Rock burst early warning method based on multi-scale microstructure parameter fusion of coal rock mass

Abstract

The application provides an outburst early warning method based on coal rock mass multi-scale microstructure parameter fusion, which comprises the following steps: collecting coal rock mass microstructure monitoring data through a multi-scale microstructure parameter monitoring system, processing and analyzing the monitoring data, and obtaining multi-scale microstructure parameters; determining the quantitative relationship between the multi-scale microstructure parameters and the outburst occurrence risk and the core correlation mechanism; based on the quantitative relationship and the core correlation mechanism, constructing and training a multi-scale microstructure parameter fusion model; collecting real-time monitoring data of the coal rock mass to be predicted through the multi-scale microstructure parameter monitoring system and inputting the trained fusion model to obtain an outburst dynamic early warning result. The method can accurately obtain the multi-scale microstructure parameter change information of the coal rock mass from the microstructure to the macrostructure, and provide real-time and accurate outburst early warning information through a multi-source network model, thereby improving the accuracy and real-time performance of the outburst early warning.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of coal mine safety engineering, and in particular to a rock burst early warning method based on coal rock mass multi-scale microstructure parameter fusion. BACKGROUND

[0002] In coal mining operations, rock burst seriously threatens safety production. The occurrence of rock burst is closely related to the internal microstructure changes of coal rock mass. From the microscopic level of mineral crystal structure, micro-pore and micro-crack to the macro level of bedding, joint and other structures, any change in the parameters of these structures can cause stress distribution changes, and then induce rock burst. Therefore, it is necessary to monitor the microstructure parameters of coal rock mass to make rock burst early warning based on the monitoring data.

[0003] In related technologies, a single microstructure parameter monitoring technology is generally used, which cannot obtain comprehensive coal rock mass microstructure information, and the process of processing and analyzing the collected data is also relatively simple, which cannot accurately obtain the information required for rock burst early warning, resulting in a large deviation in the rock burst early warning result and failing to meet the actual early warning demand. SUMMARY

[0004] The present application aims to at least partially solve one of the technical problems in the related art.

[0005] To this end, the first object of the present application is to propose a rock burst early warning method based on coal rock mass multi-scale microstructure parameter fusion. The method constructs a comprehensive and efficient monitoring system, uses advanced data processing and analysis algorithms, accurately obtains multi-scale microstructure parameter change information of coal rock mass from the micro level to the macro level, provides a solid and reliable basis for early prediction and active prevention and control of rock burst, and thus can improve the safety of coal mining.

[0006] The second object of the present application is to propose a rock burst early warning system based on coal rock mass multi-scale microstructure parameter fusion.

[0007] The third object of the present application is to propose a non-transitory computer readable storage medium.

[0008] To achieve the above objects, the first aspect of the present application is to propose a rock burst early warning method based on coal rock mass multi-scale microstructure parameter fusion, comprising the following steps:

[0009] Through a multi-scale microstructure parameter monitoring system including micro monitoring, meso monitoring and macro monitoring, coal rock mass microstructure monitoring data is collected, and the monitoring data is processed and analyzed to obtain multi-scale microstructure parameters, wherein the multi-scale microstructure parameters include microstructure sensitivity parameters, meso mechanical response parameters and macro energy release parameters.

[0010] establish a quantitative relationship between the multi-scale microstructure parameters and the rock burst occurrence risk, and determine a core correlation mechanism between the multi-scale microstructure parameters and the rock burst occurrence risk;

[0011] Based on the quantitative relationship and the core correlation mechanism, a multi-scale microstructure parameter fusion model is constructed and trained, and the fusion model is used to output a rock burst risk probability according to input multi-scale fusion parameters;

[0012] Through the multi-scale microstructure parameter monitoring system, real-time monitoring data of the coal rock mass to be predicted is collected, and the real-time monitoring data is input into the trained fusion model to obtain a rock burst dynamic early warning result.

[0013] Optionally, in an embodiment of the present application, the establishment of the quantitative relationship between the multi-scale microstructure parameters and the rock burst occurrence risk comprises: based on historical rock burst cases, a nonlinear mapping relationship between the rock burst occurrence risk and the multi-scale microstructure parameters is established; according to the nonlinear mapping relationship, the relationship between the risk early warning triggering rule and the risk probability improvement rule and the related microstructure parameters is determined respectively; and the Shapley value analysis method is used to quantify the contribution of each microstructure parameter to the rock burst risk.

[0014] Optionally, in an embodiment of the present application, the determination of the core correlation mechanism between the multi-scale microstructure parameters and the rock burst occurrence risk comprises: according to the mechanical correlation analysis result, the correlation mechanism between the microstructure deterioration of the coal rock mass and the stress concentration effect is determined; by analyzing the meso-mechanical response parameters, the correlation mechanism between the meso-damage evolution of the coal rock mass and the energy transmission anomaly is determined; and by analyzing the macro-energy release parameters, the correlation mechanism between the macro-energy release and the instability threshold of the coal rock mass is determined.

[0015] Optionally, in an embodiment of the present application, before the multi-scale microstructure parameter fusion model is constructed, it further comprises: normalizing the multi-scale microstructure parameters; calculating the correlation degree of each microstructure parameter after normalization with the rock burst event by using the mutual information method, and calculating the Spearman correlation coefficient of each microstructure parameter according to the correlation degree; and using the Spearman correlation coefficient of each microstructure parameter, a plurality of key parameters are selected from the multi-scale microstructure parameters by using the recursive feature elimination RFE algorithm, wherein the Spearman correlation coefficient of each key parameter is greater than a preset screening threshold.

[0016] Optionally, in an embodiment of the present application, the multi-scale microstructure parameter fusion model comprises a feature extraction layer, a cross-scale fusion layer and a risk assessment layer; the feature extraction layer is a bottom layer, configured to extract data features corresponding to the key parameters from the input multi-scale microstructure parameters, wherein for microscopic image data, a convolutional neural network (CNN) is used to extract data features, and for time series data, a long short-term memory (LSTM) network is used to extract data features; the cross-scale fusion layer is a middle layer, configured to map different data features to a unified feature space and calculate the contribution weight of each data feature to rock burst; and the risk assessment layer is a top layer, configured to output a rock burst risk probability.

[0017] Optionally, in an embodiment of the present application, the microstructure sensitivity parameters comprise a plurality of micro-pore parameters, a plurality of micro-crack parameters and a plurality of mineral structure parameters, and the processing and analysis of the monitoring data to obtain multi-scale microstructure parameters comprises: performing binary processing on the microscopic image data collected by a microscopic imaging technology, and calculating the plurality of micro-pore parameters according to the binary processing result; extracting a crack center line from the microscopic image data, identifying a crack initiation position and a crack propagation trajectory from in-situ CT sequence images, and calculating the plurality of micro-crack parameters according to the obtained crack information; and performing mineral phase identification and crystal boundary change analysis in combination with the in-situ CT sequence images to calculate the plurality of mineral structure parameters.

[0018] Optionally, in an embodiment of the present application, the meso-mechanics response parameters comprise an ultrasonic attenuation coefficient and a meso-strain localization degree, and the processing and analysis of the monitoring data to obtain multi-scale microstructure parameters further comprises: processing the collected ultrasonic CT signals to calculate the ultrasonic attenuation coefficient; and analyzing the coal rock mass deformation images collected by a digital image correlation (DIC) technology to calculate the meso-strain localization degree.

[0019] Optionally, in an embodiment of the present application, the macro-energy release parameters comprise a microseismic energy cumulative value and a proportion of high-frequency components of ground sound signals, and the processing and analysis of the monitoring data to obtain multi-scale microstructure parameters further comprises: analyzing microseismic event parameters collected by a microseismic monitoring system to calculate the microseismic energy cumulative value; and analyzing ground sound signals inside the coal rock mass collected by a ground sound monitoring technology to calculate the proportion of high-frequency components of the ground sound signals.

[0020] To achieve the above purpose, a second aspect of the present application further proposes a rock burst early warning system based on multi-scale microstructure parameter fusion of coal rock mass, comprising the following modules:

[0021] A monitoring and analyzing module is configured to collect coal rock mass microstructure monitoring data through a multi-scale microstructure parameter monitoring system including microscopic monitoring, mesoscopic monitoring and macroscopic monitoring, and to process and analyze the monitoring data to obtain multi-scale microstructure parameters, wherein the multi-scale microstructure parameters include microscopic structure sensitivity parameters, mesoscopic mechanical response parameters and macroscopic energy release parameters.

[0022] A determining module is configured to establish a quantitative relationship between the multi-scale microstructure parameters and the rock burst occurrence risk, and to determine a core correlation mechanism between the multi-scale microstructure parameters and the rock burst occurrence risk.

[0023] A constructing module is configured to construct and train a multi-scale microstructure parameter fusion model based on the quantitative relationship and the core correlation mechanism, and the fusion model is configured to output a rock burst risk probability according to input multi-scale fusion parameters.

[0024] An early warning module is configured to collect real-time monitoring data of a coal rock mass to be predicted through the multi-scale microstructure parameter monitoring system, and to input the real-time monitoring data into the trained fusion model to obtain a rock burst dynamic early warning result.

[0025] In order to realize the above-mentioned embodiments, the third aspect of the present application further proposes a non-transitory computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to realize the rock burst early warning method based on coal rock mass multi-scale microstructure parameter fusion in the first aspect.

[0026] The technical scheme provided by the embodiment of the present application at least brings the following beneficial effects: the present application firstly constructs a monitoring system that comprehensively covers the multi-scale microstructure of coal rock mass, organically combines microscopic, mesoscopic and macroscopic monitoring technologies, and can obtain coal rock mass structure change information from different scales and different angles, breaking the limitation of traditional single monitoring method and providing more comprehensive and more in-depth data support for rock burst prevention and control. Then, a multi-scale microstructure parameter fusion model is constructed, a series of special processing and analysis algorithms for different monitoring data are used to accurately extract microstructure parameter change characteristics related to rock burst from massive monitoring data, efficient conversion from data to knowledge is realized, and the accuracy and reliability of rock burst prediction are significantly improved. Finally, in actual application, real-time monitoring and dynamic analysis of the multi-scale microstructure parameters of the coal rock mass can be realized, a high-speed data acquisition and transmission system is constructed, the structure change of the coal rock mass in the mining process can be captured in time, the extracted key parameters are input into the multi-scale microstructure parameter fusion model, parameter fusion and rock burst risk assessment are performed. Thus, the present application can provide accurate rock burst early warning information for coal mining, help the coal production department to take preventive measures in advance, effectively reduce the risk of rock burst, ensure the safety of coal production, and improve the production efficiency.

[0027] Additional aspects and advantages of the present application will be set forth in part in the following description, will become apparent from the following description, or will be learned through practice of the present application. BRIEF DESCRIPTION OF DRAWINGS

[0028] The above and / or additional aspects and advantages of the present application will become apparent and be readily understood from the following description, taken in conjunction with the accompanying drawings, in which:

[0029] Figure 1 A flowchart of a rock burst early warning method based on multi-scale microstructure parameter fusion of coal rock mass proposed by the embodiment of the present application;

[0030] Figure 2 A flowchart of a quantitative relationship generation method proposed by the embodiment of the present application;

[0031] Figure 3 A structure schematic diagram of a rock burst early warning system based on multi-scale microstructure parameter fusion of coal rock mass proposed by the embodiment of the present application. DETAILED DESCRIPTION

[0032] Embodiments of the present application are described in detail below with reference to the attached drawing figures, wherein the same or like component having the same or similar function is denoted by the same or like reference numerals and signs throughout the attached drawing figures. The embodiments described below with reference to the drawing figures are exemplary and are intended to explain the present application, and cannot be understood as a limitation of the present application.

[0033] It should be noted that the single coal rock mass multi-scale microstructure parameter monitoring method in the related embodiments cannot comprehensively obtain the multi-scale microstructure information of the coal rock mass. For example, conventional stress monitoring can only reflect the macro stress state and cannot in-depth analyze the microstructure level; and although the scanning electron microscope or the fault CT scanning analysis method can observe the internal microstructure of the coal rock mass, it is difficult to effectively associate with the macro mining environment and cannot realize real-time and dynamic monitoring. On the other hand, the data processing and analysis method in the related embodiments is relatively simple and rough, cannot accurately extract the microstructure parameter change characteristics directly related to the rock burst from the massive monitoring data, and thus the prediction and prevention of the rock burst lack accurate basis and are difficult to meet the actual needs of the safe and efficient mining of the coal mine.

[0034] Therefore, the present application proposes a rock burst early warning method based on coal rock mass multi-scale microstructure parameter fusion, which can accurately obtain the multi-scale microstructure parameter change information of the coal rock mass from the micro level to the macro level, and provide real-time and accurate rock burst early warning information through a multi-source network model, thereby improving the accuracy and real-time performance of the rock burst early warning.

[0035] Hereinafter, a rock burst early warning method and system based on coal rock mass multi-scale microstructure parameter fusion proposed by the embodiments of the present application are described with reference to the accompanying drawings.

[0036] Figure 1 A flowchart of a rock burst early warning method based on coal rock mass multi-scale microstructure parameter fusion proposed by the embodiments of the present application is shown in FIG. 1, which comprises the following steps: Figure 1

[0037] In step S101, the coal rock mass microstructure monitoring data is collected through a multi-scale microstructure parameter monitoring system including micro monitoring, meso monitoring and macro monitoring, and the monitoring data is processed and analyzed to obtain multi-scale microstructure parameters.

[0038] The multi-scale microstructure parameters include a plurality of microstructure sensitivity parameters, a plurality of meso mechanical response parameters and a plurality of macro energy release parameters.

[0039] ​Specifically, the application first constructs a multi-scale microstructure parameter monitoring system including micro, meso and macro monitoring, then builds a data acquisition and transmission system for data transmission, and then develops a multi-scale microstructure parameter analysis algorithm to process and analyze the various monitoring data collected by the multi-scale microstructure parameter monitoring system, so as to obtain microstructure sensitivity parameters, meso-mechanical response parameters and macro energy release parameters.

[0040] In an embodiment of the application, the multi-scale monitoring system constructed for micro monitoring uses high-resolution micro imaging technology such as field emission scanning electron microscopy (FESEM) and focused ion beam scanning electron microscopy (FIB-SEM) combined technology to observe the microstructure of the coal rock sample (resolution ≤ 10 nm), directly collects data including: spatial distribution images of micropores (such as pore size, shape and three-dimensional coordinates), initiation position and propagation trajectory of microcracks (such as crack initiation point, bifurcation point coordinates and propagation direction), morphological parameters of mineral crystals (such as crystal type, particle size, arrangement orientation), and sequence images of dynamic evolution of internal microcracks of coal rock under different environmental fields (such as stress, seepage and high and low temperature) collected by multi-field coupled in-situ CT technology (resolution 1-10 μm), and stress-strain curve original data under corresponding loading conditions. In addition, the surface topography image (accuracy up to nm level) and point-by-point mechanical parameters (such as original test value matrix of elastic modulus and hardness) of the micro area are measured by atomic force microscopy (AFM).

[0041] The multi-scale monitoring system constructed by the application for meso monitoring uses ultrasonic CT technology to arrange N ultrasonic transmitters and M receivers in the coal mining area to construct a three-dimensional monitoring network with a network spacing of 5 to 10 m. Thus, the propagation speed, attenuation and other characteristics of ultrasonic waves in the coal rock body can be directly obtained. Black and white random speckles (particle size 50-100 μm) can also be sprayed on the surface of the coal rock body as deformation tracking markers. On the other hand, the digital image correlation (DIC) technology is used to collect images before and after deformation of the coal rock body by a high-speed camera (frame rate ≥ 100 fps) to monitor the surface deformation of the coal rock body.

[0042] The multi-scale monitoring system constructed by the application for macro monitoring uses a microseismic monitoring system to arrange ≥ 6 microseismic sensors (sensitivity ≥ 100 Hz) in the entire coal mining area, uses the time difference positioning method (S-P wave arrival time difference) to calculate the three-dimensional coordinates of the microseismic event, and the positioning error is ≤ 5 m. On the other hand, combined with the ground sound monitoring technology, the ground sound signals generated by the microcrack propagation and friction in the coal rock body are captured to further supplement the macrostructure change information.

[0043] Furthermore, this embodiment also constructs a data acquisition and transmission system to collect data monitored by the aforementioned multi-scale monitoring system and transmit it to the back-end center for data analysis and processing. In specific implementation, specialized data acquisition devices are used to achieve high-speed, high-precision acquisition of various monitoring signals. For example, imaging data and mechanical test data in microscopic monitoring technology are acquired by specialized data acquisition cards; ultrasonic CT, microseismic, and ground acoustic monitoring data are acquired through dedicated data acquisition devices. Moreover, a high-speed, stable wireless transmission network is constructed to transmit the acquired data to the ground data processing center in real time. 5G communication technology combined with dedicated underground wireless relay equipment is used to ensure the reliability and low latency of data transmission, meeting the needs of real-time monitoring and analysis.

[0044] Furthermore, the collected raw monitoring data is processed and analyzed to obtain multi-scale microstructure parameters. Specifically, image recognition and machine learning algorithms can be used to process the raw monitoring data to analyze and obtain multiple micropore parameters (e.g., porosity, pore fractal dimension, average pore size, and pore size distribution probability density), multiple microcrack parameters (e.g., crack density, crack ratio, crack fractal dimension, and crack connectivity), and multiple mineral structure parameters (e.g., mineral interface debonding rate and crystal orientation consistency index). The data obtained from the preliminary analysis can also be used to analyze mechanical correlation parameters and dynamic evolution characteristics.

[0045] The following is an exemplary description of the specific process by which the monitoring data of this application is processed and analyzed to obtain microstructure parameters at various scales.

[0046] In one embodiment of this application, the obtained microstructure sensitivity parameters include: multiple micropore parameters, multiple microcrack parameters, and multiple mineral structure parameters. The monitoring data is processed and analyzed to obtain multi-scale microstructure parameters, including: binarizing the microscopic image data acquired using microscopic imaging technology and calculating multiple micropore parameters based on the binarization results; extracting the crack centerline from the microscopic image data; identifying the crack initiation location and crack propagation trajectory from in-situ CT sequence images and calculating multiple microcrack parameters based on the obtained crack information; and calculating multiple mineral structure parameters by performing mineral phase identification and combining it with in-situ CT sequence images for grain boundary change analysis.

[0047] Specifically, this embodiment first performs micropore parameter analysis. The image data acquired by the monitoring system is preprocessed, including: binarizing the FESEM / CT images using a threshold segmentation algorithm (such as the Otsu method) to separate the pores (black areas) from the solid matrix (white areas). Then, based on the three-dimensional tomographic images, the three-dimensional spatial model of the pores is reconstructed using the Marching Cubes algorithm. Based on this processing, the following parameters can be obtained through analysis and calculation:

[0048] First, porosity η. The porosity is calculated by the percentage of the pore volume (the sum of the voxels in the black region in the binary image) to the total volume of the sample, i.e. it can be calculated by the following formula:

[0049] η = V 孔隙 / V 样本 × 100%

[0050] Second, the pore fractal dimension. The box-counting method is used to cover the pore boundary with different scale grids, and the relationship between the grid number and the scale in the logarithmic coordinates is fitted, and the absolute value of the slope is the pore fractal dimension D 孔隙 , which reflects the complexity of the pore shape.

[0051] Third, the average pore size and the pore size distribution. The pores in the binary image are labeled, the equivalent diameter (equivalent circle diameter) of each pore is measured, the pore size distribution probability density f(d) is counted, and the arithmetic mean is calculated, i.e. it can be calculated by the following formula:

[0052]

[0053] The micro-pore parameter analysis is also carried out in this embodiment. The data collected by the above monitoring system is processed for crack identification and tracking, including: edge detection is performed on the FESEM / FIB-SEM image (for example, the Canny operator can be used), and the crack center line is extracted by combining the skeletonization algorithm; the in-situ CT sequence image is used to identify the crack initiation position by using the dynamic difference method (Δ gray value ≥ threshold value), the crack propagation trajectory is tracked by using the optical flow method, and the starting point, bifurcation point coordinates and propagation direction are recorded. Based on the processing process, the following parameters can be obtained by analysis and calculation:

[0054] First, the crack density ρc. The number of independent cracks per unit area (mm 2 ) is counted, and then calculated by the following formula:

[0055] ρc = N 裂纹 / A 观测面积 (threads / mm 2 )

[0056] Second, the crack rate. The total area of the cracks accounts for the percentage of the area of the observation region, and the calculation method of this parameter is similar to the calculation method of the porosity, and the pores can be replaced by the cracks.

[0057] Third, the crack fractal dimension Df. Based on the crack network skeleton, the box-counting method or the counting box dimension method is used to analyze the complexity of the spatial distribution of the cracks.

[0058] Fourth, the fracture connectivity. In the calculation process of this parameter, the proportion of the number of interconnected cracks to the total number of cracks is counted, and the connectivity of crack nodes is judged by graph theory algorithm (such as adjacency matrix).

[0059] The mineral structure parameters are also analyzed in this embodiment. Mineral phase identification and interface analysis are performed on the data collected by the monitoring system, including: mineral composition analysis is performed by using the energy dispersive spectrometer (EDS) attached to the FESEM to mark the boundaries of different mineral phases; then the grain boundaries are extracted by image segmentation technology, and the changes in the grain boundaries before and after loading the above-mentioned in-situ CT sequence data are compared to identify the debonding interface (for example, an interface spacing Δ≥2nm is considered to be debonded). Based on this processing process, the following parameters can be obtained through analysis and calculation:

[0060] First, the mineral interface debonding rate δ can be calculated by the following formula:

[0061] δ=L 脱粘晶界 / L 总晶界 ×100%

[0062] Second, the crystal orientation consistency index. In the calculation process of this parameter, based on the electron backscattering diffraction (EBSD) data, the standard deviation of the angle between the crystal c-axis direction and the loading direction is calculated, and the smaller the standard deviation, the higher the orientation consistency.

[0063] In an embodiment of the present application, the obtained mesoscopic mechanical response parameters include: ultrasonic attenuation coefficient and mesoscopic strain localization degree, the monitoring data are processed and analyzed to obtain multi-scale microstructure parameters, and the processing of the collected ultrasonic CT signals to calculate the ultrasonic attenuation coefficient and the analysis of the coal and rock mass deformation images collected by the digital image correlation method DIC to calculate the mesoscopic strain localization degree are also included.

[0064] Specifically, in this embodiment, the ultrasonic CT signals collected by the monitoring system are first processed, including band-pass filtering (the filtering range can be selected as 50-500 kHz) of the received signals, and extracting the first arrival time by using the cross-correlation method; then the wave velocity distribution inside the coal and rock mass is inverted based on the algebraic reconstruction technique (SIRT) or the iterative reconstruction technique (ART) to obtain the longitudinal wave velocity Vp and the transverse wave velocity Vs. Based on this processing process, the ultrasonic attenuation coefficient can be calculated.

[0065] The ultrasonic attenuation coefficient is defined as the signal amplitude attenuation rate per unit distance, which can be calculated by the following formula:

[0066]

[0067] The unit of this parameter is dB / m, A0 is the emission amplitude, A dThe received amplitude at a distance d. Based on the above processing process, the ultrasonic longitudinal wave velocity attenuation rate (Δv / v0, unit: %) can also be derived.

[0068] The embodiment also analyzes the parameters collected by the above DIC technology, and calculates the mesoscopic strain localization degree ε l . Among them, the displacement vector is calculated first, and the displacement vector (u, v) of each point is calculated by the sub-pixel level cross-correlation algorithm (for example, the window size is 32*32 pixels, and the overlap rate is 50%). Then calculate the strain, according to the calculated displacement field, the gradient matrix is fitted by the least square method, and each engineering strain is calculated. Finally, the mesoscopic strain localization degree ε l is calculated. In the calculation process, the area ratio of the local strain concentration area (strain≥2 times the average value) is defined, which reflects the deformation inhomogeneity.

[0069] In an embodiment of the present application, the obtained macroscopic energy release parameters include: microseismic energy cumulative value and high-frequency component proportion of ground sound signal, the monitoring data are processed and analyzed to obtain multi-scale microstructure parameters, and the microseismic event parameters collected by the microseismic monitoring system are analyzed to calculate the microseismic energy cumulative value; the internal ground sound signal of the coal rock mass collected by the ground sound monitoring technology is analyzed to calculate the high-frequency component proportion of the ground sound signal.

[0070] Specifically, the embodiment first analyzes the microseismic monitoring parameters. First, the microseismic energy is calculated, which can be calculated based on the relationship between the magnitude and the energy represented by the following formula:

[0071] E=10 1.5M+4.8

[0072] Where M is the magnitude.

[0073] Or, the kinetic energy of the velocity signal can also be directly integrated, that is, calculated by the following formula:

[0074]

[0075] Where ρ is the medium density and V is the source volume.

[0076] Then, the microseismic energy cumulative value Eacc can be calculated, which can accumulate the energy according to the time window (such as 1 hour).

[0077] The spatial distribution concentration can also be calculated by calculating the standard deviation of the microseismic event coordinates, which reflects the spatial aggregation degree of energy release.

[0078] The embodiment also analyzes the ground sound monitoring parameters. The collected ground sound signals are pre-processed, including wavelet denoising of the original ground sound signals (the sensor bandwidth corresponding to the signals is 10-200 kHz), and decomposition into different frequency layers (for example, for db4 wavelet, 5-layer decomposition is performed).

[0079] Spectrum analysis is then performed, power spectral density is calculated using fast Fourier transform (FFT), main frequency fmain is extracted, and initial state main frequency f0 is compared, so that the offset can be obtained through the following formula:

[0080] Δf = |f main -f0|

[0081] Then, the high-frequency component ratio fH is calculated, that is, the percentage of energy in the frequency band above 100 kHz to the total energy can be calculated through the following formula:

[0082]

[0083] In an embodiment of the present application, mechanical correlation analysis can also be performed based on the analysis process of each scale parameter described above. Specifically, the present embodiment first establishes a quantitative relationship between micro-porosity and fracture fractal dimension and elastic modulus (E) and hardness (H) based on the stress-strain curve involved in the data analysis of the AFM collected data and in-situ CT data, which can be represented by the following formula:

[0084] E = f(η, D f )

[0085] Thus, the coupling effect of mineral interface bonding strength and micro-crack evolution can be characterized.

[0086] The present embodiment also analyzes the dynamic evolution characteristics, including determining the micro-crack propagation rate (μm / s) and pore expansion rate (% / loading step) under multi-field coupling conditions, and the influence coefficient of different environmental fields (such as stress gradient, temperature change) on microstructure parameters (for example, determining the increase Δη of porosity per 10℃ increase in temperature).

[0087] It should be noted that the monitoring techniques used by the multi-scale monitoring system described above and the corresponding data processing and analysis methods can be adjusted according to the monitoring needs, and the present application does not limit this. In actual application, the monitoring system can be arranged according to the actual situation of the mining area, and the relevant data monitoring, collection, transmission and processing and analysis processes can be performed.

[0088] As an example, first, representative monitoring locations are selected in the coal mining area, and micro, meso, and macro monitoring equipment is arranged according to different scale monitoring needs. For example, micro sample collection points are set near the mining face, ultrasonic CT probes are arranged at certain intervals, and microseismic and ground sound sensors are uniformly distributed throughout the mining area. Second, the monitoring equipment is installed and debugged to ensure normal operation. Regular calibration and maintenance of the equipment ensure the accuracy and reliability of the monitoring data. For example, the resolution of the micro imaging equipment is calibrated regularly, the performance of the ultrasonic CT probe is checked, and the sensitivity of the microseismic and ground sound sensors is tested. Third, according to the set sampling frequency and collection mode, the data collection device is started, and the data generated by each monitoring module is collected in real time. For example, micro imaging data is collected every hour, ultrasonic CT data is collected every 10 minutes, and microseismic and ground sound data is continuously collected in real time. Fourth, the collected data is transmitted to the ground data processing center in real time through a wireless transmission network. During transmission, data encryption and verification techniques are used to ensure data security and integrity. Fifth, in the ground data processing center, the transmitted data is processed. First, the raw data is preprocessed, including denoising, filtering, data format conversion, etc. For example, wavelet denoising algorithm is used for denoising of microseismic and ground sound data. Sixth, the corresponding data processing and analysis algorithm is used to extract multi-scale microstructure parameters.

[0089] Step S102, a quantitative relationship between the multi-scale microstructure parameters and the risk of rock burst is established, and the core correlation mechanism between the multi-scale microstructure parameters and the risk of rock burst is determined.

[0090] Specifically, according to the method described in the previous step, a large number of multi-scale microstructure parameters can be obtained. By analyzing the massive multi-scale microstructure parameters under different rock burst events, a quantitative relationship between the multi-scale microstructure parameters and the risk of rock burst can be established.

[0091] For example, a multiple linear regression model or a neural network model can be constructed, and trained through historical data. The trained model generates a quantitative relationship between different scale parameters and the overall mechanical properties of coal and rock mass and the risk of rock burst. The model parameter input set is the three scale parameters described above. The microstructure sensitivity parameters are: micro porosity (η), micro crack density (pc, strip / mm 2 ), mineral interface debonding rate (δ, %), and crack fractal dimension (D f , reflecting the complexity of cracks); the meso-mechanical response parameters are: ultrasonic longitudinal wave velocity attenuation rate (Δv / v0, %), meso-strain localization degree (ε lThe local strain concentration factor measured by DIC technology; macroscopic energy release parameters: cumulative energy of microseismic events (Eacc, J), proportion of high-frequency components of ground sound signals (fH, proportion of signal energy >100kHz), and stress gradient change rate (dσ / dx, MPa / m).

[0092] To more clearly illustrate the specific implementation process of generating the above-mentioned quantitative relationship in this application, the following is an example of a method for generating a quantitative relationship proposed in one embodiment of this application. Figure 2 This is a flowchart of a method for generating quantitative relationships proposed in an embodiment of this application, as shown below. Figure 2 As shown, the method includes the following steps:

[0093] Step S201: Based on historical rockburst cases, establish a nonlinear mapping relationship between the risk of rockburst occurrence and the multi-scale microstructure parameters.

[0094] Specifically, following the method described in the above embodiments, a large amount of historical data on multi-scale microstructure parameters can be obtained. For example, multi-scale microstructure parameters and related parameters of rockburst events under historical impact cases (n=500+) can be obtained. The obtained data can be used as training data to train the above-mentioned multiple linear regression model or neural network model, thereby establishing a nonlinear mapping relationship between the risk of rockburst occurrence and the multi-scale microstructure parameters, as shown in the following formula:

[0095] R=f(η,ρ c ,δ,D f ,Δv / v0,ε l ,Eacc,fH,dσ / dx)

[0096] Furthermore, this nonlinear mapping relationship is used as a quantitative formula for rockburst risk.

[0097] Step S202: Based on the nonlinear mapping relationship, determine the relationship between the risk warning triggering rule and the risk probability enhancement rule and the relevant microstructure parameters.

[0098] Specifically, by combining the nonlinear mapping relationship established above with the patterns obtained from analyzing historical impact case data, the impact of relevant microstructure parameters on risk warning triggers and the increase in risk probability is determined.

[0099] For example, the established risk warning trigger rule is: when the microcrack density ρc > 5 cracks / mm 2and the macroscopic microseismic energy Eacc single day increase > 50%, trigger risk warning. The determined risk probability promotion rule is that the product effect exists between the mineral interface debonding rate δ and the stress gradient change rate dσ / dx, that is, when δ×dσ / dx>0.1MPa, the risk probability is promoted by 30%-50%.

[0100] Step S203, the Shapley value analysis method is used to quantify the contribution of each microstructure parameter to the rock burst risk.

[0101] Specifically, the Shapley value analysis method is used to quantify the contribution of each parameter, including verifying the basic influence of the microstructure parameters on the risk (contribution ratio 40%-60%) and the triggering effect of the macro energy parameters on the risk (contribution ratio 30%-40%). The Shapley value analysis method is a method based on game theory, and the algorithm is used to quantify the contribution of each microstructure parameter to the rock burst risk result predicted by the model. The importance of each feature is evaluated by calculating the average marginal contribution of each feature in all possible feature combinations, and the SHAP framework can be used in the quantification process.

[0102] Further, the application also determines the core associated mechanism of the rock burst occurrence risk. In an embodiment of the application, the core associated mechanism between the multi-scale microstructure parameters and the rock burst occurrence risk is determined, including the following steps:

[0103] First, according to the mechanical correlation analysis result, the correlation mechanism between the microstructure degradation of the coal and rock mass and the stress concentration effect is determined. Based on the mechanical correlation analysis in the embodiment of step S101, the core correlation mechanism determined in this step is that the micro-pore expansion (η↑) and the micro-crack penetration (Df↑) cause the stiffness reduction (elastic modulus E↓) of the coal and rock mass, forming a local stress concentration area (stress concentration coefficient K≥3), which provides the energy accumulation condition for rock burst.

[0104] Second, the correlation mechanism between the mesoscopic damage evolution of the coal and rock mass and the energy transmission anomaly is determined by analyzing the mesoscopic mechanical response parameters. The core correlation mechanism determined in this step is that when the ultrasonic attenuation rate Δv / v0>15% indicates that the mesoscopic crack network is initially formed, at this time the strain localization ε l >0.8, causes the stress transmission path to change abruptly, inducing macro stress imbalance.

[0105] Third, the correlation mechanism between the macro energy release and the instability threshold of the coal and rock mass is determined by analyzing the macro energy release parameters. The core correlation mechanism determined in this step is that when the microseismic energy Eacc exceeds 2 times the historical average for 3 consecutive days and the ground sound fH>40%, it indicates that the coal and rock mass enters the instability critical state, at this time the risk probability R≥0.7, immediate pressure relief measures need to be taken.

[0106] In step S103, a multi-scale microstructure parameter fusion model is constructed and trained based on the quantitative relationship and the core association mechanism, and the fusion model is used to output the rock burst risk probability according to the input multi-scale fusion parameters.

[0107] Specifically, according to the quantitative relationship and the core association mechanism obtained in the previous step, a multi-source network model is constructed, so that the multi-source network model can use the quantitative relationship and the core association mechanism to operate the input multi-scale fusion parameters, thereby outputting the rock burst risk probability. Wherein, a large number of historical impact cases obtained in the above embodiment can be used as training data, and each set of training data includes corresponding multi-scale microstructure parameters and rock burst event data.

[0108] In an embodiment of the present application, before constructing the multi-scale microstructure parameter fusion model, the method further includes: normalizing the multi-scale microstructure parameters; calculating the correlation degree of each microstructure parameter after normalization and the rock burst event by using the mutual information method, and calculating the Spearman correlation coefficient of each microstructure parameter according to the correlation degree; using the Spearman correlation coefficient of each microstructure parameter to screen a plurality of key parameters from the multi-scale microstructure parameters by using the recursive feature elimination (RFE) algorithm, wherein the Spearman correlation coefficient of each key parameter is greater than a preset screening threshold.

[0109] Specifically, the present embodiment performs fusion analysis on the multi-scale microstructure parameters to realize the preprocessing and feature selection of the model input data. First, the multi-scale microstructure parameters are normalized. As a possible implementation, the micro parameters (η, ρc, δ), the meso parameters (Δv / v0, εl) and the macro parameters (Eacc, fH) are standardized by Z-score to eliminate the dimension effect, and the following formula can be used for standardization:

[0110]

[0111] Then, feature screening is performed, and the mutual information method and the recursive feature elimination (RFE) algorithm are used to screen key parameters with significant correlation with rock burst. Among them, the mutual information method (Mutual Information, MI for short) is used to calculate the correlation between each microstructure parameter and the rock burst event. The mutual information method is a statistical method based on information theory, which is used to measure the mutual dependence between two random variables. Then, according to the calculation result of the mutual information method, the Spearman correlation coefficient of each microstructure parameter to the rock burst event is calculated. The Spearman correlation coefficient can measure the monotonic correlation between the two variables, and the Spearman correlation coefficient is calculated by converting the original data of the two variables into ranks. Finally, according to the Spearman correlation coefficient of each parameter, the key parameters with significant correlation with rock burst are screened through recursive feature elimination, for example, the threshold is set to be the absolute value of the Spearman correlation coefficient ≥0.3, and the parameters with the Spearman correlation coefficient greater than the threshold are retained, so that 20-30 key parameters can be retained as the core features of the subsequent multi-source network model input.

[0112] Further, a three-level fusion model is constructed. In an embodiment of the present application, the multi-scale microstructure parameter fusion model constructed includes: a feature extraction layer, a cross-scale fusion layer and a risk assessment layer. Among them, the feature extraction layer is the bottom layer, which is used to extract data features corresponding to key parameters from the input multi-scale microstructure parameters. Among them, for microscopic image data, a convolutional neural network CNN is used to extract data features, and for time series data, a long short-term memory network LSTM is used to extract data features; the cross-scale fusion layer is the middle layer, which is used to map different data features to a unified feature space and calculate the contribution weight of each data feature to rock burst; the risk assessment layer is the top layer, which is used to output the rock burst risk probability.

[0113] Specifically, the model constructed in this embodiment includes a bottom feature extraction layer, a middle cross-scale fusion layer, and a top risk assessment layer. In the bottom feature extraction layer, for microscopic image data, a convolutional neural network (CNN) is used to automatically extract geometric features such as micro-crack fractal dimension and pore shape complexity; for time-series meso- and macro- data (such as ultrasonic CT inversion parameters and microseismic event sequences), a long short-term memory network (LSTM) is used to capture the dynamic evolution law of the parameters. In the middle cross-scale fusion layer, different modal features are mapped to a unified feature space through a fully connected layer, and an attention mechanism (Attention) is introduced to calculate the contribution weight of each scale parameter to rock burst, highlighting the key influencing factors (such as the coupling effect of micro-fracture connectivity and macro-microseismic energy). In the top risk assessment layer, a Softmax classifier or a regression model is used to output the rock burst risk probability (0-1 continuous value) or risk level (I-IV level), where the risk probability ≥ 0.6 is defined as the high-risk early warning threshold.

[0114] Further, the above model can be trained, and the specific training process can refer to the training method in the related art, which is not limited in the present application. After training, model verification and dynamic correction can also be performed. As a possible implementation manner, cross-validation (5-fold) is used to evaluate the prediction accuracy of the model. In a typical mine test, the recognition accuracy of the model for rock burst is ≥ 85%, and the false positive rate is ≤ 10%. By continuously updating the training set (automatic incremental learning every week) through real-time monitoring data, the parameter weights and risk threshold are dynamically corrected, so that the coal and rock mass characteristics under different geological conditions can be adapted.

[0115] Step S104, through the multi-scale microstructure parameter monitoring system, the real-time monitoring data of the coal and rock mass to be predicted is collected, and the real-time monitoring data is input into the trained fusion model to obtain the dynamic early warning result of rock burst.

[0116] Specifically, in the actual application of underground mining site, for the current coal and rock mass to be warned of rock burst, the multi-scale microstructure parameter monitoring system and data processing and analysis method described in step S101 can be used to collect the original monitoring data of the current coal and rock mass, and the original monitoring data is processed and analyzed to obtain the real-time monitoring result of the multi-scale microstructure parameters of the coal and rock mass. The real-time monitoring result is input into the trained fusion model to obtain the dynamic early warning result of rock burst for the current coal and rock mass.

[0117] Therefore, the application realizes the cross-scale mapping from the microstructure parameters to the mesoscopic damage features to the macro energy signals by organically fusing the parameters obtained through the micro, mesoscopic and macro monitoring, and builds a closed-loop logic of parameter anomaly-risk grading-warning response, thereby providing a quantitative basis for the rock burst prediction. The multi-scale microstructure parameters extracted are input into the multi-scale microstructure parameter fusion model to perform parameter fusion and rock burst risk assessment. According to the assessment result, the rock burst warning information can be timely sent to the coal mine production department through a relevant communication mode.

[0118] In summary, the rock burst early warning method based on the multi-scale microstructure parameter fusion of the coal rock mass according to the embodiments of the application firstly builds a monitoring system that comprehensively covers the multi-scale microstructure of the coal rock mass, and organically combines the micro, mesoscopic and macro monitoring technologies. The system can obtain the coal rock mass structure change information from different scales and different angles, breaks the limitation of the traditional single monitoring method, and provides more comprehensive and more in-depth data support for the rock burst prevention and control. Then, the multi-scale microstructure parameter fusion model is built, a series of special processing and analysis algorithms for different monitoring data are used to accurately extract the microstructure parameter change features related to the rock burst from the massive monitoring data, and the efficient transformation from data to knowledge is realized, thereby significantly improving the accuracy and reliability of the rock burst prediction. Finally, the real-time monitoring and dynamic analysis of the multi-scale microstructure parameters of the coal rock mass can be realized in the actual application. Through the construction of the high-speed data acquisition and transmission system, the structure change of the coal rock mass in the mining process can be timely captured, the key parameters extracted are input into the multi-scale microstructure parameter fusion model to perform parameter fusion and rock burst risk assessment. Therefore, the method can provide accurate rock burst early warning information for the coal mining, help the coal mine production department to take preventive measures in advance, effectively reduce the risk of rock burst, ensure the safety of the coal mine production, and improve the production efficiency.

[0119] In order to realize the above-mentioned embodiments, the application further provides a rock burst early warning system based on the multi-scale microstructure parameter fusion of the coal rock mass, Figure 3 A structure diagram of a rock burst early warning system based on the multi-scale microstructure parameter fusion of the coal rock mass according to the embodiments of the application is shown in Figure 3 The system comprises a monitoring and analysis module 100, a determination module 200, a construction module 300 and a warning module 400.

[0120] The monitoring and analyzing module 100 is configured to collect coal rock mass microstructure monitoring data through a multi-scale microstructure parameter monitoring system including microscopic monitoring, mesoscopic monitoring and macroscopic monitoring, and process and analyze the monitoring data to obtain multi-scale microstructure parameters, wherein the multi-scale microstructure parameters include microscopic structure sensitivity parameters, mesoscopic mechanical response parameters and macroscopic energy release parameters.

[0121] The determining module 200 is configured to determine a quantitative relationship between the multi-scale microstructure parameters and the rock burst occurrence risk and a core correlation mechanism between the multi-scale microstructure parameters and the rock burst occurrence risk.

[0122] The constructing module 300 is configured to construct and train a multi-scale microstructure parameter fusion model based on the quantitative relationship and the core correlation mechanism, and the fusion model is configured to output a rock burst risk probability according to input multi-scale fusion parameters.

[0123] The early warning module 400 is configured to collect real-time monitoring data of a coal rock mass to be predicted through the multi-scale microstructure parameter monitoring system, and input the real-time monitoring data into the trained fusion model to obtain a rock burst dynamic early warning result.

[0124] It should be noted that the foregoing description of the embodiment of the rock burst early warning method based on multi-scale microstructure parameter fusion of a coal rock mass also applies to the system of the embodiment, and the implementation principles are the same, which will not be described here.

[0125] To sum up, the rock burst early warning system based on multi-scale microstructure parameter fusion of a coal rock mass can provide accurate rock burst early warning information for coal mining, help the coal production department to take preventive measures in advance, effectively reduce the risk of rock burst occurrence, ensure the safety of coal production, and improve the production efficiency.

[0126] In order to realize the above-mentioned embodiments, the present application further provides a non-transitory computer readable storage medium having a computer program stored thereon, and the computer program is executed by a processor to realize the rock burst early warning method based on multi-scale microstructure parameter fusion of a coal rock mass according to any one of the above-mentioned embodiments of the first aspect.

[0127] In the description of the application, reference to "one embodiment", "some embodiments", "an example", "a specific example", or "some examples" means that a particular feature, structure, material, or characteristic being described is included in at least one embodiment or example of the application. The appearances of the phrase in various places in the specification are not necessarily all referring to the same embodiment or example. Furthermore, the described specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples. Moreover, the usage of the terms "first", "second" or "third" does not limit the quantity or order of the specific features, structures, materials or characteristics, but rather the terms are used to distinguish between different sets of the same or similar features, structures, materials or characteristics. In addition, the described specific features, structures, materials or characteristics can be combined in any suitable manner in one or more embodiments or examples, for example, without reference to the specific term "combination". Furthermore, in non-contradictory embodiments or examples, different embodiments or examples described in the application, as well as features of different embodiments or examples, can be combined and combined with each other.

[0128] Furthermore, the terms "first", "second", or the like merely mean that different instances and do not imply a relative importance or a specific characteristic. Thus, a feature defined with "first", "second" or the like can include at least one of the feature. In the description of the application, the meaning of "a plurality" is at least two, for example, two, three, etc., unless otherwise explicitly specified.

[0129] Any process or method descriptions or blocks in flow charts or otherwise described herein represent embodiments which can be managed as one or more modules, segments, or portions of code which include one or more executable instructions for implementing specific logic functions or steps, and alternate implementations are possible. In some embodiments, the processes and methods described can be executably encoded on a machine- readable medium in a data signal embodied in an electromagnetic signal, a wireless signal, or a propagated signal.

[0130] The logic and / or steps represented in flow diagrams or otherwise described herein, for example, can be considered as a sequence of instructions to implement logic functions, and can be embodied in any computer-readable medium for use by an instruction execution system, apparatus, or device, such as a computer-based system, processor- containing system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions. In the context of this specification, a "computer-readable medium" can be any means that can contain, store, communicate, propagate or transport the program for use by or in connection with the instruction execution system, apparatus, or device. The computer-readable medium can be a machine-readable storage device (e.g., magnetic, optical or other) a machine-readable storage diskette (e.g., floppy disk, optical disk, CD- ROM, etc.), a machine- readable storage card (e.g., PCMCIA card, etc.), a machine-readable storage tape (e.g., magnetic tape, optical tape, etc.), a machine-readable storage medium (e.g., RAM, ROM, etc.), a machine-readable signal (e.g., electrical, optical, etc.), a machine-readable medium (e.g., carrier wave, etc.) or any other suitable medium or means of embodying the program. More specific examples (a non-exhaustive list) of the computer-readable medium include the following: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a RAM, a ROM, an EPROM, a FLASH memory card, an optical fiber, and a portable compact disc read-only memory (CD-ROM). Additionally, the computer-readable medium can be paper or another suitable medium upon which the program is printed, as the program can be electronically captured, for example, via optical scanning of the paper or other medium, then compiled, interpreted or otherwise processed in a suitable manner if necessary, and stored in a computer memory.

[0131] It should be understood that aspects of the application can be implemented in hardware, software, firmware or combinations thereof. In the above embodiments, various steps or methods can be implemented in software or firmware that is stored in memory and executed by a suitable instruction execution system. As such, in some embodiments, the steps or methods can be implemented in hardware such as a special purpose computer, a programmed microprocessor or microcontroller, a microprocessor-based or a microcontroller-based application-specific integrated circuit, a peripheral integrated circuit element, a digital signal processor, a highly-parallelized architecture or other similar or well-known computing devices. In other embodiments, the steps or methods can be implemented in software that is stored in a memory and executed on a suitable instruction execution system. In other embodiments, the steps or methods can be implemented in a combination of both software and hardware.

[0132] Those of skill in the art would understand that information and signals can be represented using any of a variety of technologies and techniques. For example, data, instructions, commands, information, signals, bits, symbols, and chips that can be referenced throughout the above description can be represented by voltages, currents, electromagnetic waves, magnetic fields or particles, optical fields or particles, or any combination thereof.

[0133] In addition, each of the functional units in the various embodiments of the present application can be integrated in one processing module, or each of the units can be physically present separately, or two or more units can be integrated in one module. The integrated module can be implemented in the form of hardware or in the form of a software functional module. When the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it can also be stored in a computer readable storage medium.

[0134] The storage medium mentioned above can be a read-only memory, a magnetic disk or an optical disk, etc. Although the embodiments of the present application have been shown and described above, it should be understood that the above embodiments are exemplary and should not be construed as limiting the present application, and those skilled in the art can make changes, modifications, replacements and variations to the above embodiments within the scope of the present application.

Claims

1. A method for early warning of rockburst based on the fusion of multi-scale microstructure parameters of coal and rock mass, characterized in that, Includes the following steps: A multi-scale microstructure parameter monitoring system, including micro-monitoring, meso-monitoring, and macro-monitoring, is used to collect coal and rock mass microstructure monitoring data. The monitoring data is then processed and analyzed to obtain multi-scale microstructure parameters, which include microstructure sensitivity parameters, meso-mechanical response parameters, and macro-energy release parameters. Establish a quantitative relationship between the multi-scale microstructure parameters and the risk of rockburst, and determine the core correlation mechanism between the multi-scale microstructure parameters and the risk of rockburst. Based on the quantitative relationship and the core correlation mechanism, a multi-scale microstructure parameter fusion model is constructed and trained. The fusion model is used to output the probability of rockburst risk according to the input multi-scale fusion parameters. The multi-scale microstructure parameter monitoring system collects real-time monitoring data of the coal and rock mass to be predicted, and inputs the real-time monitoring data into the trained fusion model to obtain dynamic early warning results of rockburst.

2. The method according to claim 1, characterized in that, The establishment of a quantitative relationship between the multi-scale microstructure parameters and the risk of rockburst includes: Based on historical rockburst cases, a nonlinear mapping relationship between the risk of rockburst occurrence and the multi-scale microstructure parameters is established. Based on the aforementioned nonlinear mapping relationship, the relationships between the risk warning triggering rule and the risk probability enhancement rule and the relevant microstructure parameters are determined respectively; The Shapley value analysis method was used to quantify the contribution of each microstructure parameter to rockburst risk.

3. The method according to claim 1, characterized in that, The core mechanism for determining the correlation between the multi-scale microstructure parameters and the risk of rockburst includes: Based on the results of mechanical correlation analysis, the correlation mechanism between the microstructural deterioration of coal and rock mass and stress concentration effect was determined; By analyzing the aforementioned micromechanical response parameters, the correlation mechanism between the microscopic damage evolution and energy transfer anomalies in coal and rock masses was determined. By analyzing the macroscopic energy release parameters, the correlation mechanism between macroscopic energy release and the instability criticality of coal and rock mass was determined.

4. The method according to claim 1, characterized in that, Before constructing the multi-scale microstructure parameter fusion model, the following is also included: The multi-scale microstructure parameters are normalized. The correlation between each normalized microstructure parameter and the rockburst event is calculated using the mutual information method, and the Spearman correlation coefficient of each microstructure parameter is calculated based on the correlation. Using the Spearman correlation coefficient of each microstructure parameter, a number of key parameters are selected from the multi-scale microstructure parameters through the recursive feature elimination (RFE) algorithm, wherein the Spearman correlation coefficient of each key parameter is greater than a preset screening threshold.

5. The method according to claim 4, characterized in that, The multi-scale microstructure parameter fusion model includes: a feature extraction layer, a cross-scale fusion layer, and a risk assessment layer; wherein, The feature extraction layer is the bottom layer, used to extract data features corresponding to the key parameters from the input multi-scale microstructure parameters. For microscopic image data, a convolutional neural network (CNN) is used to extract data features, and for time-series data, a long short-term memory network (LSTM) is used to extract data features. The cross-scale fusion layer is a middle layer used to map different data features to a unified feature space and calculate the contribution weight of each data feature to rockburst. The risk assessment layer is the top layer, used to output the probability of rockburst risk.

6. The method according to claim 1, characterized in that, The microstructure sensitivity parameters include: multiple microporosity parameters, multiple microcrack parameters, and multiple mineral structure parameters. The processing and analysis of the monitoring data to obtain multi-scale microstructure parameters includes: The microscopic image data acquired using microscopic imaging technology is binarized, and the multiple micropore parameters are calculated based on the binarization results. The crack centerline is extracted from the microscopic image data, the crack initiation location and crack propagation trajectory are identified from the in-situ CT sequence images, and the multiple microcrack parameters are calculated based on the acquired crack information. The multiple mineral structural parameters are calculated by performing mineral phase identification and analyzing grain boundary changes in conjunction with the in-situ CT sequence images.

7. The method according to claim 1, characterized in that, The micromechanical response parameters include: ultrasonic attenuation coefficient and microstrain localization degree. The processing and analysis of the monitoring data to obtain multi-scale microstructure parameters further includes: The acquired ultrasound CT signals are processed to calculate the ultrasound attenuation coefficient. The deformation images of coal and rock mass acquired using the digital image correlation (DIC) technique are analyzed to calculate the degree of localization of micro-strain.

8. The method according to claim 1, characterized in that, The macroscopic energy release parameters include: the cumulative value of microseismic energy and the proportion of high-frequency components of ground sound signals. The processing and analysis of the monitoring data to obtain multi-scale microstructure parameters also includes: The microseismic event parameters collected by the microseismic monitoring system are analyzed, and the cumulative value of the microseismic energy is calculated. The ground sound signals inside the coal and rock mass collected using ground sound monitoring technology are analyzed, and the proportion of high-frequency components of the ground sound signals is calculated.

9. A rockburst early warning system based on the fusion of multi-scale microstructure parameters of coal and rock mass, characterized in that, Includes the following modules: The monitoring and analysis module is used to collect coal and rock mass microstructure monitoring data through a multi-scale microstructure parameter monitoring system that includes micro-monitoring, meso-monitoring and macro-monitoring, and to process and analyze the monitoring data to obtain multi-scale microstructure parameters, wherein the multi-scale microstructure parameters include microstructure sensitivity parameters, meso-mechanical response parameters and macro-energy release parameters; The determination module is used to establish a quantitative relationship between the multi-scale microstructure parameters and the risk of rockburst occurrence, and to determine the core correlation mechanism between the multi-scale microstructure parameters and the risk of rockburst occurrence. A construction module is used to construct and train a multi-scale microstructure parameter fusion model based on the quantitative relationship and the core correlation mechanism. The fusion model is used to output the probability of rockburst risk based on the input multi-scale fusion parameters. The early warning module is used to collect real-time monitoring data of the coal and rock mass to be predicted through the multi-scale microstructure parameter monitoring system, and input the real-time monitoring data into the trained fusion model to obtain dynamic early warning results of rockburst.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the rockburst early warning method based on the fusion of multi-scale microstructure parameters of coal and rock mass as described in any one of claims 1-8.

Images