th sample point within the window; The number of microseismic events counts the number of triggers within the window , Number of microseismic events, Window length, , Indicator function, 1 if the condition is met, 0 if not;; The stress volatility calculates the standard deviation , Stress volatility, Number of sample points within the window, Stress value of the i-th sample point within the window, Average value of the stress values within the window; The displacement accumulation uses window integration , where is the instantaneous velocity, Displacement accumulation, is the window length, Time interval, Instantaneous velocity; The geothermal gradient calculates the linear regression coefficient , Geothermal gradient, is the window length, is the geothermal value of the i-th sample point, is the time of the i-th sample point; All feature parameters are calculated and stored in the time series feature matrix, providing input for subsequent entropy analysis.

[0034] Step 3. Multi-modal information entropy evolution analysis; Information entropy calculation: Within each time window, calculate the Shannon entropy of each modal feature respectively, reflecting the degree of disorder and uncertainty of the system. Joint entropy and mutual information analysis: Calculate the joint entropy and mutual information between different modalities to quantify the coupling and correlation changes between monitored quantities. Entropy evolution curve construction: Plot the evolution curves of information entropy, joint entropy, and mutual information on the time series.

[0035] The implementation process of multi-modal information entropy evolution analysis relies on the time series feature matrix constructed in the early stage, and reveals the system state evolution law through entropy theory. For each modal feature parameter (acoustic emission energy , the number of microseismic events , stress volatility etc.), first, dynamic probability distribution modeling is adopted: taking the sliding window as the basic unit, discretize the value range of each characteristic parameter into equal-width intervals, count the frequency of each interval within the window and calculate the probability mass function ; The probability of the i-th interval, The frequency of the j-th interval, The total number of intervals.

[0036] The calculation of Shannon entropy uses an improved weight formula , where, Shannon entropy, Weight coefficient is used to enhance the entropy value contribution of the abnormal interval. The calculation of joint entropy is realized by constructing a two-dimensional joint probability distribution: for any two modal features X and Y, divide their joint value range into grids, count the joint frequency to obtain the joint probability , then the joint entropy , Joint entropy, Joint probability. Mutual information analysis introduces a standardized mutual information index , Standardized mutual information index, which eliminates the difference in entropy value dimensions, Shannon entropy of modal X, Shannon entropy of modal Y, Joint entropy of modal X and modal Y, effectively characterizing the coupling strength between modes; for high-dimensional modal spaces (such as five-modal systems), the maximum information coefficient (MIC) is used to evaluate the multivariate non-linear correlation: , where, Maximum information coefficient, Mutual information, Gateway division parameter, grid division parameter related function , search for the optimal division method through the dynamic programming algorithm. In the stage of constructing the entropy evolution curve, cubic spline interpolation is used to smooth the entropy value sequence of the discrete time window to generate a continuous evolution trajectory. For mutation point detection, design an entropy change gradient index , Entropy change gradient index, Entropy value at time t, Entropy value at time t−1, Time interval, when ( is the historical gradient standard deviation) triggers an anomaly warning. The innovation of the algorithm is reflected in two aspects: one is the introduction of a dynamic weight factor Enhance the entropy characterization ability for non-stationary signals; second, propose a multi-scale entropy correlation matrix (k is the number of modes, Multi-scale entropy correlation matrix, The elements in the matrix represent the modes and mode The standardized mutual information index between them), and reveal the evolution law of system-level stability through matrix eigenvalue spectrum analysis. The final output is a multi-dimensional visualization result including Shannon entropy curve, mutual information heat map, and joint entropy surface, providing a quantitative basis for tunnel stability assessment.

[0037] Step 4. Catastrophe point detection and precursor identification; design a catastrophe point detection algorithm based on catastrophe theory: use the entropy mutation criterion: when the information entropy or joint entropy shows a sharp jump within a short time (that is, the first derivative or second derivative exceeds the adaptive threshold), it is determined as a catastrophe critical point; introduce a multi-modal collaborative mutation criterion: only when the entropy changes of multiple modes occur simultaneously, it is determined as an effective fracture precursor to reduce the false alarm rate.

[0038] Adopt an adaptive threshold algorithm with a sliding window to automatically adjust the sensitivity of the mutation criterion in combination with historical data.

[0039] The core of the catastrophe point detection and precursor identification algorithm lies in constructing a multi-level mutation criterion system to capture the precursor of system instability through the spatio-temporal coupling characteristics of entropy dynamic characteristics. First, based on the Shannon entropy sequence generated in Step 3 (m represents the m-th mode), adopt an improved derivative mutation detection algorithm: calculate the first derivative and the second derivative for each mode, where is the sliding window step size, The first derivative of the m-th mode at time t, The entropy value of the m-th mode at time t + 1, The entropy value of the m-th mode at time t - 1, Time interval. The second derivative of the m-th mode at time t. The adaptive threshold setting adopts the dynamic quantile method: based on the data of the past L = 24 hours, calculate the moving quantile of the absolute value of the derivative as the critical threshold. When and are satisfied, a single-mode mutation alarm is triggered, where , are sensitivity adjustment coefficients, The first derivative of the m-th mode at time t. The second derivative of the m-th mode at time t.

[0040] Introduction of Multimodal Collaboration Criterion into Modal Coupling Coefficient Matrix , is the normalized mutual information between modes i and j. When , it is determined as an effective precursor ( is the number of single-modal alarms triggered, and k is the number of modes). This criterion quantifies the correlation strength between modes through mutual information and dynamically adjusts the minimum number of alarms required for collaborative mutation. The joint entropy mutation detection uses the wavelet modulus maximum theory: decompose the joint entropy sequence to the 4th layer by Mallat wavelet decomposition. When the local maximum of the detail coefficient exceeds the historical mean and lasts for 3 sampling points, it is determined as a system-level mutation.

[0041] The false alarm suppression algorithm designs a double verification mechanism: first, verify the authenticity of the probability distribution mutation through the KL divergence test (the threshold is set to 0.4). Among them, is the KL divergence of the probability distributions at time t and t−1, is the probability distribution at time t, is the probability at time t - 1. Then, verify the time series relationship of multimodal mutations through the Granger causality test (F statistic > 4.0). The innovation of the algorithm lies in the integration of triple criteria of derivative mutation, wavelet singularity detection, and probability distribution migration, and the establishment of a dynamic collaboration mechanism based on mutual information weights. Its mathematical expression is: ; where is the standard normal distribution function, is the collaborative penalty factor, is the early warning probability, is the first derivative of the m-th mode at time t, is the mean of the first derivative of the m-th mode, is the standard deviation of the first derivative of the m-th mode, is the collaborative penalty factor, is the normalized mutual information between modes i and j. The final output is an early warning signal containing the timestamp of the disaster point, the precursor confidence level (in the range of 0 - 1), and the list of dominant modes, providing quantitative support for engineering emergency decision-making.

[0042] Step 5. Risk assessment and early warning output; according to the disaster point detection results, calculate the rupture risk level of the current tunnel area. If a disaster point is detected, an early warning is automatically triggered, and the corresponding spatio-temporal position and risk level are output.

[0043] The implementation process of risk assessment and early warning output is based on the spatio-temporal markers and confidence parameters output in the catastrophe point detection stage, and realizes the hierarchical early warning of tunnel stability through a multi-dimensional risk quantification model. The system first constructs a dynamic risk index , The dynamic risk index at time t, , , is the weight coefficient, is the confidence of precursor events, represents the entropy change gradient of each mode, is the KL divergence similarity between the current state and the historical catastrophe case library, and the weight coefficient is dynamically adjusted by the analytic hierarchy process. The risk level is divided into five levels: when it is level I (safe state), is level II (low risk), is level III (moderate risk), is level IV (high risk), then triggers level V (emergency state). When generating early warning information, the system automatically associates the spatial topology data in the 3D geological model, maps the risk events to the corresponding lining segments of the tunnel BIM model, and realizes the visual positioning of risks through color gradient rendering.

[0044] For risks above level IV, multi-channel early warnings are synchronously triggered: pushing structured alarm information including risk coordinates, dominant modal characteristics, and recommended disposal measures to the monitoring center; activating on-site audible and visual alarm devices through Internet of Things relay devices; for areas with a risk of collapse chain reaction, automatically generating a protection and support plan and sending it to the construction robot queue. The system has a built-in feedback learning mechanism, continuously tracking the frequency of microseismic events and the energy release curve within 72 hours after early warning. If the actual rupture intensity does not reach the predicted value, the sensitivity parameters of similar scenarios are automatically reduced, otherwise the weight coefficients of related modes are enhanced to realize the dynamic optimization of the risk assessment model.

[0045] As Figure 2 shown, a prediction system for rock rupture disasters in tunnel engineering includes: a multi-modal monitoring data acquisition and synchronization module for deploying multi-type sensors at key tunnel parts to unify the spatio-temporal benchmarks of multi-source heterogeneous data; a data preprocessing and feature calculation module for preprocessing the collected various original monitoring data and generating a feature time series matrix; a multi-modal information entropy evolution analysis module for analyzing the non-linear correlation between multi-modal features; a catastrophe point detection and precursor identification module for determining whether there are catastrophe precursor events; a risk assessment and early warning output module for evaluating the risk level of the tunnel area and performing early warning output after detecting high risks and emergency states.

[0046] The present invention focuses on rock burst disasters in tunnel engineering, such as rock burst and collapse. By deploying various types of sensors such as acoustic emission, microseismic, stress, displacement, and ground temperature sensors at key positions in the tunnel, multi-modal monitoring data is collected.

[0047] When collecting data, the present invention pays attention to achieving the spatio-temporal benchmark unification of multi-source heterogeneous data, and uses the precise time protocol and fiber optic transmission delay compensation algorithm to ensure the alignment of the time axes of each modal data. The prior art has not reached the level of the present invention in terms of the synchronization of data collection and the processing of data type heterogeneity.

[0048] The present invention models the probability distribution of each modal feature using a sliding window, calculates information theory indicators such as Shannon entropy, joint entropy, and mutual information, and analyzes the non-linear correlation between multi-modal features. Prior art 1 uses multi-source information fusion technology to comprehensively identify bad geology, and applies multi-scale geological model theory and multi-scale topological reconstruction technology to construct a three-dimensional geological model; prior art 2 constructs a data feature space for feature extraction and clustering analysis; prior art 3 predicts future time series monitoring data based on historical disaster warning monitoring information, and calculates the risk probability using a risk prediction model; prior art 4 does not explicitly mention methods such as information entropy analysis.

[0049] In terms of data preprocessing, the present invention adopts specific denoising and compensation technologies for different monitoring modalities. For example, the wavelet threshold denoising algorithm is used for acoustic emission signals, and multi-order smoothing algorithms and temperature drift compensation technologies are used for stress and displacement data. The prior art is relatively weak in terms of the pertinence of data preprocessing and the diversity of methods.

[0050] The present invention comprehensively uses various types of sensors for multi-modal monitoring, realizes the spatio-temporal benchmark unification of multi-source heterogeneous data, and can comprehensively and accurately reflect the true state of the tunnel rock formation. The multi-modal information entropy evolution analysis method deeply explores the non-linear correlation between each modal feature, providing a more accurate basis for the identification of disaster precursors, which is an innovation point not covered by the prior art.

[0051] The present invention uses a variety of methods to discriminate disaster precursor events, combines single-modal mutation detection, multi-modal collaborative criteria, joint entropy extreme value detection, and causal test, etc., to form a comprehensive and accurate disaster point detection and precursor identification system, significantly reducing the false alarm and missed alarm probabilities. The prior art lacks the multi-dimensionality and comprehensiveness of the present invention in terms of disaster detection and precursor identification.

[0052] The multi-dimensional risk quantification model of the present invention combines multiple parameters for risk level assessment. It can not only issue warnings in a timely and accurate manner, but also automatically push structured alarm information, generate protection support plans and have a feedback learning mechanism, improving the response efficiency and safety of rock burst disasters in tunnel engineering. In terms of disaster response, the early warning and decision-making mechanism of the present invention is more comprehensive and efficient, and can better ensure the safety of tunnel engineering.

[0053] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The said program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above various methods. Among them, the said storage medium can be a magnetic disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM), etc.

[0054] It should be understood that the detailed description of the technical solutions of the present invention with the aid of the preferred embodiments above is illustrative rather than restrictive. Those of ordinary skill in the art can modify the technical solutions recorded in each embodiment or perform equivalent replacements for some of the technical features on the basis of reading the specification of the present invention; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of each embodiment of the present invention.

Claims

1. A prediction method for rock mass rupture disasters in tunnel engineering, characterized in that, The described method includes: Multi-modal monitoring data acquisition and synchronization: Deploy multiple types of sensors at key parts of the tunnel, including acoustic emission sensors, microseismic sensors, stress sensors, displacement sensors, and ground temperature sensors. Based on the geological exploration report and numerical simulation results, deploy a multi-modal sensor array in high-risk areas. All sensors are connected to the edge computing gateway through an industrial fieldbus. Use the Precision Time Protocol to achieve clock synchronization. Synchronize the clock through the master clock source and combine the fiber optic transmission delay compensation algorithm to eliminate the link delay difference. The data acquisition terminal is configured with a multi-channel synchronous sampling card, and the time axes of data in each modality are aligned through a hardware trigger signal. The original data stream is preprocessed and encapsulated into data packets with a unified timestamp, and uploaded to the central server through a dedicated network to achieve the spatio-temporal reference unification of multi-source heterogeneous data; Data preprocessing and feature calculation: Perform preprocessing operations such as denoising, standardization, and drift compensation on various types of original monitoring data collected. Extract features such as acoustic emission energy, microseismic event count, stress volatility, displacement accumulation, and ground temperature gradient through the sliding window statistical method, and generate a feature time series matrix; Multi-modal information entropy evolution analysis: Use sliding window probability distribution modeling for each modality feature, calculate information theory indicators such as Shannon entropy, joint entropy, and mutual information. Extract the entropy change gradient through the entropy evolution curve and smoothing process, and analyze the non-linear correlation between multi-modal features using the multi-scale entropy correlation matrix and the maximum information coefficient method; Catastrophe point detection and precursor identification: Based on single-modal mutation detection, multi-modal collaborative criteria, joint entropy extreme value detection, and causal test methods, determine whether there are catastrophe precursor events, and output the spatio-temporal markers of catastrophe points, precursor confidence levels, and dominant modality features; Risk assessment and early warning output: Based on the catastrophe point detection results, combined with precursor event confidence levels, entropy change gradients, and historical case similarity parameters, use a multi-dimensional risk quantification model to evaluate the risk level of the tunnel area. The risk level is divided into five levels. The system automatically associates the three-dimensional geological model with the BIM model to achieve spatial visualization positioning of risk events, and synchronously triggers multi-channel early warnings in high-risk and emergency states, including information push, audible and visual alarms, automatically generating protection support plans, and a feedback learning mechanism to dynamically optimize the model parameters.

2. The prediction method for rock stratum rupture disasters in a tunnel project according to claim 1, wherein: The deployment of the multi-modal sensor array includes: Install acoustic emission sensors at intervals of 10 meters on the tunnel vault and side walls. The microseismic sensors use three-component geophones and are arranged in a grid pattern at intervals of no more than 20 meters. The stress sensors use fiber Bragg grating sensors and form circumferential and longitudinal cross-measurement points. The displacement sensors use laser rangefinders and total stations to form a three-dimensional monitoring network. The ground temperature sensors use a distributed fiber optic temperature measurement system and are arranged in the tunnel surrounding rock to achieve all-round and multi-dimensional monitoring of high-risk areas.

3. A prediction method for rock burst disasters in tunnel engineering according to claim 1, characterized in that: The data synchronization is connected to the edge computing gateway through an industrial fieldbus, and the master-slave clock synchronization is carried out using the Precision Time Protocol. The master clock source uses a satellite time service module deployed at the tunnel entrance. Each sub-node eliminates the time delay through the fiber optic link delay compensation algorithm. All monitoring data acquisition terminals use multi-channel synchronous sampling cards to ensure the time consistency of various data acquisitions through hardware triggering. Finally, the data is encapsulated into data packets with a unified timestamp for remote transmission and storage.

4. A prediction method for rock formation rupture disasters in tunnel engineering according to claim 1, characterized in that: In the risk assessment and early warning output step, after the system detects high-risk and emergency states, it can automatically push structured alarm information including risk coordinates, dominant modal characteristics, and recommended disposal measures to the monitoring center, activate the on-site audible and visual alarm device through the Internet of Things relay device. At the same time, for areas with the risk of collapse chain reaction, it automatically generates a protection and support plan and sends it to the construction robot queue. The system also has a feedback learning mechanism to automatically optimize the model parameters according to the frequency and energy release of subsequent microseismic events.

5. A prediction method for rock formation rupture disasters in tunnel engineering according to claim 1, characterized in that: In the multi-modal information entropy evolution analysis step, a sliding window is used to model the probability distribution of each modal characteristic, and information theory indexes including Shannon entropy, joint entropy, and mutual information are calculated. The entropy curve is smoothed by cubic spline interpolation, and the coupling relationship between each monitoring mode is analyzed using the entropy change gradient and the maximum information coefficient, which is used as an important basis for the identification of disaster precursors.

6. The prediction method for rock formation rupture disasters in a tunnel project according to claim 1, wherein: In the data preprocessing and feature calculation step, wavelet denoising and band-pass filtering are respectively used for acoustic emission and microseismic signals to suppress noise, and multi-order smoothing algorithms and temperature drift compensation techniques are used for stress and displacement data to achieve high-precision preprocessing of various monitoring data.

7. A prediction method for rock formation fracture disasters in tunnel engineering according to claim 1, characterized in that: In the disaster point detection and precursor identification step, a method combining multi-modal mutation detection and comparison with historical cases is adopted. Only when anomalies appear simultaneously in multiple monitoring modes and are highly similar to historical disaster samples, it is determined as an effective disaster point, significantly reducing the false alarm and missed alarm probabilities.

8. A prediction method for rock mass rupture disasters in tunnel engineering according to claim 1, characterized in that: In the risk assessment and early warning output step, the system can automatically spatially associate risk events with the tunnel three-dimensional geological model and BIM model, and through color grading and dynamic annotation, display the risk level and precursor location on the visualization platform in real time, facilitating managers to timely grasp the risk distribution and take targeted measures.

9. A prediction system for rock burst disasters in tunnel engineering, wherein the system is applicable to the method described in any one of claims 1-8, and is characterized in that, The described system includes: a multi-modal monitoring data acquisition and synchronization module for deploying multiple types of sensors at key tunnel parts to achieve the spatio-temporal reference unification of multi-source heterogeneous data; a data preprocessing and feature calculation module for preprocessing various original monitoring data collected and generating a feature time series matrix; a multi-modal information entropy evolution analysis module for analyzing the non-linear correlation between multi-modal characteristics; a disaster point detection and precursor identification module for determining whether there are disaster precursor events; a risk assessment and early warning output module for evaluating the risk level of the tunnel area and performing early warning output after detecting high-risk and emergency states.

Images