A hidden structure detection system for intelligent monitoring, early warning, prevention and treatment of mine water disasters

Abstract

This invention relates to the field of coal mine water hazard prevention and control technology, specifically disclosing a concealed structure detection system for intelligent monitoring, early warning, and prevention of mine water hazards. It addresses the problems of low accuracy in detecting concealed structures in mines, delayed mine water hazard early warning, and large errors in simulating seismic wave propagation paths. The system includes a multi-source data integrated detection module, a real-time imaging module, a microseismic multi-scale intelligent analysis module, a seismic wave signal identification module, a regional difference analysis module, and an intelligent data processing module. By integrating multi-source data, real-time imaging, microseismic analysis, seismic wave identification, and intelligent data processing technologies, this invention achieves accurate detection of concealed structures in mines, high-precision prediction of mine earthquakes and water hazards, and real-time risk early warning, thereby improving the timeliness and accuracy of mine water hazard prevention and control.

Description

Technical Field

[0001] This invention relates to the field of coal mine water hazard prevention and control technology, and more specifically, to a concealed structure detection system for intelligent monitoring, early warning and prevention of mine water hazards. Background Technology

[0002] Coal resources are one of my country's main energy sources, abundant and widely distributed. However, the hydrogeological conditions vary greatly across different coalfields, leading to diverse and complex mine water hazards. Therefore, mine water hazard prevention and control is a crucial task that needs to be addressed. Hidden geological structures are underground or subsurface geological formations or phenomena that are difficult to detect directly, yet they play a vital role in the formation of mine water inrush channels. Factors such as deep strata, surface cover, and geological diversity can all make hidden structures difficult to predict. Therefore, intelligent detection of hidden structures has become a significant challenge in current mine water hazard prevention and control. Chinese invention patent application number 2020106010182 discloses a comprehensive advanced detection device and method for hidden water hazards at the bottom of boreholes. This device utilizes a time-domain electromagnetic detection system with multi-source variable current combined transmission and multi-component parallel reception, ensuring that the equipment does not need to be moved. This technology allows for comprehensive observation of low-resistivity anomalies in front of the equipment, ultimately enabling advanced geological prediction of hidden water hazards at the bottom of the borehole. It addresses the limitations of insufficient borehole space in coal mines, the inability of detection equipment to move or rotate in a plane, and the inability to conduct two-dimensional or three-dimensional observations to obtain sufficient background data. This lays the foundation for detailed analysis of the distribution of geological anomalies ahead. However, with the development and advancement of detection technology, more and more advanced intelligent detection methods are being introduced to help study and understand hidden structures more accurately. The identification and regional differences in seismic wave signals involve multiple factors, including underground geological structure, seismic wave propagation, signal processing, and data analysis, leading to signal variations. The underground environment can interfere with signals, and different underground structures cannot be distinguished by current equipment. This hinders timely monitoring and early warning of mine water hazards caused by hidden structures, preventing timely preventative measures and reducing casualties and engineering losses. Summary of the Invention

[0003] To achieve the above objectives, the present invention provides the following technical solution: A concealed geological structure detection system for intelligent monitoring, early warning, and prevention of mine water hazards includes a multi-source data integrated detection module, a real-time imaging module, a microseismic multi-scale intelligent analysis module, a seismic wave signal identification module, a regional difference analysis module, and an intelligent data processing module. The multi-source data integrated detection module employs differential GPS positioning technology, integrates Kalman filtering and adaptive noise reduction algorithms, and utilizes Kriging interpolation and joint inversion methods to construct a three-dimensional geological structure model. The real-time imaging module constructs real-time subsurface structure images through an adaptive nonlinear regularized inversion method and uses GPU parallel computing to accelerate the imaging process. The microseismic multi-scale intelligent analysis module utilizes the ICEEMDAN mode decomposition algorithm and consistent... A random sampling strategy and a rapid sample entropy assessment method are combined with variational mode decomposition and Autoformer model multi-step prediction of mine-related microseismic events. The regional difference analysis module employs short-time Fourier transform, adaptive wavelet analysis, and high-order difference hybrid numerical methods to simulate and analyze the regional characteristics of mine earthquake propagation. The intelligent data processing module integrates multi-source geophysical data, real-time dynamic imaging data, microseismic prediction data, and spatial difference data of seismic wave propagation. It uses a deep learning intelligent risk assessment model, dynamic weighted ensemble learning method, and attention mechanism to provide real-time dynamic assessment and automatic early warning of hidden geological structures and mine water hazard risks. The specific input-output relationships between the modules are as follows: The multi-source data integrated detection module takes GPS positioning data, geophysical exploration data, and electromagnetic interference data as inputs and outputs a high-precision three-dimensional geological structure model. The real-time imaging module takes a three-dimensional geological structure model and resistivity monitoring data as input, and outputs a real-time underground resistivity imaging image, which finely displays the underground concealed structural areas. The input to the microseismic multi-scale intelligent analysis module is the raw data of mine microseismic monitoring, and the output is multi-scale prediction data of mine microseismic events, including the energy, frequency of occurrence and spatial location of microseismic events. The input to the seismic wave signal identification module is the real-time acquired mine seismic wave propagation signal, and the output is the identified seismic wave characteristic signal, including the source location, magnitude, and propagation path characteristics. The input to the regional difference analysis module is the seismic wave characteristic signal output by the seismic wave signal identification module and the geological parameters of each region. The output is the spatial difference characteristic data of the seismic wave propagation path. The intelligent data processing module takes into account a three-dimensional geological structure model, real-time imaging images, microseismic prediction data, and spatial difference data of seismic wave propagation. Its output includes real-time dynamic assessment results of hidden structures and mine water hazard risks, as well as automatic early warning information.

[0004] As a further aspect of the present invention, the multi-source data integrated detection module acquires GPS positioning data, geophysical exploration data, and electromagnetic interference data in real time. Based on adaptive spatial interpolation technology, it automatically detects local errors in the underground three-dimensional geological model of the mine in real time, dynamically updates the interpolation grid resolution, and quickly repairs missing data areas in the three-dimensional geological model.

[0005] As a further aspect of the present invention, in the multi-source data integrated detection module, the local errors of the three-dimensional geological model include, but are not limited to, the first local error caused by the lack of data acquisition due to sensor failure, and the second local error caused by the lack of data due to local geological disturbances caused by encountering complex geological structures and mining activities. For the local error caused by the lack of data acquisition due to sensor failure, the first error is repaired by real-time monitoring of the sensor status and automatic activation of a redundant interpolation method that fuses adjacent sensor data within a set range. The second error is repaired by real-time detection of geological structure disturbance areas and the establishment of a targeted interpolation model using disturbance characteristics.

[0006] As a further aspect of the present invention, in the multi-source data integrated detection module, during the repair of the first error, data from four to six sensors closest to the faulty sensor are selected for weighted fusion interpolation centered on the faulty sensor; during the repair of the second error, fault fracture feature matching interpolation model, goaf stress release dynamic interpolation model, and groundwater seepage feature interpolation model are established according to the spatial distribution characteristics of faults, fracture zones, goaf areas, and groundwater bodies, respectively.

[0007] As a further aspect of the present invention, during the process of repairing the second error in the multi-source data integrated detection module: The fault fracture feature matching interpolation model identifies the spatial distribution characteristics, fracture density, and extension direction of fault fracture zones. It then utilizes anisotropic interpolation algorithms to achieve refined spatial trend repair of missing data areas. The repair formula is as follows:

[0008] In the formula: , These are the x and y coordinates of the location to be interpolated, respectively, with the reference coordinate system using a benchmark point set within the detection area as its origin. The positive direction of the axis points towards the direction of the main mine roadway. The positive axis is perpendicular to the main tunnel and extends to the right. The value to be inserted at the location in the fault fracture feature matching interpolation model. For the index of the known data points, The number of known data points. The anisotropy weighting coefficients are obtained by calculating the cosine similarity between the fracture density and the fracture direction vector. , For the point to be interpolated to the th The angle between the crack directions of the known data points For the first Crack density at known data points For the index variable in the summation process, For the first The angle between the interpolation point and the crack direction of the known data point For the first Crack density at known data points For the first Observations at known data points; The dynamic interpolation model for stress release in goaf areas is based on real-time monitoring of stress changes in the surrounding rock strata. A dynamic interpolation function is established to reflect real-time changes in the geological disturbance area caused by stress release, achieving accurate data supplementation. The supplementation formula is as follows:

[0009] in: For the current moment, Current position in the dynamic interpolation model for stress release in the goaf Current moment interpolation, This is the initial state data. The stress sensitivity coefficient is obtained through the properties of the rock strata. For the current moment Real-time monitoring of rock strata stress changes; The groundwater seepage characteristic interpolation model analyzes dynamic monitoring data of groundwater seepage path, velocity, and groundwater level. Using flow-oriented interpolation technology, it supplements missing data in the groundwater-affected area in real time. The repair formula is as follows:

[0010] In the formula: For a moment The groundwater level to be interpolated. , , The x-coordinate, y-coordinate, and initial time of the initial water level monitoring point are given. The initial water level, The groundwater flow field guiding factor is determined by real-time flow field monitoring data. For integration variables, Groundwater flow velocity, This represents the groundwater level gradient.

[0011] As a further aspect of this invention, the real-time imaging module is based on dynamic adaptive grid refinement technology. It performs differential processing on the resistivity data sequence acquired at consecutive time intervals, calculates the resistivity change gradient between adjacent time intervals, identifies regions with abnormal resistivity trends, and judges the significance of resistivity anomalies in real time by setting gradient thresholds. It uses the adaptive differential gradient method to accurately identify the boundary positions of regions with abnormal resistivity, and dynamically adjusts the grid resolution based on the magnitude and spatial distribution range of the gradient values ​​in the abnormal regions. The grid size is automatically refined in regions where the gradient values ​​are within the set first-level threshold range, and the grid size is automatically enlarged in regions where the gradient values ​​are within the set second-level threshold range. The grid resolution and refinement region are dynamically adjusted and optimized in real time, enabling real-time and precise capture and identification of hidden structures.

[0012] As a further aspect of the present invention, the seismic wave signal identification module acquires real-time collected mine seismic wave propagation signals, identifies the source location, magnitude, and propagation path characteristics of the seismic waves, and transmits the identified characteristics to the regional difference analysis module. As a further aspect of the present invention, the microseismic multi-scale intelligent analysis module dynamically determines the frequency and amplitude of threshold updates by calculating the sample entropy change trend of the data sequence within the sliding window in real time. It uses the real-time sample entropy threshold to guide a consistent random sampling strategy. When the sample entropy exceeds the current dynamic threshold, it shortens the distance between sampling bands, increases the local sampling density, and adds an additional 20% to 30% of the number of sampling points in that area. When the sample entropy is lower than the current dynamic threshold, it extends the distance between sampling points and reduces the number of sampling points to 60% to 80% of the original design value, thus adaptively and dynamically adjusting the spatial distribution and density of sampling points.

[0013] As a further aspect of this invention, the regional difference analysis module acquires real-time data on the elastic modulus, Poisson's ratio, density, and medium damping parameters of rocks in different areas of the mine, as well as real-time seismic wave propagation data. It then establishes geological characteristic models for each region, uses the real-time seismic wave propagation data to solve for the spatial partial derivatives of the continuously acquired seismic wave data, obtains the wavefield spatial gradient distribution, analyzes local wavefield errors in real-time based on the wavefield spatial gradient, sets a dynamic adjustment threshold, and evaluates local Courant stability conditions in real-time by analyzing local wave velocity differences and seismic waveform frequency variation trends. The upper limit of the time step is automatically calculated from the real-time seismic wave velocity data, and the optimal time step that meets stability requirements is dynamically selected to accurately simulate the spatial differentiation characteristics of the seismic wave propagation path.

[0014] As a further aspect of this invention, the intelligent data processing module monitors the rate of change of input data features in real time using a sliding window method and calculates the degree of anomaly of the features using a real-time updated exponentially weighted moving average method. Based on the current degree of anomaly, it adjusts the update frequency and weight distribution density of the spatial weight matrix in real time. Through a hierarchical attention mechanism, it realizes hierarchical interaction between local and global features. The local attention layer quickly responds to the location where anomalies occur, and the global attention layer dynamically coordinates the attention intensity of each local feature. By using real-time feedback of latent high-risk area data, it triggers an adaptive attention adjustment algorithm to quickly focus on the most critical risk area.

[0015] Technical advantages of the concealed structure detection system for intelligent monitoring, early warning and prevention of mine water hazards in this invention: This invention utilizes multi-scale data fusion and real-time dynamic analysis techniques, integrating differential GPS positioning, Kalman filtering, adaptive noise reduction, Kriging interpolation, joint inversion, adaptive nonlinear regularized inversion, GPU parallel computing, ICEEMDAN mode decomposition, consistent random sampling strategy, variational mode decomposition, and Autoformer deep learning model. It also incorporates short-time Fourier transform, adaptive wavelet analysis, high-order differential hybrid numerical methods, and dynamic weighted learning and attention mechanisms for intelligent data processing. This enables precise detection of hidden structures and real-time dynamic prediction and early warning of mine water hazards, significantly reducing casualties and economic losses, and ensuring safe mine production and smart mine construction. Attached Figure Description

[0016] Figure 1 This is a system block diagram of the present invention; Figure 2 This is a flowchart of the modal decomposition of the present invention; Figure 3 The process of constructing a multi-feature input-output dataset for this invention; Figure 4 This is a diagram of the Autoformer model architecture of the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described content is only a part of the present invention, and not all of it. Based on the content of the present invention, all other technical solutions obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] like Figure 1As shown, the concealed structure detection system for intelligent monitoring, early warning, and prevention of mine water hazards proposed in this invention includes a multi-source data integrated detection module, a real-time imaging module, a microseismic multi-scale intelligent analysis module, a seismic wave signal identification module, a regional difference analysis module, and an intelligent data processing module. The multi-source data integrated detection module employs differential GPS positioning technology, integrates Kalman filtering and adaptive noise reduction algorithms, and utilizes Kriging interpolation and joint inversion methods to construct a three-dimensional geological structure model. The real-time imaging module constructs real-time underground structure imaging through an adaptive nonlinear regularized inversion method and uses GPU parallel computing to accelerate the imaging process. The microseismic multi-scale intelligent analysis module utilizes ICEE... The MDAN mode decomposition algorithm, consistent random sampling strategy, and sample entropy fast evaluation method are combined with variational mode decomposition and Autoformer model multi-step prediction of mine seismic microseismic events. The regional difference analysis module uses short-time Fourier transform, adaptive wavelet analysis technology, and high-order difference hybrid numerical method to simulate and analyze the regional characteristics of mine earthquake propagation. The intelligent data processing module integrates multi-source geophysical data, real-time dynamic imaging data, microseismic prediction data, and seismic wave propagation spatial difference data. It adopts a deep learning intelligent risk assessment model, dynamic weight ensemble learning method, and attention mechanism to conduct real-time dynamic assessment and automatic early warning of hidden structures and mine water hazard risks. like Figure 2 The diagram shows the flowchart of mode decomposition in the microseismic multi-scale intelligent analysis module. Before constructing the dataset, the original data is preprocessed. By performing secondary mode decomposition on the preprocessed dataset, signals can be analyzed at multiple scales, and detailed information from different frequency bands can be extracted, resulting in several characteristic mode functions (EMFs). Subsequently, these EMFs are combined with the time-series data of the original dataset to obtain a multivariate characteristic time-series sample set. Then, as... Figure 3 As shown, a sliding window method is used to construct a multivariate feature input-output dataset. The sliding window method divides the data into multiple subsequences by sliding a fixed-length window across the time series, forming sample pairs. These subsequences serve as the model input, and combined with the corresponding output labels, constitute the final dataset. This multivariate feature time series sample set is used to construct a multivariate feature input-single-output sample dataset.

[0019] First, the original time-series data is decomposed using ICEEMDAN. ICEEMDAN employs multiple noise additions and adjustments to enhance the robustness of the decomposition results against noise and adaptively adjusts the noise and mode extraction process, thereby reducing mode aliasing and resulting in a more defined frequency distribution for each mode. ICEEMDAN decomposition yields several C-IMF sequences. Next, a selective sampling method is used to uniformly and randomly sample each C-IMF sequence, calculating the sample entropy. Generally, a higher sample entropy indicates a more complex and disordered signal with higher frequency components; conversely, a lower sample entropy indicates a more stable signal with lower frequency components or a stronger trend. Based on the sample entropy, the C-IMF sequences are reconstructed into high-frequency, mid-frequency, and low-frequency sequences. Subsequently, VMD decomposition is performed on the reconstructed high-frequency, mid-frequency, and low-frequency sequences. The effectiveness of VMD decomposition is highly dependent on the choice of the number of modes. When the number of modes is too small, the VMD algorithm may behave as an adaptive filter bank, causing key information in the original signal to be filtered out, thus reducing prediction accuracy. Conversely, when the number of modes is too large, the center frequencies of adjacent mode components are close, potentially leading to redundancy or additional noise. This paper initially estimates the number of decompositions K based on the sample entropy of each sequence and selects an appropriate number of decompositions based on the center frequency distribution under different decomposition numbers. Finally, n V-IMF sequences are obtained through decomposition. These V-IMF sequences are combined with the original data features to obtain n+1 multivariate feature time series sample sets. Next, the sliding window method is used to construct the dataset. The specific method is as follows: First, a time window ∆t (i.e., time step) is set, and a sliding step S is defined. Within each sliding window, the first ∆t time series values, along with the corresponding V-IMF component data, are used as input features for the model. Subsequently, at the end of each time window, the data values ​​of the next time step or multiple subsequent time steps are used as output. Through this process, the model will learn how to predict one or more future time series values ​​based on the features of the first ∆t time steps.

[0020] like Figure 4The diagram shows the architecture of the Autoformer model used in this invention. Specifically, the model decomposes the input time series into trend and seasonal components. This decomposition process utilizes a sliding window for smoothing to reduce noise interference with the model. The input features of the encoder and decoder are embedded in a high-dimensional space. Each encoder module consists of multiple encoding layers responsible for extracting time series features. Each encoder layer includes an autocorrelation layer, a feedforward neural network, a LayerNormalization layer, and a Dropout layer. The autocorrelation layer is used to capture long-term dependencies in the time series; the feedforward neural network consists of two fully connected layers combined with a non-linear activation function to enhance feature representation capabilities; the LayerNormalization layer and Dropout layer help stabilize the training process and prevent overfitting. Each decoder module also consists of multiple decoding layers, each including an autocorrelation layer and a cross-autocorrelation layer, the interaction between the decoder layer and the encoder layer, and the reconstruction of trend and seasonal components. In long-term series prediction tasks, the decoder output is combined with the encoder features to reconstruct the trend and seasonal components, ultimately generating the prediction result.

[0021] The input-output relationships between the modules are as follows: The multi-source data integrated detection module takes GPS positioning data, geophysical exploration data, and electromagnetic interference data as inputs and outputs a high-precision three-dimensional geological structure model. The real-time imaging module takes a three-dimensional geological structure model and resistivity monitoring data as input, and outputs a real-time underground resistivity imaging image, which finely displays the underground concealed structural areas. The input to the microseismic multi-scale intelligent analysis module is the raw data of mine microseismic monitoring, and the output is multi-scale prediction data of mine microseismic events, including the energy, frequency of occurrence and spatial location of microseismic events. The input to the seismic wave signal identification module is the real-time acquired mine seismic wave propagation signal, and the output is the identified seismic wave characteristic signal, including the source location, magnitude, and propagation path characteristics. The input to the regional difference analysis module is the seismic wave characteristic signal output by the seismic wave signal identification module and the geological parameters of each region. The output is the spatial difference characteristic data of the seismic wave propagation path. The intelligent data processing module takes into account a three-dimensional geological structure model, real-time imaging images, microseismic prediction data, and spatial difference data of seismic wave propagation. Its output includes real-time dynamic assessment results of hidden structures and mine water hazard risks, as well as automatic early warning information.

[0022] To address the technical problems raised in the background section, the concealed structure detection system for intelligent monitoring, early warning, and prevention of mine water hazards proposed in this invention has the following workflow: (1) Multi-source data integrated detection module: Differential GPS positioning technology is adopted, Kalman filtering and adaptive noise reduction algorithm are integrated to process the input GPS positioning data, geophysical exploration data and electromagnetic interference data, and Kriging interpolation and joint inversion method are used to construct a high-precision three-dimensional geological structure model; (2) Real-time imaging module: Taking the three-dimensional geological structure model and real-time monitored resistivity data as input, the module constructs real-time underground resistivity imaging images through an adaptive nonlinear regularization inversion method, and uses GPU parallel computing to accelerate the imaging process, and accurately identifies and displays underground hidden structural areas. (3) Microseismic Multi-scale Intelligent Analysis Module: Taking the raw data of mine microseismic monitoring as input, the module uses the ICEEMDAN mode decomposition algorithm, the consistent random sampling strategy and the sample entropy fast evaluation method, combined with variational mode decomposition and Autoformer deep learning model to realize multi-scale prediction of mine microseismic events and output the energy, frequency of occurrence and spatial location information of microseismic events. (4) Seismic wave signal identification module: Input the real-time acquired mine seismic wave propagation signal, and output the identified seismic wave feature signal through signal processing and feature extraction, including source location, magnitude and propagation path characteristics; (5) Regional difference analysis module: Taking the seismic wave characteristic signal output by the seismic wave signal identification module and the geological parameters of each region as input, the module uses short-time Fourier transform, adaptive wavelet analysis and high-order difference hybrid numerical method to simulate and analyze the regional characteristic differences of seismic wave propagation in the mine, and outputs the spatial difference characteristic data of seismic wave propagation path. (6) Intelligent data processing module: It integrates multi-source geophysical data, real-time dynamic imaging data, microseismic prediction data and seismic wave propagation spatial difference data as input, and uses deep learning intelligent risk assessment model, dynamic weight ensemble learning method and attention mechanism to realize real-time dynamic assessment and automatic early warning of hidden structure and mine water hazard risk, and outputs risk assessment results and early warning information.

[0023] It should be noted that the multi-source data integrated detection module acquires GPS positioning data, geophysical exploration data, and electromagnetic interference data in real time. Based on adaptive spatial interpolation technology, it automatically detects local errors in the underground three-dimensional geological model of the mine in real time, dynamically updates the interpolation grid resolution, and quickly repairs missing data areas in the three-dimensional geological model.

[0024] It should be further explained that in the multi-source data integrated detection module, the local errors of the three-dimensional geological model include, but are not limited to, the first local error caused by the lack of data acquisition due to sensor failure, and the second local error caused by the lack of data due to local geological disturbances caused by encountering complex geological structures and mining activities. For the local error caused by the lack of data acquisition due to sensor failure, the first error is repaired by real-time monitoring of sensor status and automatic activation of redundant interpolation method for fusing adjacent sensor data within a set range. The second error is repaired by real-time detection of geological structure disturbance areas and the establishment of a targeted interpolation model using disturbance characteristics.

[0025] It should be further explained that in the multi-source data integrated detection module, during the repair of the first error, data from four to six sensors closest to the faulty sensor are selected for weighted fusion interpolation. During the repair of the second error, fault fracture feature matching interpolation model, goaf stress release dynamic interpolation model, and groundwater seepage feature interpolation model are established according to the spatial distribution characteristics of faults, fracture zones, goaf areas, and groundwater bodies, respectively.

[0026] What needs further explanation is that during the process of the multi-source data integrated detection module repairing the second error: The fault fracture feature matching interpolation model identifies the spatial distribution characteristics, fracture density, and extension direction of fault fracture zones. It then utilizes anisotropic interpolation algorithms to achieve refined spatial trend repair of missing data areas. The repair formula is as follows:

[0027] In the formula: , These are the x and y coordinates of the location to be interpolated, respectively, with the reference coordinate system using a benchmark point set within the detection area as its origin. The positive direction of the axis points towards the direction of the main mine roadway. The positive axis is perpendicular to the main tunnel and extends to the right. The value to be inserted at the location in the fault fracture feature matching interpolation model. For the index of the known data points, The number of known data points. The anisotropy weighting coefficients are obtained by calculating the cosine similarity between the fracture density and the fracture direction vector. , For the point to be interpolated to the th The angle between the crack directions of the known data points For the first Crack density at known data points For the index variable in the summation process, For the first The angle between the interpolation point and the crack direction of the known data point For the first Crack density at known data points For the first Observations at known data points; The dynamic interpolation model for stress release in goaf areas is based on real-time monitoring of stress changes in the surrounding rock strata. A dynamic interpolation function is established to reflect real-time changes in the geological disturbance area caused by stress release, achieving accurate data supplementation. The supplementation formula is as follows:

[0028] in: For the current moment, Current position in the dynamic interpolation model for stress release in the goaf Current moment interpolation, This is the initial state data. The stress sensitivity coefficient is obtained through the properties of the rock strata. For the current moment Real-time monitoring of rock strata stress changes; The groundwater seepage characteristic interpolation model analyzes dynamic monitoring data of groundwater seepage path, velocity, and groundwater level. Using flow-oriented interpolation technology, it supplements missing data in the groundwater-affected area in real time. The repair formula is as follows:

[0029] In the formula: For a moment The groundwater level to be interpolated. , , The x-coordinate, y-coordinate, and initial time of the initial water level monitoring point are given. The initial water level, The groundwater flow field guiding factor is determined by real-time flow field monitoring data. For integration variables, Groundwater flow velocity, This represents the groundwater level gradient.

[0030] The multi-source data integration and detection module acquires GPS positioning data, geophysical exploration data, and electromagnetic interference data in real time. Based on adaptive spatial interpolation technology, it dynamically detects and repairs local errors in the three-dimensional geological model of the mine's underground. In specific execution, for example, during exploration in a certain mine area, the system detects a malfunction in sensor S10, resulting in data loss. The system automatically selects data from the six nearest sensors S7, S8, S9, S11, S12, and S13, and uses a weighted fusion interpolation method to repair the missing areas, rapidly restoring the model data integrity from 85% to 100%. Simultaneously, when the exploration area encounters complex fault fracture zones, the spatial location and density changes of the fracture zones are determined in real time through geophysical exploration, and anisotropic interpolation methods are used to repair the missing data areas. In the example, the initial fracture density at location (50, 100) in the fracture zone is 0.8, with an azimuth angle of 30°. Through model calculation, the real-time interpolation repair accuracy reaches over 0.95. In the goaf area, taking a goaf location (120, 150) as an example, the initial rock stress data was 5 MPa. Real-time monitoring showed the rock stress change to 0.8 MPa. After calculation using a dynamic interpolation model, the interpolated data was accurately updated to 5.64 MPa. For the groundwater seepage area, at the initial monitoring location (200, 250), the initial water level was 10 m. Real-time monitoring showed a flow velocity of 0.05 m / s and a water level gradient of 0.002 m / m. The flow field guidance factor was determined to be 0.95. After real-time calculation and interpolation correction, the accuracy of the water level data was improved to over 99%.

[0031] This invention acquires GPS positioning data, geophysical exploration data, and electromagnetic interference data in real time through a multi-source data integrated detection module. Based on adaptive spatial interpolation technology, it dynamically detects and repairs local errors in the underground three-dimensional geological model of the mine. Redundant data interpolation and feature matching interpolation methods are used to address sensor failures and disturbances in complex geological structures, including fault fracture feature matching, dynamic interpolation of stress release in goaf areas, and groundwater seepage feature interpolation models. This enables real-time and refined supplementation of missing data areas, effectively improving the detection accuracy of hidden structures, significantly enhancing the real-time prediction and early warning capability of mine water hazards, greatly reducing mine safety risks, and ensuring safe production in the mine.

[0032] It should be noted that the real-time imaging module is based on dynamic adaptive grid refinement technology. It performs differential processing on the resistivity data sequence acquired at continuous time intervals, calculates the resistivity change gradient between adjacent time intervals, identifies regions with abnormal resistivity trends, and judges the significance of resistivity anomalies in real time by setting gradient thresholds. It uses the adaptive differential gradient method to accurately identify the boundary position of the resistivity anomaly region, and dynamically adjusts the grid resolution based on the gradient value and spatial distribution range of the anomaly region. The grid size is automatically refined in regions where the gradient value is within the set first-level threshold range, and the grid size is automatically enlarged in regions where the gradient value is within the set second-level threshold range. The grid resolution and refinement region are dynamically adjusted and optimized in real time, enabling real-time and precise capture and identification of hidden structures.

[0033] In a specific implementation example, real-time monitoring was conducted in the main roadway area of ​​the mine. Resistivity data was continuously collected at two time points. At location (80, 120), the resistivity decreased from an initial 20 Ω·m to 15 Ω·m, a resistivity gradient change of 0.5 Ω·m / m, exceeding the set primary threshold of 0.4 Ω·m / m. The system automatically refined the grid size of this area to 50% of its original size, improving the grid accuracy from 1 m to 0.5 m, effectively enhancing the identification accuracy of abnormal areas. Simultaneously, at location (180, 200), the resistivity gradient change was only 0.2 Ω·m / m, below the secondary threshold of 0.3 Ω·m / m. The system automatically enlarged the grid size to 150% of its original size, i.e., 1.5 m, reducing unnecessary computational resource waste. Through these dynamic adjustments, the system accurately captured and identified concealed structural areas in the mine in real time, improving detection accuracy to over 98%. This significantly enhanced the mine's safety risk prediction and early warning capabilities, reducing the probability of on-site safety accidents by more than 30%.

[0034] The real-time imaging module employs dynamic adaptive grid refinement technology. By processing resistivity data sequences in real time and calculating the gradient of change, it accurately identifies resistivity anomaly regions and their boundary locations. Based on the gradient value and spatial distribution, it dynamically adjusts the grid resolution in real time, achieving efficient and accurate capture and real-time fine imaging of concealed structures. This effectively improves the detection accuracy of concealed structures, enhances the ability to predict and warn of mine water hazards, and significantly reduces mine safety risks and the probability of accidents.

[0035] It should be noted that the seismic wave signal identification module acquires real-time seismic wave propagation signals from the mine, identifies the seismic wave source location, magnitude, and propagation path characteristics, and transmits the identified characteristics to the regional difference analysis module. It should be noted that the microseismic multi-scale intelligent analysis module dynamically determines the frequency and amplitude of threshold updates by calculating the sample entropy change trend of the data sequence within the sliding window in real time. It uses the real-time sample entropy threshold to guide a consistent random sampling strategy. When the sample entropy exceeds the current dynamic threshold, it shortens the distance between sampling bands, increases the local sampling density, and adds an extra 20% to 30% of the number of sampling points in that area. When the sample entropy is lower than the current dynamic threshold, it extends the distance between sampling points and reduces the number of sampling points to 60% to 80% of the original design value, adaptively and dynamically adjusting the spatial distribution and density of sampling points.

[0036] The microseismic multi-scale intelligent analysis module dynamically determines the frequency and amplitude of threshold updates by calculating the sample entropy change trend of the data sequence within a sliding window in real time, and uses the real-time sample entropy threshold to guide a consistent random sampling strategy. The specific implementation process is as follows: For example, a continuous monitoring microseismic data sequence in a mining area has a length of 1000 data points, with an initial design sampling point density of once every 10 data points. Real-time sample entropy calculation reveals that when the sequence is monitored in the 200-300 range, the sample entropy rapidly increases from the initial threshold of 0.5 to 0.75, exceeding the dynamically updated threshold of 0.7. At this point, the sampling strategy is immediately adjusted, increasing the sampling density from once every 10 data points to once every 7 data points, equivalent to adding approximately 30% more sampling points in that range to more precisely capture the changing trends of microseismic events. When the sequence location was detected in the 700-800 range, the sample entropy decreased from 0.5 to 0.3, falling below the dynamic update threshold of 0.4. The system automatically reduced the sampling point density to 70% of the original design value, i.e., sampling once every approximately 14 data points. This saved computational resources while maintaining high analytical accuracy. This adaptive dynamic adjustment method of sampling point density enables precise dynamic monitoring and efficient data processing of microseismic data.

[0037] It should be noted that the regional difference analysis module acquires real-time data on the rock elastic modulus, Poisson's ratio, density, and medium damping parameters of different areas in the mine, as well as real-time seismic wave propagation data. It establishes geological characteristic models for each area, uses real-time seismic wave propagation data to solve for spatial partial derivatives of seismic wave data collected at continuous intervals, obtains the wavefield spatial gradient distribution, analyzes local wavefield errors in real-time based on the wavefield spatial gradient, sets dynamic adjustment thresholds, and evaluates local Courant stability conditions in real-time by analyzing local wave velocity differences and seismic waveform frequency variation trends. The upper limit of the time step is automatically calculated from the real-time seismic wave velocity data, and the optimal time step that meets the stability requirements is dynamically selected to accurately simulate the spatial differentiation characteristics of the seismic wave propagation path.

[0038] In a specific implementation example, during the exploration of a mine, the rock elastic modulus was measured in real time in regions A, B, and C as 15 GPa, 18 GPa, and 12 GPa, respectively; Poisson's ratios were 0.25, 0.28, and 0.22, respectively; densities were 2600 kg / m³, 2700 kg / m³, and 2500 kg / m³, respectively; and medium damping parameters were 0.02, 0.015, and 0.025, respectively. The system continuously acquired seismic wave data propagating in different regions and solved the spatial partial derivatives in real time. The calculated average spatial gradient of the seismic waves was 0.35 for region A, 0.45 for region B, and 0.25 for region C. Based on this, the system evaluated the local wavefield error in each region in real time and set a dynamic adjustment threshold. For example, when the wavefield gradient in region B exceeded the set threshold of 0.4, the system automatically increased the spatial difference order to improve the calculation accuracy. Meanwhile, based on the real-time seismic wave velocity differences in the regions, such as 3200 m / s in region A, 3500 m / s in region B, and 3000 m / s in region C, the system analyzes the frequency variation trend of seismic waveforms in real time and automatically calculates the Courant stability condition. It dynamically selects the optimal time step that meets the stability requirements, with the time step set at 0.0002 s for region A, 0.00018 s for region B, and 0.00022 s for region C. This enables high-precision real-time simulation of the spatial differentiation characteristics of seismic wave propagation paths, reducing the overall wavefield error to within 3%, and significantly improving the ability to identify concealed structures and predict mine safety risks.

[0039] It should be noted that the intelligent data processing module uses a sliding window method to monitor the rate of change of input data features in real time, and uses a real-time updated exponentially weighted moving average method to calculate the degree of anomaly of the features. Based on the current degree of anomaly, it adjusts the update frequency and weight distribution density of the spatial weight matrix in real time. Through a hierarchical attention mechanism, it realizes hierarchical interaction between local features and global features. The local attention layer quickly responds to the location of anomalies, and the global attention layer dynamically coordinates the attention intensity of each local feature. Through real-time feedback of latent high-risk area data, it triggers an adaptive attention adjustment algorithm to quickly focus on the most critical risk area.

[0040] The intelligent data processing module monitors the rate of change of input data features in real time using a sliding window method and calculates the degree of anomaly of the features using a real-time updated exponentially weighted moving average (EWMA) method. Based on the current degree of anomaly, it adjusts the update frequency and weight distribution density of the spatial weight matrix in real time. Specifically, in a mine concealed structure detection task, if the system detects that the EWMA of the mine data anomaly rises from 0.2 to 0.6 within time window T1 (0–10 minutes), indicating a drastic change in data features, the system automatically increases the update frequency of the spatial weight matrix to once every 5 seconds and increases the weight distribution density of the anomalous area by 50%, giving it higher attention. Within time window T2 (10–20 minutes), when the EWMA drops to 0.3, the system reduces the update frequency to once every 10 seconds and moderately increases the weight distribution density of the normal area by 15% to ensure overall monitoring coverage. The system simultaneously employs a hierarchical attention mechanism for interaction between local and global features. The local attention layer reduces the response time to abnormal data to 1.5 seconds, ensuring that high-risk areas receive priority computing resources. The global attention layer dynamically coordinates features from multiple areas of the mine, comprehensively analyzing the evolution trend of hidden structural risks. Ultimately, at time window T3 (20-30 minutes), the system, through an adaptive attention adjustment algorithm, identified the most critical high-risk area of ​​hidden structures and triggered an early warning mechanism, predicting the specific area where a mine tremor might occur 20 minutes in advance. This successfully averted a potential mine safety accident and improved the accuracy of hidden structural detection and the response speed of mine water hazard early warning.

[0041] In summary, this invention utilizes multi-scale data fusion and real-time dynamic analysis techniques, integrating differential GPS positioning, Kalman filtering, adaptive noise reduction, Kriging interpolation, joint inversion, adaptive nonlinear regularized inversion, GPU parallel computing, ICEEMDAN mode decomposition, consistent random sampling strategy, variational mode decomposition, and Autoformer deep learning model. It also incorporates short-time Fourier transform, adaptive wavelet analysis, high-order differential hybrid numerical methods, and dynamic weighted integrated learning and attention mechanisms for intelligent data processing. This enables precise detection of hidden structures and real-time dynamic prediction and early warning of mine water hazards, significantly reducing casualties and economic losses, and ensuring safe mine production and the construction of smart mines.

[0042] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0043] Finally: The above description is only an embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A concealed structure detection system for intelligent monitoring, early warning, prevention and treatment of mine water disasters, characterized in that, The system includes a multi-source data integrated detection module, a real-time imaging module, a microseismic multi-scale intelligent analysis module, a seismic wave signal identification module, a regional difference analysis module, and an intelligent data processing module. The multi-source data integrated detection module employs differential GPS positioning technology, integrates Kalman filtering and adaptive noise reduction algorithms, and utilizes Kriging interpolation and joint inversion methods to construct a three-dimensional geological structure model. The real-time imaging module constructs real-time subsurface structure images through an adaptive nonlinear regularized inversion method and uses GPU parallel computing to accelerate the imaging process. The microseismic multi-scale intelligent analysis module utilizes the ICEEMDAN modal decomposition algorithm and uniform random sampling... The system employs a rapid assessment method based on strategy and sample entropy, combined with variational mode decomposition and Autoformer model multi-step prediction of mine-related microseismic events. The regional difference analysis module utilizes short-time Fourier transform, adaptive wavelet analysis, and high-order difference hybrid numerical methods to simulate and analyze the regional characteristics of mine earthquake propagation. The intelligent data processing module integrates multi-source geophysical data, real-time dynamic imaging data, microseismic prediction data, and spatial difference data of seismic wave propagation. It employs a deep learning intelligent risk assessment model, dynamic weight ensemble learning method, and attention mechanism to provide real-time dynamic assessment and automatic early warning of hidden structures and mine water hazard risks.

2. The concealed structure detection system for intelligent monitoring, early warning, and prevention of mine water hazards according to claim 1, characterized in that, The multi-source data integrated detection module acquires GPS positioning data, geophysical exploration data, and electromagnetic interference data in real time. Based on adaptive spatial interpolation technology, it automatically detects local errors in the underground three-dimensional geological model of the mine in real time, dynamically updates the interpolation grid resolution, and quickly repairs missing data areas in the three-dimensional geological model.

3. The concealed structure detection system for intelligent monitoring, early warning, and prevention of mine water hazards according to claim 2, characterized in that, In the multi-source data integrated detection module, the local errors of the three-dimensional geological model include, but are not limited to, the first local error caused by the lack of data acquisition due to sensor failure, and the second local error caused by the lack of data due to local geological disturbances caused by encountering complex geological structures and mining activities. For the local error caused by the lack of data acquisition due to sensor failure, the first error is repaired by real-time monitoring of sensor status and automatic activation of redundant interpolation method for fusing adjacent sensor data within a set range. The second error is repaired by real-time detection of geological structure disturbance areas and the establishment of a targeted interpolation model using disturbance characteristics.

4. The concealed structure detection system for intelligent monitoring, early warning, and prevention of mine water hazards according to claim 3, characterized in that, In the multi-source data integrated detection module, during the repair of the first error, data from four to six sensors closest to the faulty sensor are selected for weighted fusion interpolation. During the repair of the second error, fault fracture feature matching interpolation model, goaf stress release dynamic interpolation model, and groundwater seepage feature interpolation model are established according to the spatial distribution characteristics of faults, fracture zones, goaf areas, and groundwater bodies, respectively.

5. The concealed structure detection system for intelligent monitoring, early warning, prevention and treatment of mine water disaster according to claim 4, characterized in that, During the process of repairing the second error using the multi-source data integrated detection module: The fault fracture feature matching interpolation model identifies the spatial distribution characteristics, fracture density and extension direction of fault fracture zones, and uses anisotropic interpolation algorithms to achieve fine-grained repair of spatial trends in areas with missing data. The dynamic interpolation model for stress release in goaf areas is based on real-time monitoring of stress changes in the surrounding rock strata. A dynamic interpolation function is established to reflect the changes in the geological disturbance area caused by stress release in real time, thereby achieving accurate data supplementation. The groundwater seepage characteristic interpolation model analyzes the seepage path, flow velocity, and dynamic monitoring data of groundwater level, and uses flow field-guided interpolation technology to supplement the missing data in the groundwater-affected area in real time.

6. The concealed structure detection system for intelligent monitoring, early warning, prevention and treatment of mine water disaster according to claim 1, characterized in that, The real-time imaging module is based on dynamic adaptive grid refinement technology. It performs differential processing on the resistivity data sequence acquired at continuous time intervals, calculates the resistivity change gradient between adjacent time intervals, identifies regions with abnormal resistivity trends, and judges the significance of resistivity anomalies in real time by setting gradient thresholds. It uses the adaptive differential gradient method to accurately identify the boundary position of the resistivity anomaly region, and dynamically adjusts the grid resolution based on the gradient value and spatial distribution range of the anomaly region. The grid size is automatically refined in regions where the gradient value is within the set first-level threshold range, and the grid size is automatically enlarged in regions where the gradient value is within the set second-level threshold range. The grid resolution and refinement region are dynamically adjusted and optimized in real time, enabling real-time and precise capture and identification of hidden structures.

7. The concealed structure detection system for intelligent monitoring, early warning, prevention and treatment of mine water disaster according to claim 1, characterized in that, The seismic wave signal identification module acquires real-time seismic wave propagation signals from the mine, identifies the source location, magnitude, and propagation path characteristics of the seismic waves, and transmits the identified characteristics to the regional difference analysis module.

8. The concealed structure detection system for intelligent monitoring, early warning, prevention and treatment of mine water disaster according to claim 1, characterized in that, The microseismic multi-scale intelligent analysis module dynamically determines the frequency and amplitude of threshold updates by calculating the sample entropy change trend of the data sequence within the sliding window in real time. It uses the real-time sample entropy threshold to guide a consistent random sampling strategy. When the sample entropy exceeds the current dynamic threshold, it shortens the distance between sampling bands, increases the local sampling density, and adds an extra 20% to 30% of the number of sampling points in that area. When the sample entropy is lower than the current dynamic threshold, it extends the distance between sampling points and reduces the number of sampling points to 60% to 80% of the original design value, adaptively and dynamically adjusting the spatial distribution and density of sampling points.

9. The concealed structure detection system for intelligent monitoring, early warning, prevention and treatment of mine water disaster according to claim 7, characterized in that, The regional difference analysis module acquires real-time data on the elastic modulus, Poisson's ratio, density, and damping parameters of rocks in different areas of the mine, as well as real-time seismic wave propagation data. It establishes geological characteristic models for each region, calculates spatial partial derivatives of the continuously acquired seismic wave data using real-time seismic wave propagation data, obtains the spatial gradient distribution of the wavefield, analyzes local wavefield errors in real-time based on the wavefield spatial gradient, sets dynamic adjustment thresholds, and evaluates local Courant stability conditions in real-time by analyzing local wave velocity differences and seismic waveform frequency variation trends. The module automatically calculates the upper limit of the time step based on real-time seismic wave velocity data, dynamically selects the optimal time step that meets stability requirements, and accurately simulates the spatial differentiation characteristics of seismic wave propagation paths.

10. The concealed structure detection system for intelligent monitoring, early warning, prevention and treatment of mine water disaster according to claim 1, characterized in that, The intelligent data processing module monitors the rate of change of input data features in real time using a sliding window method and calculates the degree of anomaly of features using a real-time updated exponentially weighted moving average method. Based on the current degree of anomaly, it adjusts the update frequency and weight distribution density of the spatial weight matrix in real time. Through a hierarchical attention mechanism, it realizes hierarchical interaction between local and global features. The local attention layer quickly responds to the location of anomalies, while the global attention layer dynamically coordinates the attention intensity of each local feature. By constructing high-risk area data through real-time feedback, it triggers an adaptive attention adjustment algorithm to quickly focus on the most critical risk area.

Images