A method for intelligent identification and prevention of potential instability of a roadway roof rock layer

Abstract

This invention discloses an intelligent identification and control method for potential unstable rock strata in roadway roofs, relating to the field of mine safety monitoring and intelligent control technology. The specific steps of this method are as follows: synchronously acquiring microseismic, acoustic emission, and stress data; after preprocessing, using a weighted cross-correlation algorithm to achieve spatiotemporal synchronization and fracture location; constructing a multi-field three-dimensional coupling model to achieve integrated spatial representation; extracting the energy-frequency-stress spatiotemporal evolution fingerprint; using a weighted dynamic time warping algorithm and instability template matching to identify risk levels and implement graded control measures; this invention achieves spatiotemporal synchronization and multi-field three-dimensional coupling of roof fracture data through a multimodal sensor network and a weighted cross-correlation algorithm, improving monitoring reliability; combining an identification database and a weighted dynamic time warping algorithm to extract spatiotemporal fingerprint features of fractures and identify early instability risks; and formulating graded control measures based on risk levels to achieve differentiated management and improve the scientific nature and effectiveness of roof safety management.

Description

Technical Field

[0001] This invention relates to the field of mine safety monitoring and intelligent prevention and control technology, specifically to a method for intelligent identification and prevention and control of potentially unstable rock strata in the roof of a roadway. Background Technology

[0002] Roof instability is one of the most frequent and serious geological hazards in underground mining. Its sudden onset and destructive power can easily cause casualties and equipment damage, severely restricting mine safety and production efficiency. With the gradual depletion of shallow mineral resources in my country and the continuous increase in mining depth, the combined effects of high ground stress, complex geological structures, and strong mining activities at depth have made the evolution of roof strata fracturing more complex, significantly increasing the risk of instability. Traditional roof safety management models relying on manual inspections and single-parameter monitoring are no longer sufficient to meet the high-precision, real-time roof risk prevention and control needs of deep mines. There is an urgent need to develop intelligent technologies for identifying and controlling potential roof instability.

[0003] Existing technologies for monitoring roof instability in roadways mostly employ single-mode monitoring, acquiring only one type of data—surrounding rock stress, microseismic activity, or acoustic emission—failing to comprehensively reflect the coupling relationship between energy release, frequency evolution, and stress redistribution during roof strata fracturing. This leads to frequent omissions and misjudgments. Some multi-source monitoring systems lack effective methods for spatiotemporal synchronization of multi-source data, resulting in temporal and spatial discrepancies between data from different sensors, making it difficult to construct a unified roof condition evaluation model. Furthermore, existing instability identification methods are largely based on static threshold judgments, failing to consider the spatiotemporal evolution characteristics of strata fracturing. This makes it impossible to identify the location and extent of potentially unstable strata in advance, and prevention and control measures lack tiered targeting, leading to either excessive prevention increasing costs or insufficient prevention leaving hidden dangers.

[0004] To address the shortcomings of existing technologies, this invention proposes an intelligent identification and prevention method for potentially unstable rock strata in roadway roofs. This method achieves simultaneous acquisition of microseismic, acoustic emission, and surrounding rock stress data by constructing a multimodal monitoring sensor network. A weighted cross-correlation algorithm is used to achieve spatiotemporal synchronization of multi-source data, establishing a three-dimensional coupled field and spatiotemporal evolution fingerprint database. A weighted dynamic time warping algorithm is employed to accurately identify potentially unstable rock strata and formulate tiered prevention and control measures. This invention effectively improves the accuracy and timeliness of roof instability early warning, providing technical support for the safety management of roadway roofs in deep mines. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of existing technologies and provide a method for intelligent identification and prevention of potential unstable rock strata in roadway roofs. This method synchronously collects microseismic, acoustic emission, and surrounding rock stress data through a multimodal sensor network, and achieves spatiotemporal synchronization and fracture location through wavelet packet denoising and weighted cross-correlation algorithms. A three-dimensional coupled model of the microseismic energy field, acoustic emission main frequency field, and surrounding rock stress field is constructed, and spatiotemporal evolution fingerprints containing energy, frequency, and stress are extracted. An instability identification database is established, and a weighted dynamic time warping algorithm is used to match real-time fingerprints with historical templates to identify the instability location, range, and risk level, and implement graded prevention and control to achieve intelligent management and control of roof instability risks.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a method for intelligent identification and prevention of potentially unstable rock strata in roadway roofs, the specific steps of which are as follows:

[0007] S100, Multimodal Data Acquisition: A multimodal monitoring sensor network is deployed on the roof of the roadway to simultaneously acquire microseismic signals, acoustic emission signals, and surrounding rock stress data;

[0008] S200, Multi-source data preprocessing: The collected multi-modal data is preprocessed to remove environmental noise and support vibration interference. The weighted cross-correlation algorithm is used to achieve spatiotemporal synchronization of multi-source data and locate the occurrence of fracture signals in different rock strata.

[0009] S300, 3D Coupled Field Construction: Construct a 3D coupled model of the microseismic energy field, acoustic emission main frequency field and surrounding rock stress field, and map multimodal data to a unified 3D spatial coordinate system;

[0010] S400, Evolutionary fingerprint database establishment: Call the pre-built instability identification database, which contains spatiotemporal evolution fingerprint templates of different instability cases; extract the spatiotemporal evolution fingerprint of energy-frequency-stress during the fracturing process of unstable rock strata;

[0011] S500 Risk Identification and Prevention: The spatiotemporal evolution fingerprint extracted in real time is matched with the spatiotemporal evolution fingerprint template in the instability identification database. The similarity is calculated using a weighted dynamic time warping algorithm to identify the location, extent, and risk level of potentially unstable rock strata. Corresponding graded prevention and control measures are implemented according to the risk level.

[0012] Furthermore, the multimodal data acquisition utilizes a unified clock triggering mechanism. All monitoring devices are connected to the same time synchronization network, and each sensor is triggered to start data acquisition simultaneously via a pulse signal. During the acquisition process, the sampling frequency of each sensor is kept stable, with a default acquisition frequency of one set of data every 5 minutes and a data acquisition duration of 10 seconds per set. The microseismic signal acquisition covers the frequency band from 10Hz to 500Hz, the acoustic emission signal acquisition covers the frequency band from 100Hz to 2000Hz, and the surrounding rock stress data acquisition covers the range from 0MPa to 30MPa. The acquired raw data is stored in a local cache in timestamp order, with a cache capacity sufficient for 72 consecutive hours of data storage. The cached data is transmitted to the ground control center in real time via industrial Ethernet. During transmission, a cyclic redundancy check method is used to ensure data integrity. A check is performed every 1024 bytes of data transmitted, and if the check fails, the data segment is automatically retransmitted.

[0013] Furthermore, the denoising process of the multi-source data preprocessing involves performing 5-level wavelet packet decomposition on the acquired microseismic signals and acoustic emission signals to obtain wavelet packet coefficients for 32 different frequency bands; statistically analyzing the energy proportion of coefficients in each frequency band, retaining frequency band coefficients with an energy proportion greater than 5%, and removing high-frequency noise and low-frequency trend terms with an energy proportion less than 5%; reconstructing the retained wavelet packet coefficients to obtain the denoised microseismic signals and acoustic emission signals; and performing moving average filtering on the surrounding rock stress data with a moving window size of 10 sampling points to remove random fluctuation interference in the surrounding rock stress data.

[0014] Furthermore, the calculation formula for the weighted cross-correlation algorithm is as follows:

[0015]

[0016] in, Let be the weighted cross-correlation coefficient between signal i and signal j at a time offset of τ; This is an exponentially decaying weighting function, where α is the decay coefficient in units of s⁻¹ and has a value of 0.01. Let be the amplitude of the denoised signal i at time t; For the denoised signal j at time... The amplitude; The signal length;

[0017] By searching The maximum value corresponding to The time offset was determined to achieve time synchronization between the microseismic signal and the acoustic emission signal. The weighted cross-correlation coefficient distribution of the stress sensor and the denoised signal at different depths was statistically analyzed, and the source rock layer of the fracture signal was determined based on the depth corresponding to the peak value.

[0018] Furthermore, the construction process of the three-dimensional coupling model involves establishing a three-dimensional rectangular coordinate system with the geometric center of the roadway roof as the origin. The x-axis runs along the roadway direction, the y-axis runs along the roadway width, and the z-axis points vertically upward. The monitoring area is divided into three-dimensional grid cells with a resolution of 0.5m × 0.5m × 0.5m. Each grid cell contains three scalar attributes: microseismic energy value, acoustic emission dominant frequency value, and surrounding rock stress value. Kriging interpolation is used to spatially interpolate the discrete monitoring data of each attribute to obtain the attribute value of each grid cell. The attribute values ​​of all grid cells are integrated to construct a three-dimensional coupling model of the microseismic energy field, acoustic emission dominant frequency field, and surrounding rock stress field covering the entire monitoring area, thus completing the unified spatial representation of multimodal data.

[0019] Furthermore, the extraction process of the spatiotemporal evolution fingerprint is as follows: From the three-dimensional coupled model, the microseismic energy release rate, acoustic emission dominant frequency, acoustic emission dominant frequency offset, surrounding rock stress concentration factor, correlation between microseismic energy and acoustic emission dominant frequency, and coupling degree between surrounding rock stress and microseismic energy are extracted from each grid cell; the time variation rate and spatial gradient value of the microseismic energy release rate, the time variation rate and spatial gradient value of the surrounding rock stress concentration factor, and the time variation rate and spatial gradient value of the acoustic emission dominant frequency offset are calculated; the weighted comprehensive value of the microseismic energy release rate, surrounding rock stress concentration factor, acoustic emission dominant frequency offset, correlation between microseismic energy and acoustic emission dominant frequency, coupling degree between surrounding rock stress and microseismic energy, and the aforementioned time variation rate of microseismic energy release rate and spatial gradient value of the surrounding rock stress concentration factor are used as seven key indicators, arranged in order to form a spatiotemporal evolution fingerprint feature vector; the formula for calculating the time variation rate is... ,in The parameter value at the current moment. The parameter value at the previous sampling time. The sampling time interval is in seconds; the formula for calculating the spatial gradient value is as follows: ,in The parameter value of the current mesh cell. These are the parameter values ​​of the adjacent mesh cells in the negative x-axis direction. The grid spacing is set to 0.5m; all extracted parameters are arranged in order to form a spatiotemporal evolution fingerprint feature vector.

[0020] Furthermore, the process of establishing the instability identification database is as follows: Data on roadway roof instability cases under different geological conditions, mining depths, and support methods are collected. Each case includes complete multimodal monitoring data and instability process records. The multimodal monitoring data for each case is preprocessed according to the S200 method to extract the corresponding spatiotemporal evolution fingerprint. K-means clustering is used to classify all extracted spatiotemporal evolution fingerprints, with three clusters corresponding to low, medium, and high risk levels. Based on the actual instability damage degree of historical cases within each cluster, the cluster with the least damage is labeled as low risk, the cluster with the most severe damage is labeled as high risk, and the rest are labeled as medium risk. The classified fingerprints are stored as templates in the instability identification database, with each template containing a fingerprint feature vector and corresponding risk level information.

[0021] Furthermore, the calculation formula for the weighted dynamic time warping algorithm is as follows:

[0022]

[0023] in, Fingerprint feature vector time series and The weighted dynamic time-normalized distance between them; For time-ordered paths; and Paths exist and Time index on; is the path length; F is the total number of features, with a value of 7. The weight of the f-th feature is determined based on the contribution of each feature to instability identification in historical instability cases. The weights are as follows: microseismic energy release rate (0.30), surrounding rock stress concentration factor (0.30), acoustic emission dominant frequency offset (0.20), correlation between microseismic energy and acoustic emission dominant frequency (0.05), coupling between surrounding rock stress and microseismic energy (0.05), and weighted average of the time variation rate of microseismic energy release rate and the spatial gradient of surrounding rock stress concentration factor (0.10).

[0024] Convert the above weighted dynamic time-normalized distance into similarity: When the similarity S between the real-time fingerprint and the template is greater than 0.85, it is determined that there is a potential risk of instability.

[0025] Furthermore, the risk level classification criteria are as follows:

[0026] Level 1 risk is low risk, corresponding to a spatiotemporal evolution fingerprint with a similarity greater than 0.85 to the low-risk template, a microseismic energy release rate of less than 100 J / h, and a surrounding rock stress concentration factor of less than 1.5.

[0027] Level 2 risk is medium risk, corresponding to a spatiotemporal evolution fingerprint with a similarity greater than 0.85 with the medium risk template, and a microseismic energy release rate of 100-500 J / h, and a surrounding rock stress concentration factor of 1.5-2.0;

[0028] Level 3 risk is high risk, corresponding to a spatiotemporal evolution fingerprint with a similarity greater than 0.85 to a high-risk template, a microseismic energy release rate greater than 500 J / h, and a surrounding rock stress concentration factor greater than 2.0;

[0029] When the risk level determined by fingerprint similarity is inconsistent with the risk level determined by microseismic energy release rate, the risk level determined by microseismic energy release rate shall prevail; when the risk level determined by microseismic energy release rate is inconsistent with the risk level determined by surrounding rock stress concentration coefficient, the higher risk level shall prevail.

[0030] Furthermore, the aforementioned tiered prevention and control measures:

[0031] Level 1 Risk: Increase the data acquisition frequency to 5 times the default acquisition frequency and monitor continuously for 72 hours; if the microseismic energy release rate is consistently less than 100 J / h and the surrounding rock stress concentration factor is consistently less than 1.5 within 72 hours, then restore the default frequency.

[0032] Level 2 risk: Increase the pre-tightening force of all anchor bolts in the risk area to 1.2 times the initial pre-tightening force, and maintain a 5 times sampling frequency for intensive monitoring until the microseismic energy release rate drops below 100 J / h and the stress concentration factor of the surrounding rock drops below 1.5;

[0033] Level 3 risk: Within the risk zone, anchor cables with a length of 8m and a preload of 200kN are densely arranged at 1.5m×1.5m intervals. Grouting holes are arranged between the anchor cables, with a drilling depth of 7m. A two-component grout with a water-cement ratio of 1:1 and a cement-water glass volume ratio of 4:1 is injected at a grouting pressure of 2-3MPa. After curing for 7 days, the grouting is retested. If the microseismic energy release rate is confirmed to be less than 100J / h and the surrounding rock stress concentration factor is less than 1.5, the risk is eliminated.

[0034] Compared with existing technologies, this intelligent identification and prevention method for potential instability of roadway roof strata has the following advantages:

[0035] I. This invention overcomes the limitations of traditional single-modal monitoring technology by constructing a multimodal monitoring sensor network to simultaneously collect various key data reflecting the state of the roof strata in roadways. It can comprehensively capture multi-dimensional information during the roof strata fracturing process. By employing targeted data preprocessing methods to remove various environmental and support interference signals, and combining a weighted cross-correlation algorithm to achieve spatiotemporal synchronization of multi-source data and location of fracturing signal sources, it effectively eliminates the temporal and spatial deviations of data from different sensors, improving data consistency and reliability. By constructing a multi-field three-dimensional coupling model, multimodal data is mapped to a unified spatial coordinate system, realizing an integrated spatial expression of the roof strata state.

[0036] Second, this invention establishes an identification database containing instability cases under different working conditions, extracts spatiotemporal evolution fingerprint features during the rock strata fracturing process, and can fully reflect the dynamic evolution process of roof strata from micro-fracture to instability. By employing a weighted dynamic time warping algorithm to achieve similarity matching between real-time fingerprints and historical templates, and combining multiple indicators to comprehensively determine the location, range, and risk level of potentially unstable rock strata, it solves the problem that traditional static threshold identification methods cannot identify early potential risks. By formulating graded prevention and control measures corresponding to risk levels, it achieves differentiated management of roof instability risks, ensuring prevention and control effectiveness while avoiding resource waste, and significantly improving the scientificity and effectiveness of roadway roof safety management.

[0037] Other advantages, objectives and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination or study, or may be learned from the practice of the invention. Attached Figure Description

[0038] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0039] Figure 1 A step-by-step framework diagram of a method for intelligent identification and prevention of potentially unstable rock strata in roadway roof;

[0040] Figure 2 A flowchart of multi-source data preprocessing and spatiotemporal synchronization for a method of intelligent identification and prevention of potentially unstable rock strata in roadway roof;

[0041] Figure 3 This is a flowchart illustrating the risk identification and graded prevention and control process of a smart identification and prevention and control method for potential unstable rock strata in roadway roof. Detailed Implementation

[0042] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0043] Example 1:

[0044] During the multimodal data acquisition phase, a multimodal monitoring sensor network was deployed on the roof of the target roadway in deep mining to simultaneously acquire microseismic signals, acoustic emission signals, and surrounding rock stress data. A unified clock triggering mechanism was adopted, with all monitoring devices connected to the same time synchronization network. Pulse signals triggered each sensor to simultaneously start data acquisition. During acquisition, the sampling frequency of each sensor remained stable, with a default acquisition frequency of one data set every 5 minutes and each data set acquisition duration of 10 seconds. This ensured that the data acquired by different sensors were completely aligned on the time axis, eliminating timing deviations caused by independent timing of different devices and laying the foundation for subsequent correlation analysis of multi-source data. Microseismic signal acquisition covered the 10Hz to 500Hz frequency band, acoustic emission signal acquisition covered the 100Hz to 2000Hz frequency band, and surrounding rock stress data acquisition covered the 0MPa to 30MPa range. This accurately matched the signal frequency range and stress change interval of micro-fractures in the roof strata of deep roadways, comprehensively capturing various precursory information before rock instability and avoiding the omission of critical signals. The collected raw data is stored in a local cache in chronological order according to timestamps. The cache capacity meets the data storage requirements for 72 consecutive hours, ensuring complete retention of all monitoring data during brief downhole network outages and preventing monitoring gaps caused by data loss. The cached data is transmitted to the ground control center in real time via industrial Ethernet. Cyclic redundancy check (CRC) is used during transmission to ensure data integrity; a check is performed every 1024 bytes of data transmitted. If a check fails, the data segment is automatically retransmitted, guaranteeing the accuracy of every set of data received by the ground control center and providing reliable raw data support for subsequent data analysis and risk assessment.

[0045] In the multi-source data preprocessing stage, the collected multimodal data are preprocessed to remove environmental noise and support vibration interference. A weighted cross-correlation algorithm is used to achieve spatiotemporal synchronization of the multi-source data and locate the occurrence of different rock strata fracture signals. The collected microseismic and acoustic emission signals are decomposed into 5-layer wavelet packets, yielding wavelet packet coefficients for 32 different frequency bands. The energy proportion of each frequency band coefficient is statistically analyzed, retaining coefficients with an energy proportion greater than 5%, while removing high-frequency noise and low-frequency trend terms with an energy proportion less than 5%. This accurately separates the effective signals generated by rock strata fracture, filtering out environmental interference signals generated by downhole fan operation, equipment movement, and support installation, significantly improving the signal-to-noise ratio of the effective signals. The retained wavelet packet coefficients are reconstructed to obtain the denoised microseismic and acoustic emission signals, restoring the true form of the rock strata fracture signals. The surrounding rock stress data is processed using a moving average filter with a sliding window size of 10 sampling points to smooth out random fluctuations in the surrounding rock stress data caused by minute sensor vibrations and instantaneous elastic deformation of the surrounding rock, clearly presenting the true trend of surrounding rock stress changes over time. The weighted cross-correlation algorithm is used to achieve spatiotemporal synchronization of multi-source data. The formula is as follows: ,in, Let be the weighted cross-correlation coefficient between signal i and signal j at a time offset of τ; This is an exponentially decaying weighting function, where α is the decay coefficient in units of s⁻¹ and has a value of 0.01. Let be the amplitude of the denoised signal i at time t; For the denoised signal j at time... The amplitude; The signal length is used as the reference point. By finding the time offset corresponding to the maximum value of the weighted cross-correlation coefficient, precise time synchronization between the microseismic signal and the acoustic emission signal is achieved. This allows monitoring data from different modes to be analyzed in the same time dimension, accurately reflecting the energy release, frequency response, and stress state of the rock strata at the same moment. The distribution of weighted cross-correlation coefficients between stress sensors at different depths and the denoised signals is statistically analyzed. The source rock stratum of the rupture signal is determined based on the depth corresponding to the peak value, accurately locating the vertical stratum where the rupture occurred. This provides precise locational basis for subsequent risk area delineation and the development of targeted prevention and control measures.

[0046] In the three-dimensional coupled field construction stage, a three-dimensional coupled model of the microseismic energy field, acoustic emission dominant frequency field, and surrounding rock stress field is constructed, mapping multimodal data to a unified three-dimensional spatial coordinate system. A three-dimensional rectangular coordinate system is established with the geometric center of the tunnel roof as the origin. The x-axis follows the tunnel direction, the y-axis follows the tunnel width, and the z-axis points vertically upwards. This coordinate system perfectly matches the spatial structure of the tunnel, facilitating a direct visualization of the spatial distribution of various parameters within the monitoring area. The monitoring area is divided into three-dimensional grid cells with a resolution of 0.5m × 0.5m × 0.5m. This sufficiently high spatial resolution allows for detailed characterization of local variations in the roof strata and accurate identification of small-scale stress concentration zones and energy release points. Each grid cell contains three scalar attributes: microseismic energy value, acoustic emission dominant frequency value, and surrounding rock stress value. Kriging interpolation is used to spatially interpolate the discrete monitoring data for each attribute, extending the limited sensor monitoring point data to the entire monitoring area, resulting in continuous and smooth spatial distribution data and filling information gaps between monitoring points. By integrating the attribute values ​​of all grid cells, a three-dimensional coupled model of microseismic energy field, acoustic emission dominant frequency field, and surrounding rock stress field covering the entire monitoring area is constructed. This completes the unified spatial representation of multimodal data, realizes the synchronous display of microseismic energy, acoustic emission dominant frequency, and surrounding rock stress at the same spatial location, and intuitively presents the spatial correlation between the three. This helps to identify the concentrated areas and evolution directions of rock strata fracture, and provides a complete spatial data foundation for subsequent extraction of spatiotemporal evolution fingerprints.

[0047] In the evolutionary fingerprint database establishment phase, a pre-constructed instability identification database is invoked. This database contains spatiotemporal evolutionary fingerprint templates for different instability cases, extracting the spatiotemporal evolutionary fingerprint of energy-frequency-stress during the fracturing process of unstable rock strata. From the three-dimensional coupled model, the microseismic energy release rate, acoustic emission dominant frequency, acoustic emission dominant frequency offset, surrounding rock stress concentration factor, correlation between microseismic energy and acoustic emission dominant frequency, and coupling degree between surrounding rock stress and microseismic energy are extracted from each grid cell. The temporal and spatial gradient values ​​of the microseismic energy release rate, surrounding rock stress concentration factor, and acoustic emission dominant frequency offset are calculated. The weighted composite value of the microseismic energy release rate, surrounding rock stress concentration factor, acoustic emission dominant frequency offset, correlation between microseismic energy and acoustic emission dominant frequency, coupling degree between surrounding rock stress and microseismic energy, and the aforementioned temporal and spatial gradient values ​​of the microseismic energy release rate and surrounding rock stress concentration factor are used as seven key indicators, arranged sequentially to form a spatiotemporal evolutionary fingerprint feature vector. The formula for calculating the rate of change over time is: ,in The parameter value at the current moment. The parameter value at the previous sampling time. The sampling time interval is in seconds; the formula for calculating the spatial gradient value is as follows: ,in The parameter value of the current mesh cell. These are the parameter values ​​of the adjacent mesh cells in the negative x-axis direction. The grid spacing is set to 0.5m. In this embodiment, the seven key indicators ultimately formed can comprehensively cover energy release, stress concentration, frequency shift, and their spatiotemporal evolution characteristics, providing standardized feature expressions for subsequent matching. The instability identification database is established by collecting roadway roof instability case data under different geological conditions, mining depths, and support methods. Each case includes complete multimodal monitoring data and instability process records, ensuring that the database can cover a variety of complex deep mining engineering scenarios and has broad applicability. The multimodal monitoring data of each case is uniformly preprocessed according to the multi-source data preprocessing steps to extract the corresponding spatiotemporal evolution fingerprint, ensuring that the processing standards of all fingerprint data are consistent and eliminating identification errors caused by differences in data processing. K-means clustering was used to classify all extracted spatiotemporal evolution fingerprints. Three clusters were set, corresponding to low, medium, and high risk levels. Based on the actual instability and damage severity of historical cases within each cluster, the cluster with the least damage was labeled low-risk, the cluster with the most severe damage was labeled high-risk, and the rest were labeled medium-risk, ensuring that the risk level classification completely corresponds to the actual engineering damage. The classified fingerprints were stored as templates in the instability identification database. Each template contains a fingerprint feature vector and corresponding risk level information, providing a clear benchmark for risk identification of real-time monitoring data.

[0048] In the risk identification and prevention phase, the spatiotemporal evolution fingerprints extracted in real time are matched with spatiotemporal evolution fingerprint templates in the instability identification database. A weighted dynamic time warping algorithm is used to calculate the similarity, identifying the location, extent, and risk level of potentially unstable rock strata. Corresponding graded prevention and control measures are then implemented based on the risk level. The weighted dynamic time warping distance between the real-time fingerprint and each template is calculated using the weighted dynamic time warping algorithm, with the following formula: ,in, Fingerprint feature vector time series and The weighted dynamic time-normalized distance between them; For time-ordered paths; and Paths exist and Time index on; is the path length; F is the total number of features, with a value of 7. The weight of the f-th feature is determined based on the contribution of each feature to instability identification in historical instability cases. The weights are: microseismic energy release rate (0.30), surrounding rock stress concentration factor (0.30), acoustic emission dominant frequency offset (0.20), correlation between microseismic energy and acoustic emission dominant frequency (0.05), coupling between surrounding rock stress and microseismic energy (0.05), and the weighted composite value of the time variation rate of microseismic energy release rate and the spatial gradient value of the surrounding rock stress concentration factor (0.10). The weighted dynamic time warp distance is then converted into a similarity score. This algorithm effectively handles the scaling and deformation issues of time series during different instability processes, accurately matches instability precursor processes with varying evolution rates, and improves the accuracy of risk identification. Calculations showed that the similarity between the real-time fingerprint and the medium-risk template was greater than 0.85. Simultaneously, the monitored microseismic energy release rate was 260 J / h, and the surrounding rock stress concentration factor was 1.7, meeting the criteria for a level-two risk. The potential instability area was determined to be located in the middle of the roadway roof, with a range of approximately 6m × 4m × 1.5m. Based on the graded prevention and control measures corresponding to level-two risk, the pre-tightening force of all anchor bolts in the risk area was increased to 1.2 times the initial pre-tightening force, enhancing the support system's constraint on the roof strata and limiting further delamination and fracture propagation. Furthermore, monitoring was intensified by maintaining a 5x acquisition frequency, significantly improving the temporal resolution of data acquisition, enabling more intensive acquisition of roof strata state data, and timely understanding of the dynamic changes in the strata after the implementation of prevention and control measures. During continuous monitoring, the microseismic energy release rate showed a gradual decreasing trend, and the stress concentration coefficient of the surrounding rock also decreased slowly. After 48 hours, the microseismic energy release rate dropped to 92 J / h, and the stress concentration coefficient of the surrounding rock dropped to 1.4. All indicators met the risk relief conditions, indicating that the anchor bolt pre-tightening force enhancement measures effectively suppressed the rock strata's fracturing activity, and the top rock strata's state tended to be stable. At this point, the default data acquisition frequency was restored to reduce the system's operating energy consumption and data processing pressure while ensuring the effectiveness of monitoring.

[0049] This embodiment addresses the complex stress environment and concealed precursors of instability in the roof strata of deep mining roadways. It fully applies a comprehensive intelligent identification method encompassing multimodal data acquisition, multi-source data preprocessing, three-dimensional coupled field construction, evolutionary fingerprint extraction, and risk matching. Through the fusion and analysis of multi-dimensional monitoring data, it accurately identifies potential roof instability areas and risk levels, and implements targeted, tiered prevention and control measures. This effectively suppresses the further propagation of rock strata fractures, validating the method's early warning and proactive prevention capabilities for roof instability risks in deep mining projects. This provides reliable technical support for the safety management of roof strata in deep roadways. Figure 1 As shown.

[0050] Example 2:

[0051] During the multimodal data acquisition phase, a multimodal monitoring sensor network is deployed on the roof of the target roadway affected by mining activities in adjacent working faces to simultaneously acquire microseismic signals, acoustic emission signals, and surrounding rock stress data. Under the influence of mining, the roadway roof is subjected to the superposition of advance support pressure and lateral support pressure, resulting in intense rock strata activity and complex precursor signals of fracture. The multimodal monitoring sensor network can comprehensively capture the dynamic information of the rock strata from three different dimensions: energy release, frequency response, and stress changes, avoiding the information blind spots inherent in single monitoring methods. A unified clock triggering mechanism is adopted, with all monitoring devices connected to the same time synchronization network. Pulse signals trigger each sensor to simultaneously start data acquisition. During the acquisition process, the sampling frequency of each sensor remains stable, with a default acquisition frequency of one set of data every 5 minutes and each set of data acquisition lasting 10 seconds. This ensures that the microseismic, acoustic emission, and surrounding rock stress data are strictly synchronized in time, accurately reflecting the comprehensive state of the rock strata at the same moment under mining disturbance. The system acquires microseismic signals covering the 10Hz to 500Hz frequency band, acoustic emission signals covering the 100Hz to 2000Hz frequency band, and surrounding rock stress data covering the 0MPa to 30MPa range. This fully adapts to the broadband signals and large-amplitude stress changes generated by rock fractures under mining conditions, comprehensively recording all precursory information from microfracture initiation to macroscopic instability. The acquired raw data is stored in a local cache in timestamp order, with a cache capacity sufficient for 72 consecutive hours of data storage. Even if mining causes fluctuations or interruptions in the downhole network, it can completely preserve the monitoring data, ensuring the continuity of the monitoring process. The cached data is transmitted to the ground control center in real time via industrial Ethernet. During transmission, a cyclic redundancy check (CRC) method is used to ensure data integrity. A check is performed every 1024 bytes of data transmitted; if the check fails, the data segment is automatically retransmitted, preventing misjudgments and omissions due to data transmission errors, and providing reliable data support for roof safety monitoring under mining conditions.

[0052] In the multi-source data preprocessing stage, the collected multimodal data is preprocessed to remove environmental noise and support vibration interference. A weighted cross-correlation algorithm is used to achieve spatiotemporal synchronization of the multi-source data and locate the occurrence positions of different rock strata fracture signals. Under the influence of mining, there are numerous strong interference signals generated underground by coal mining machine operation, scraper conveyor vibration, blasting operations, etc. The collected microseismic signals and acoustic emission signals are decomposed into 5-layer wavelet packets, obtaining wavelet packet coefficients in 32 different frequency bands. The energy proportion of the coefficients in each frequency band is statistically analyzed, retaining the coefficients of frequency bands with an energy proportion greater than 5%, and removing high-frequency noise and low-frequency trend terms with an energy proportion less than 5%. This effectively separates the mining interference signals from the rock strata fracture signals, significantly improving the identification of the effective signals. The retained wavelet packet coefficients are reconstructed to obtain the denoised microseismic signals and acoustic emission signals, restoring the true characteristics of the rock strata fracture signals under mining disturbance. Moving average filtering was applied to the surrounding rock stress data, with a moving window size of 10 sampling points, to smooth out instantaneous fluctuations in surrounding rock stress caused by mining and clearly present the transmission and evolution of bearing pressure. A weighted cross-correlation algorithm was used to achieve spatiotemporal synchronization of multi-source data. By finding the time offset corresponding to the maximum value of the weighted cross-correlation coefficient, precise time synchronization between microseismic signals and acoustic emission signals was achieved, eliminating timing errors caused by differences in the propagation time of signals from different sensors and providing an accurate time reference for the fusion analysis of multi-source data. The distribution of weighted cross-correlation coefficients between stress sensors at different depths and the denoised signals was statistically analyzed. Based on the depth corresponding to the peak value, the source rock layer of the fracture signal was determined to be a sandstone layer 6m to 8m below the roof, accurately locating the rock fracture layer caused by mining disturbance and clarifying the vertical distribution range of high-risk rock layers, such as... Figure 2 As shown.

[0053] In the three-dimensional coupled field construction stage, a three-dimensional coupled model of the microseismic energy field, acoustic emission dominant frequency field, and surrounding rock stress field is constructed, mapping multimodal data to a unified three-dimensional spatial coordinate system. A three-dimensional rectangular coordinate system is established with the geometric center of the tunnel roof as the origin. The x-axis follows the tunnel direction, the y-axis follows the tunnel width, and the z-axis points vertically upwards. The coordinate system is perfectly consistent with the spatial orientation and geometry of the tunnel, facilitating a direct visualization of the spatial distribution characteristics of various roof parameters under mining influence. The monitoring area is divided into three-dimensional grid cells with a resolution of 0.5m × 0.5m × 0.5m. This high-resolution grid division allows for precise characterization of local stress concentration zones and energy release zones caused by mining, accurately identifying small-scale high-risk fracture areas. Each grid cell contains three scalar attributes: microseismic energy value, acoustic emission dominant frequency value, and surrounding rock stress value. Kriging interpolation is used to spatially interpolate the discrete monitoring data for each attribute, transforming the dispersed sensor monitoring data into a continuous spatial distribution field, fully presenting the energy, frequency, and stress state of the rock strata throughout the entire monitoring area. By integrating the attribute values ​​of all grid units, a three-dimensional coupled model of the microseismic energy field, acoustic emission frequency field, and surrounding rock stress field covering the entire monitoring area is constructed. This completes the unified spatial representation of multimodal data and enables the synchronous visualization of the three physical fields. It intuitively presents the spatial overlap relationship between the microseismic energy concentration area, the acoustic emission frequency anomaly area, and the surrounding rock stress concentration area under the influence of mining. The three-dimensional coupled model intuitively shows that the rupture signal is concentrated in the north-central part of the roadway roof, with a range of approximately 8m×6m×2m. This provides accurate spatial basis for subsequent precise delineation of risk areas and formulation of prevention and control measures.

[0054] In the evolutionary fingerprint database establishment phase, a pre-constructed instability identification database is invoked. This database contains spatiotemporal evolutionary fingerprint templates for different instability cases, extracting the spatiotemporal evolutionary fingerprint of energy-frequency-stress during the fracturing process of unstable rock strata. From the three-dimensional coupled model, the microseismic energy release rate, acoustic emission dominant frequency, acoustic emission dominant frequency offset, surrounding rock stress concentration factor, correlation between microseismic energy and acoustic emission dominant frequency, and coupling degree between surrounding rock stress and microseismic energy are extracted from each grid cell. The temporal and spatial gradient values ​​of the microseismic energy release rate, surrounding rock stress concentration factor, and acoustic emission dominant frequency offset are calculated. The weighted composite value of the microseismic energy release rate, surrounding rock stress concentration factor, acoustic emission dominant frequency offset, correlation between microseismic energy and acoustic emission dominant frequency, coupling degree between surrounding rock stress and microseismic energy, and the aforementioned temporal and spatial gradient values ​​of the microseismic energy release rate and surrounding rock stress concentration factor are used as seven key indicators, arranged sequentially to form a spatiotemporal evolutionary fingerprint feature vector. This database comprehensively covers the core physical characteristics of rock strata fracturing under mining influence, fully reflecting the damage evolution process of rock strata under mining stress. The instability identification database was established by collecting roof instability case data under different geological conditions, mining depths, and support methods. It includes a large number of roof instability cases under mining influence, with each case containing complete multimodal monitoring data and instability process records, ensuring the database accurately reflects the instability patterns of rock strata under mining disturbance. The multimodal monitoring data for each case undergoes unified preprocessing according to multi-source data preprocessing steps to extract corresponding spatiotemporal evolution fingerprints, ensuring complete consistency in the processing procedures and standards for all fingerprint data. K-means clustering is used to classify all extracted spatiotemporal evolution fingerprints, with three clusters corresponding to low, medium, and high risk levels. Based on the actual instability damage severity of historical cases within each cluster, the cluster with the least damage is labeled as low risk, the cluster with the most severe damage as high risk, and the rest as medium risk, ensuring a high degree of consistency between the risk level classification and the actual damage under mining influence. The classified fingerprints are stored as templates in the instability identification database. Each template contains fingerprint feature vectors and corresponding risk level information, providing a reliable comparison benchmark for real-time risk identification under mining impact.

[0055] In the risk identification and prevention phase, the spatiotemporal evolution fingerprints extracted in real time are matched with spatiotemporal evolution fingerprint templates in the instability identification database. A weighted dynamic time warping algorithm is used to calculate the similarity, identifying the location, extent, and risk level of potentially unstable rock strata. Corresponding graded prevention and control measures are then implemented based on the risk level. The weighted dynamic time warping algorithm calculates the weighted dynamic time warping distance between the real-time fingerprint and each template, converting this distance into a similarity score. This algorithm effectively adapts to the rapid and drastic changes in rock strata instability under mining influence, accurately matching instability precursor sequences with different evolution rates. Calculations show that the similarity between the real-time fingerprint and the high-risk template is greater than 0.85. Simultaneously, the monitored microseismic energy release rate is 620 J / h, and the surrounding rock stress concentration factor is 2.3, meeting the criteria for a level-three risk. This confirms that the sandstone strata in the roof of the area have experienced large-scale fracturing, posing a risk of immediate instability. Based on the graded prevention and control measures corresponding to the three-level risk, anchor cables with a length of 8m and a preload of 200kN are densely arranged at 1.5m × 1.5m intervals within the risk area to form a strong suspension structure deep into the stable rock strata, restraining the deformation and collapse of deep fractured rock layers. Grouting holes are arranged between the anchor cables, with a drilling depth of 7m. A two-component grout with a water-cement ratio of 1:1 and a cement-water glass volume ratio of 4:1 is injected, with the grouting pressure controlled between 2MPa and 3MPa. The two-component grout can quickly penetrate and fill the rock fissures, cementing the fractured rock mass and improving the integrity and bearing capacity of the rock strata. After grouting, curing is carried out for 7 days to allow the grout to fully solidify and bond tightly to the rock mass. After the curing period, a retest was conducted on the area. The results showed that the microseismic energy release rate had decreased to 85 J / h, and the stress concentration factor of the surrounding rock had decreased to 1.3. All indicators met the risk relief conditions, indicating that the anchor cable reinforcement and grouting reinforcement measures effectively repaired the fractured rock strata, and the stability of the roof was significantly improved. At this point, the default data acquisition frequency was restored, and the area was switched to routine monitoring. Figure 3 As shown.

[0056] This embodiment addresses the engineering challenges posed by mining-induced roof stress, intense rock movement, and sudden instability risks. It fully implements the intelligent identification and control process of this invention. A multi-modal monitoring system comprehensively captures dynamic information of the rock strata under mining disturbance. After preprocessing, effective signals are separated, and the specific strata of fractured rock layers are precisely located. A three-dimensional coupled model is constructed to intuitively present the spatial correlation characteristics of multi-physics fields. Spatiotemporal evolution fingerprints are extracted and quickly matched with database templates to accurately identify the range and state of high-risk instability areas. For high-risk areas, comprehensive control measures combining anchor cable reinforcement and grouting are adopted to repair fractured rock masses and improve the overall bearing capacity of the roof, ultimately achieving safe management of high-risk areas. This method can adapt to the complex and variable rock strata environment under mining influence, providing reliable technical support for the safety protection of the roof in mining-affected areas.

[0057] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for intelligent identification and prevention of potentially unstable rock strata in roadway roof, characterized in that, The specific steps of this method are as follows: S100, Multimodal Data Acquisition: A multimodal monitoring sensor network is deployed on the roof of the roadway to simultaneously acquire microseismic signals, acoustic emission signals, and surrounding rock stress data; S200, Multi-source data preprocessing: The collected multi-modal data is preprocessed to remove environmental noise and support vibration interference. The weighted cross-correlation algorithm is used to achieve spatiotemporal synchronization of multi-source data and locate the occurrence of fracture signals in different rock strata. S300, 3D Coupled Field Construction: Construct a 3D coupled model of the microseismic energy field, acoustic emission main frequency field and surrounding rock stress field, and map multimodal data to a unified 3D spatial coordinate system; S400, Evolutionary fingerprint database establishment: Call the pre-built instability identification database, which contains spatiotemporal evolution fingerprint templates of different instability cases; extract the spatiotemporal evolution fingerprint of energy-frequency-stress during the fracturing process of unstable rock strata; S500 Risk Identification and Prevention: The spatiotemporal evolution fingerprint extracted in real time is matched with the spatiotemporal evolution fingerprint template in the instability identification database. The similarity is calculated using a weighted dynamic time warping algorithm to identify the location, extent, and risk level of potentially unstable rock strata. Corresponding graded prevention and control measures are implemented according to the risk level.

2. The intelligent identification and prevention method for potential instability of roadway roof strata according to claim 1, characterized in that, In step S100, the multimodal data acquisition utilizes a unified clock triggering mechanism. All monitoring devices are connected to the same time synchronization network. Each sensor is triggered to start data acquisition simultaneously via a pulse signal. During the acquisition process, the sampling frequency of each sensor is kept stable. The default acquisition frequency is one set of data every 5 minutes, and the acquisition duration of each set of data is 10 seconds. The microseismic signal acquisition covers the frequency band from 10Hz to 500Hz, the acoustic emission signal acquisition covers the frequency band from 100Hz to 2000Hz, and the surrounding rock stress data acquisition covers the range from 0MPa to 30MPa. The acquired raw data is stored in a local cache according to the timestamp order. The cache capacity meets the data storage requirements for 72 consecutive hours. The cached data is transmitted to the ground control center in real time via industrial Ethernet. During the transmission, a cyclic redundancy check method is used to ensure data integrity. A check is performed every 1024 bytes of data transmitted. If the check fails, the data segment is automatically retransmitted.

3. The intelligent identification and prevention method for potential instability of roadway roof strata according to claim 1, characterized in that, In step S200, the noise reduction process of the multi-source data preprocessing is to perform 5-level wavelet packet decomposition on the acquired micro-vibration signal and acoustic emission signal to obtain wavelet packet coefficients of 32 different frequency bands; to count the energy proportion of the coefficients of each frequency band, retain the frequency band coefficients with an energy proportion greater than 5%, and remove high-frequency noise and low-frequency trend terms with an energy proportion less than 5%. The retained wavelet packet coefficients are reconstructed to obtain the denoised microseismic signal and acoustic emission signal. The surrounding rock stress data is then subjected to moving average filtering with a moving window size of 10 sampling points to remove random fluctuation interference in the surrounding rock stress data.

4. The intelligent identification and prevention method for potential instability of roadway roof strata according to claim 1, characterized in that, In step S200, the calculation formula for the weighted cross-correlation algorithm is as follows: in, Let be the weighted cross-correlation coefficient between signal i and signal j at a time offset of τ; This is an exponentially decaying weighting function, where α is the decay coefficient in units of s⁻¹ and has a value of 0.

01. Let be the amplitude of the denoised signal i at time t; For the denoised signal j at time... The amplitude; The signal length; By searching The maximum value corresponding to The time offset is determined to achieve time synchronization between the microseismic signal and the acoustic emission signal. The weighted cross-correlation coefficient distribution of the stress sensor and the denoised signal at different depths is statistically analyzed, and the source rock layer of the fracture signal is determined based on the depth corresponding to the peak value.

5. The intelligent identification and prevention method for potential instability of roadway roof strata according to claim 1, characterized in that, In step S300, the construction process of the three-dimensional coupling model is as follows: a three-dimensional rectangular coordinate system is established with the geometric center of the roadway roof as the origin, the x-axis along the roadway direction, the y-axis along the roadway width direction, and the z-axis vertically upward. The monitoring area is divided into three-dimensional grid units with a resolution of 0.5m×0.5m×0.5m. Each grid unit contains three scalar attributes: microseismic energy value, acoustic emission dominant frequency value, and surrounding rock stress value. The discrete monitoring data of each attribute are spatially interpolated using the Kriging interpolation method to obtain the attribute value of each grid unit. The attribute values ​​of all grid units are integrated to construct a three-dimensional coupling model of the microseismic energy field, acoustic emission dominant frequency field, and surrounding rock stress field covering the entire monitoring area, thus completing the unified spatial expression of multimodal data.

6. The intelligent identification and prevention method for potential instability of roadway roof strata according to claim 1, characterized in that, In step S400, the spatiotemporal evolution fingerprint extraction process is as follows: extracting the microseismic energy release rate, acoustic emission dominant frequency, acoustic emission dominant frequency offset, surrounding rock stress concentration factor, correlation between microseismic energy and acoustic emission dominant frequency, and coupling degree between surrounding rock stress and microseismic energy from each grid cell in the three-dimensional coupled model; and calculating the time change rate and spatial gradient value of the microseismic energy release rate, the time change rate and spatial gradient value of the surrounding rock stress concentration factor, and the time change rate and spatial gradient value of the acoustic emission dominant frequency offset; and taking the weighted comprehensive value of the microseismic energy release rate, surrounding rock stress concentration factor, acoustic emission dominant frequency offset, correlation between microseismic energy and acoustic emission dominant frequency, coupling degree between surrounding rock stress and microseismic energy, and the aforementioned time change rate of microseismic energy release rate and spatial gradient value of surrounding rock stress concentration factor as 7 key indicators, and arranging them in order to form a spatiotemporal evolution fingerprint feature vector.

7. The intelligent identification and prevention method for potential instability of roadway roof strata according to claim 1, characterized in that, In step S400, the process of establishing the instability identification database is as follows: collect roadway roof instability case data under different geological conditions, different mining depths, and different support methods. Each case contains complete multimodal monitoring data and instability process records. The multimodal monitoring data of each case is preprocessed according to the method in S200 to extract the corresponding spatiotemporal evolution fingerprint. K-means clustering was used to classify all extracted spatiotemporal evolution fingerprints. The number of clusters was set to 3, corresponding to low, medium and high risk levels. Based on the actual instability and damage of the historical cases contained in each cluster, the cluster with the least damage was marked as low risk, the cluster with the most damage was marked as high risk, and the rest were marked as medium risk. The classified fingerprints were stored as templates in the instability identification database. Each template contained fingerprint feature vectors and corresponding risk level information.

8. The intelligent identification and prevention method for potential instability of roadway roof strata according to claim 1, characterized in that, In step S500, the calculation formula for the weighted dynamic time warping algorithm is as follows: in, Fingerprint feature vector time series and The weighted dynamic time-normalized distance between them; For time-ordered paths; and Paths exist and Time index on; is the path length; F is the total number of features, with a value of 7. The weight of the f-th feature is determined based on the contribution of each feature to instability identification in historical instability cases. The weights are as follows: microseismic energy release rate (0.30), surrounding rock stress concentration factor (0.30), acoustic emission dominant frequency offset (0.20), correlation between microseismic energy and acoustic emission dominant frequency (0.05), coupling between surrounding rock stress and microseismic energy (0.05), and weighted average of the time variation rate of microseismic energy release rate and the spatial gradient of surrounding rock stress concentration factor (0.10). Convert the above weighted dynamic time-normalized distance into similarity: When the similarity S between the real-time fingerprint and the template is greater than 0.85, it is determined that there is a potential risk of instability.

9. The intelligent identification and prevention method for potential instability of roadway roof strata according to claim 1, characterized in that, In step S500, the risk level classification criteria are as follows: Level 1 risk is low risk, corresponding to a spatiotemporal evolution fingerprint with a similarity greater than 0.85 to the low-risk template, a microseismic energy release rate of less than 100 J / h, and a surrounding rock stress concentration factor of less than 1.

5. Level 2 risk is medium risk, corresponding to a spatiotemporal evolution fingerprint with a similarity greater than 0.85 with the medium risk template, and a microseismic energy release rate of 100-500 J / h, and a surrounding rock stress concentration factor of 1.5-2.0; Level 3 risk is high risk, corresponding to a spatiotemporal evolution fingerprint with a similarity greater than 0.85 to a high-risk template, a microseismic energy release rate greater than 500 J / h, and a surrounding rock stress concentration factor greater than 2.0; When the risk level determined by fingerprint similarity is inconsistent with the risk level determined by microseismic energy release rate, the risk level determined by microseismic energy release rate shall prevail. When the risk level determined by the microseismic energy release rate and the surrounding rock stress concentration factor is inconsistent, the higher risk level shall prevail.

10. The intelligent identification and prevention method for potential instability of roadway roof strata according to claim 1, characterized in that, In step S500, the tiered prevention and control measures are as follows: Level 1 Risk: Increase the data acquisition frequency to 5 times the default acquisition frequency and monitor continuously for 72 hours; if the microseismic energy release rate is consistently less than 100 J / h and the surrounding rock stress concentration factor is consistently less than 1.5 within 72 hours, then restore the default frequency. Level 2 risk: Increase the pre-tightening force of all anchor bolts in the risk area to 1.2 times the initial pre-tightening force, and maintain a 5 times sampling frequency for intensive monitoring until the microseismic energy release rate drops below 100 J / h and the stress concentration factor of the surrounding rock drops below 1.5; Level 3 risk: Within the risk zone, anchor cables with a length of 8m and a preload of 200kN are densely arranged at 1.5m×1.5m intervals. Grouting holes are arranged between the anchor cables, with a drilling depth of 7m. A two-component grout with a water-cement ratio of 1:1 and a cement-water glass volume ratio of 4:1 is injected at a grouting pressure of 2-3MPa. After curing for 7 days, the grouting is retested. If the microseismic energy release rate is confirmed to be less than 100J / h and the surrounding rock stress concentration factor is less than 1.5, the risk is eliminated.

Images