Coal mine goaf monitoring method and system

Abstract

The application provides a kind of coal mine goaf monitoring method and system, belong to coal mine monitoring technical field.The method includes: through the multi-dimensional signal of surrounding rock of target goaf collected by the sensing network arranged in target goaf, the multi-dimensional signal of surrounding rock collected is inferred and fused, and the fusion signal is obtained;Based on the fusion signal, the boundary identification and stress distribution inversion of the current target goaf are executed one by one, and the boundary coordinates and stress distribution data of the current target goaf are obtained;According to the boundary coordinates and stress distribution data of the current target goaf, the multi-index comprehensive risk assessment of the current target goaf is carried out, based on the risk assessment result, the risk level of the current target goaf is matched, and the corresponding risk warning scheme is triggered according to the matched risk level;And, based on the risk assessment result, the sensing network arranged is adjusted.The purpose of goaf boundary accurate positioning, stress global monitoring and real-time risk warning is realized.

Description

Technical Field

[0001] This invention relates to the field of coal mine monitoring technology, specifically to a method for monitoring coal mine goaf areas, a coal mine goaf area monitoring system, a machine-readable storage medium, and an electronic device. Background Technology

[0002] Goaf areas are underground cavities formed after coal mining. Their boundaries are unstable, and stress distribution is complex, making them prone to safety accidents such as roof collapse, gas accumulation, and surface subsidence, seriously threatening safe coal mine production. The safety hazards posed by goaf areas include: first, low accuracy in goaf boundary positioning, making it impossible to accurately identify the extent of hidden goaf areas; second, lagging stress state monitoring, making it difficult to capture sudden stress changes in the surrounding rock of the goaf; and third, weak anti-interference capabilities of the monitoring system, resulting in severe data distortion in the complex underground environment and untimely early warnings.

[0003] Current monitoring of coal mine goaf areas mainly employs traditional physical detection and single-point sensing technologies, which have many limitations. For example, traditional physical detection (such as ground-penetrating radar and seismic wave methods) can only achieve phased detection and cannot track the dynamic changes of the goaf area in real time. Moreover, the detection accuracy is greatly affected by geological structure, resulting in large boundary positioning errors. Single-point sensing monitoring (such as pressure sensors and displacement sensors) can only acquire stress data at local points and cannot achieve full-area stress distribution monitoring in the goaf area. Furthermore, the sensors are susceptible to electromagnetic interference and dust corrosion, leading to a high failure rate. Existing intelligent monitoring technologies are usually based on a single sensing dimension, lacking the collaborative fusion of sound wave and vibration data. This makes it difficult to simultaneously achieve boundary positioning and stress monitoring, and the data processing latency is high, failing to meet the needs of real-time early warning.

[0004] Existing coal mine goaf monitoring technologies mainly fall into two categories: First, goaf boundary monitoring technology based on distributed optical fiber vibration, which identifies the goaf area by sensing surrounding rock vibration signals through optical fibers. However, it can only locate the boundary and cannot monitor stress state; moreover, the vibration signals are easily interfered with by mining operations, resulting in low identification accuracy. Second, stress monitoring technology based on fiber optic grating sensing can achieve local stress data acquisition, but the monitoring range is limited, it cannot simultaneously locate the goaf boundary, and the system deployment is complex, with poor adaptability. It cannot adapt to dynamic working conditions such as surrounding rock deformation and stress changes in the goaf, and is prone to monitoring blind spots. Both types of technologies suffer from problems such as single monitoring dimensions, low goaf boundary positioning accuracy, weak anti-interference ability, insufficient data collaboration, and poor dynamic adaptability, failing to meet the full-chain requirements of accurate goaf boundary positioning, comprehensive stress monitoring, and real-time risk early warning. Summary of the Invention

[0005] The purpose of this invention is to provide a method and system for monitoring coal mine goaf areas, so as to at least solve the problems mentioned above, such as the inability of existing technologies to meet the full-chain requirements of accurate goaf area boundary positioning, full-domain stress monitoring, and real-time risk early warning, due to the single monitoring dimension, low accuracy of goaf area boundary positioning, weak anti-interference ability, insufficient data collaboration, and poor dynamic adaptability.

[0006] To achieve the above objectives, a first aspect of the present invention provides a method for monitoring coal mine goaf areas, comprising: Multi-dimensional signals of the surrounding rock in the target goaf are collected by a sensor network deployed in the target goaf, and the collected multi-dimensional signals of the surrounding rock are inferred and fused to obtain a fused signal. Based on the fused signal, the boundary identification and stress distribution inversion of the current target goaf are performed one by one to obtain the boundary coordinates and stress distribution data of the current target goaf. Based on the boundary coordinates and stress distribution data of the current target goaf area, a comprehensive risk assessment using multiple indicators is conducted. Based on the risk assessment results, the risk level of the current target goaf area is matched, and a corresponding risk warning scheme is triggered according to the matched risk level; and... Based on the risk assessment results, the deployed sensor network was adjusted.

[0007] Optional, the initial deployment rules for sensor networks include: Based on the initial surrounding rock information of the target goaf, the potential boundary area and potential instability area of ​​the target goaf are determined; among them, the potential instability area includes at least the fault area and the area where the strength of the roof rock layer is lower than the preset strength threshold. Distributed optical fibers are deployed in the potential boundary area of ​​the target goaf, and optical fiber acoustic sensors are deployed in the potential instability area of ​​the target goaf. Based on the deployed distributed optical fibers and optical fiber acoustic sensors, a sensor network is formed. Among them, both the distributed optical fiber and the optical fiber acoustic sensor are fixed to the surrounding rock of the target goaf using explosion-proof anchor bolts.

[0008] Optionally, the rules for adjusting the deployed sensor network include: Based on the boundary coordinates of the target goaf evolving over time, the abnormal boundary regions of the target goaf are identified; based on the stress distribution data of the target goaf evolving over time, the abnormal stress gradient regions of the target goaf are identified; and based on the latest risk assessment results of the target goaf, the safety risk regions of the target goaf are determined. The identified boundary anomaly areas, stress gradient anomaly areas, and safety risk areas of the target goaf are designated as key monitoring areas. The coverage of the current sensor network's monitoring spatial distribution over the key monitoring areas is determined, and the monitoring defect areas of the current sensor network are identified based on the coverage determination results. Based on the current sensor network's monitoring defect areas, generate and execute sensor network adjustment strategies until the adjusted sensor network achieves the preset coverage standard for the key monitoring areas.

[0009] Optionally, inference and fusion are performed on the multi-dimensional signals of the collected surrounding rock, including: The acquired multi-dimensional signals of the surrounding rock were sequentially subjected to signal denoising, interference removal, and signal fusion; among them, The rules for signal noise reduction include: Variational mode decomposition is performed on the collected signals of the same dimension according to the preset number of candidates, so as to obtain the modal components of the corresponding dimension signal under each number of candidates. Determine the minimum interval between the energy distribution entropy of each modal component and the center frequency of adjacent modal components for a signal of the same dimension with the same number of candidates, and select the candidate number with the highest comprehensive fitness, which includes the minimum interval between the energy distribution entropy and the center frequency, as the optimal number of modes for the current dimension signal. Under the optimal number of modes, each modal component that satisfies the effective modal component condition in the corresponding dimension signal is selected, and the selected modal components are linearly superimposed to obtain the denoised corresponding dimension signal. The rules for interference removal include: Abnormal signals in the same dimension of the denoised signal are identified based on the 3σ criterion and the identified abnormal signals are removed. At the same time, the wavelet thresholding algorithm is used to remove interference signals in the same dimension of the signal. The rules for signal fusion include: Based on the feature vector of the same dimension signal after interference removal, a basic probability assignment function for the corresponding dimension signal is constructed, and the signal quality correction factor determined by the signal-to-noise ratio and sampling rate of the corresponding dimension signal is used to perform weighted correction on the constructed basic probability assignment function. Based on the conflict coefficients between the modified basic probability assignment functions of the signals in each dimension, the signals in each dimension after interference removal are fused according to the Dempster synthesis rule.

[0010] Optional rules for identifying the boundaries of the target goaf include: The fused signal is time-reversed to obtain the signal reception time of each sensor node in the corresponding sensor network. Based on the signal reception time of each sensor node, the propagation time difference of acoustic signal and vibration signal in the surrounding rock of the current target goaf is determined. The surrounding rock of the current target goaf is discretized into multiple candidate boundary points. For each candidate boundary point, the determined propagation time difference is used as the observation value. Within the corresponding preset position range, the optimal position of the current candidate boundary point is obtained by minimizing the error between the theoretical propagation time difference of the acoustic signal and vibration signal from the candidate boundary point to each sensor node and the observation value. The optimal position of each candidate boundary point is obtained by traversing each candidate boundary point. The optimal positions of each candidate boundary point are connected to form the boundary coordinates of the current target goaf.

[0011] Optionally, the sensor network includes distributed optical fibers for acquiring vibration signals from the surrounding rock; The rules for stress distribution inversion in the target goaf include: Phase demodulation and conversion from the phase of the vibration signal to the displacement caused by vibration at each sensing node of the distributed optical fiber are performed sequentially on the vibration signal in the fused signal. The time derivative of the converted displacement is then calculated to obtain the vibration acceleration of each sensing node of the corresponding distributed optical fiber. Based on the vibration acceleration of each sensing node of the distributed optical fiber, the elastic modulus and Poisson's ratio of the rock mass in the current target goaf, the stress value of each sensing node of the distributed optical fiber is determined, and the determined stress value is spatially interpolated to obtain the stress distribution data of the current target goaf.

[0012] Optionally, the multiple indicators may include at least the boundary stability index, which characterizes the intensity and development trend of the convergence deformation of the surrounding rock in the target goaf; the stress concentration index, which characterizes the stress accumulation degree and the degree of uneven stress distribution of the surrounding rock in the target goaf; and the surrounding rock deformation rate index, which characterizes the speed of deformation of the surrounding rock in the target goaf. A comprehensive risk assessment of the current target goaf area will be conducted based on multiple indicators, including: The current target goaf area is assessed by boundary stability index, stress concentration index, and surrounding rock deformation rate index. The risk assessment result of the current target goaf area is obtained by combining the results of the boundary stability index assessment, stress concentration index assessment, and surrounding rock deformation rate index assessment.

[0013] A second aspect of the present invention provides a coal mine goaf monitoring system, comprising: The signal fusion module is used to collect multi-dimensional signals of the surrounding rock in the target goaf through a sensor network deployed in the target goaf, and to perform inference and fusion on the collected multi-dimensional signals of the surrounding rock to obtain a fused signal. The boundary identification module is used to perform boundary identification and stress distribution inversion of the current target goaf area one by one based on the fused signal, so as to obtain the boundary coordinates and stress distribution data of the current target goaf area. The risk warning module is used to conduct a multi-indicator comprehensive risk assessment of the current target goaf area based on the boundary coordinates and stress distribution data of the current target goaf area, match the risk level of the current target goaf area based on the risk assessment results, and trigger the corresponding risk warning scheme according to the matched risk level; and adjust the deployed sensor network based on the risk assessment results.

[0014] In a third aspect, the present invention provides a machine-readable storage medium storing instructions that, when executed by a processor, configure the processor to perform the aforementioned coal mine goaf monitoring method.

[0015] In a fourth aspect, an electronic device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the aforementioned method for monitoring coal mine goaf areas.

[0016] The above technical solution provides a method and system for monitoring coal mine goaf areas. It collects multi-dimensional signals from the surrounding rock of the target goaf area through a sensor network deployed there, achieving simultaneous acquisition of multi-source signals. The collected multi-dimensional signals are then fused to obtain a fused signal, avoiding the problem of single-dimensional monitoring. Based on the fused signal, boundary identification and stress distribution inversion of the current target goaf area are performed sequentially to obtain its boundary coordinates and stress distribution data. This effectively identifies hidden goaf areas such as those near faults and below the roof, avoiding the problems of low boundary positioning accuracy and delayed stress monitoring. Based on the boundary coordinates and stress distribution data of the current target goaf area, a multi-indicator comprehensive risk assessment is conducted. Based on the risk assessment results, the risk level of the current target goaf area is matched, and a corresponding risk warning scheme is triggered according to the matched risk level. Furthermore, based on the risk assessment results, the deployed sensor network is adjusted, thus ensuring not only the timeliness of risk warnings but also real-time adaptation of the sensor network to dynamic conditions such as deformation and stress changes in the surrounding rock of the goaf area, eliminating monitoring blind spots and improving monitoring stability.

[0017] Other features and advantages of the embodiments of the present invention will be described in detail in the following detailed description section. Attached Figure Description

[0018] The accompanying drawings are provided to further illustrate embodiments of the present invention and form part of the specification. They are used together with the following detailed description to explain the embodiments of the present invention, but do not constitute a limitation thereof. In the drawings: Figure 1 This is a flowchart of a coal mine goaf monitoring method provided in one embodiment of the present invention; Figure 2This is a flowchart of a closed-loop monitoring system corresponding to a coal mine goaf monitoring method provided in one embodiment of the present invention; Figure 3 This is a block diagram of a coal mine goaf monitoring system provided in one embodiment of the present invention. Detailed Implementation

[0019] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0020] Figure 1 This is a flowchart of a coal mine goaf monitoring method provided by one embodiment of the present invention. Figure 1 As shown, an embodiment of the present invention provides a method for monitoring coal mine goaf areas, comprising: S110: Collect multi-dimensional signals of the surrounding rock in the target goaf through a sensor network deployed in the target goaf, and perform reasoning and fusion on the collected multi-dimensional signals of the surrounding rock to obtain a fused signal; The multidimensional signals of the surrounding rock include acoustic and vibration signals. The sensor network includes fiber optic acoustic sensors for acquiring acoustic signals from the surrounding rock and distributed optical fibers for acquiring vibration signals from the surrounding rock.

[0021] Specifically, by synchronously acquiring vibration signals from distributed optical fibers and acoustic signals from fiber optic acoustic sensors through a sensor network pre-deployed in the target goaf area, the synchronous acquisition of multi-source signals is achieved. Subsequently, the acquired acoustic and vibration signals are sequentially processed for noise reduction, interference removal, and fusion, ultimately yielding a high-quality fused signal that can be used for subsequent analysis.

[0022] Furthermore, the synchronous acquisition of acoustic and vibration signals from the surrounding rock also includes: applying timestamps to the acquired acoustic and vibration signals to ensure that the acquisition time deviation between the two does not exceed 10 milliseconds. The acquired acoustic and vibration signals are then transmitted to an edge computing terminal and / or a cloud server via a dual-mode transmission method (including fiber optic communication and 5G communication). Specifically, acoustic and vibration signals are transmitted via fiber optic communication, while early warning information and status data are transmitted via 5G communication; when the 5G network signal strength is detected to be below a threshold, the early warning information is switched to fiber optic communication transmission.

[0023] In some embodiments of this example, dual-mode transmission further includes encrypting the transmitted data using the AES-256 encryption algorithm.

[0024] In the above implementation process, this method addresses the problems of single monitoring dimension and unreliable data transmission by constructing a multi-source signal synchronous acquisition and dual-mode transmission system. Specifically, it includes: ① Signal acquisition: Fiber optic acoustic wave sensors acquire acoustic wave signals from the surrounding rock (frequency range 10Hz-10kHz), and a distributed vibration monitoring unit (in this application, the distributed vibration monitoring unit is composed of distributed optical fibers) acquires vibration signals (vibration acceleration range 0.01-100m / s²) based on phase-sensitive optical time-domain reflectometry (Φ-OTDR) technology. Simultaneous acquisition of multiple source signals and association with timestamps reduces acquisition delay. ≤10ms; ② Data transmission: Adopts dual-mode transmission of fiber optic and 5G. Fiber optic is responsible for the stable transmission of massive signals (transmission rate ≥1Gbps), and 5G is responsible for the rapid transmission of early warning information and status data. The transmission delay is ≤30ms, and the data transmission success rate is ≥98% in harsh environments; ③ Data encryption: The transmission process adopts the AES-256 encryption algorithm to ensure the security of monitoring data and prevent data leakage or tampering; ④ Low power consumption design: Adopts dual power supply mode of solar energy and battery, which is suitable for underground power supply scenarios. A single charge can be used continuously for ≥72 hours to meet the long-term monitoring needs.

[0025] In some embodiments of this example, inference fusion is performed on the multi-dimensional signals of the acquired surrounding rock, including: sequentially performing signal denoising, interference removal, and signal fusion on the multi-dimensional signals of the acquired surrounding rock; wherein, the rules for signal denoising include: performing variational mode decomposition on the acquired signals of the same dimension according to preset candidate numbers to obtain the modal components of the corresponding dimension signal under each candidate number; determining the minimum interval of the energy distribution entropy and the center frequency of adjacent modal components of the same dimension signal under the same candidate number, and selecting the candidate number with the highest comprehensive fitness containing the minimum interval of energy distribution entropy and center frequency as the optimal number of modes for the current dimension signal; under the optimal number of modes, filtering the modal components in the corresponding dimension signal that meet the conditions for effective modal components, and linearly superimposing the filtered modal components to obtain the denoised corresponding dimension signal; The preset number of candidates is 3 to 8.

[0026] Specifically, signals of the same dimension are decomposed into 3 to 8 modal components, respectively. For each modal decomposition of the same dimension signal with different candidate numbers, the energy distribution entropy and the minimum interval of the center frequencies of adjacent modal components are calculated for each candidate number. The candidate number with the highest overall fitness is selected as the optimal number of modal components for the current dimension signal. From the modal decomposition of the current dimension signal with the optimal number of modal components, modal components that satisfy the conditions for effective modal components are selected, and the selected modal components are linearly superimposed to obtain the denoised signal of the corresponding dimension.

[0027] The rules for interference removal include: identifying abnormal signals in the same dimension of the denoised signal based on the 3σ criterion and removing the identified abnormal signals; and using the wavelet thresholding algorithm to remove interference signals in the same dimension. Interference signals refer to external interference signals such as those from mining operations and equipment operation.

[0028] The rules for signal fusion include: constructing basic probability assignment functions for signals of the same dimension based on the feature vectors of the signals after interference removal, and using signal quality correction factors determined by the signal-to-noise ratio and sampling rate of the signals of the corresponding dimensions to perform weighted correction on the constructed basic probability assignment functions; and fusing the signals of each dimension after interference removal according to the Dempster synthesis rules based on the conflict coefficients between the corrected basic probability assignment functions of each dimension.

[0029] In the above implementation process, this method addresses the problems of high signal interference and low data quality in complex downhole environments by designing a three-stage signal processing and collaborative fusion process, specifically including: ① Signal denoising: An improved variational mode decomposition (VMD) algorithm is used for denoising (the improved VMD algorithm is the process of first determining the optimal number of modes, then performing mode decomposition under the optimal number of modes, and judging the effective mode components) to decompose 3 to 8 modes, removing electromagnetic and vibration interference, thus improving the signal-to-noise ratio by more than 60%; ② Interference removal: A 3σ criterion combined with a wavelet threshold algorithm is used to accurately remove external interference signals such as those from mining operations and equipment operation, with an interference removal accuracy of ≥95%; ③ Multi-source signal fusion: A fusion model is constructed based on improved DS evidence theory to fuse acoustic and vibration signals, outputting a highly complementary and reliable fused signal with a data fusion accuracy of ≥92%. Through signal processing and fusion, data quality and anti-interference capabilities are significantly improved, providing support for subsequent analysis. Through the above signal processing and fusion process, the obtained fused signal has stronger anti-interference ability and higher reliability. The data accuracy rate in complex downhole environments is ≥92%, the anti-electromagnetic and vibration interference ability is significantly improved, the equipment failure rate is reduced by 60%, and the monitoring stability is improved by more than 30%.

[0030] S120: Based on the fused signal, perform boundary identification and stress distribution inversion of the current target goaf area one by one to obtain the boundary coordinates and stress distribution data of the current target goaf area; Specifically, this method addresses the issues of low boundary positioning accuracy and lag in stress monitoring by constructing a collaborative analysis process for boundary identification and stress distribution inversion. This process includes: ① Goaf boundary positioning: Based on fused signals, time reversal positioning is used, combined with the propagation time difference between acoustic and vibration signals, to accurately calculate the goaf boundary coordinates, ensuring a positioning error ≤0.5m, a concealed goaf identification accuracy ≥95%, and a positioning delay ≤80ms; ② Stress state inversion: Based on distributed vibration signals, stress inversion is performed. Through the correlation between vibration acceleration and stress, the stress distribution across the entire goaf is inverted, achieving a stress monitoring accuracy of ±0.1MPa and a stress mutation response delay ≤100ms; ③ Dynamic updating: Boundary positioning and stress monitoring data are updated at a frequency of 10Hz to track dynamic changes in the goaf in real time, ensuring the timeliness and accuracy of the monitoring data. This method not only boasts higher positioning accuracy and stronger identification capabilities (boundary positioning error ≤0.5m, more than 75% improvement over existing technologies, accuracy rate for identifying concealed goaf areas ≥95%, effectively identifying concealed goaf areas near faults and below the roof, with a positioning delay ≤80ms), but also provides more real-time stress monitoring and faster response (stress monitoring accuracy ±0.1MPa, stress mutation response delay ≤100ms, more than 66.7% reduction over existing technologies, enabling real-time inversion of stress distribution across the entire domain, and allowing for early detection of stress concentration and mutation risks).

[0031] In some embodiments of this example, the boundary identification rules for the target goaf include: performing time reversal processing on the fused signal to obtain the signal reception time of each sensor node in the corresponding sensor network; determining the propagation time difference of the acoustic signal and vibration signal in the fused signal at the surrounding rock of the current target goaf based on the obtained signal reception time of each sensor node; discretizing the surrounding rock of the current target goaf into multiple candidate boundary points; for each candidate boundary point, using the determined propagation time difference as the observation value; within the corresponding preset position range, aiming to minimize the error between the theoretical propagation time difference of the acoustic signal and vibration signal in the fused signal from the candidate boundary point to each sensor node and the observation value, obtaining the optimal position of the current candidate boundary point; traversing each candidate boundary point to obtain the optimal position of each candidate boundary point; and connecting the optimal positions of each candidate boundary point to form the boundary coordinates of the current target goaf.

[0032] Specifically, by fitting a continuous and smooth closed curve to the optimal position of all candidate boundary points obtained through traversal, the complete boundary contour of the target goaf can be obtained, thus achieving accurate positioning of the goaf boundary and effectively identifying hidden goaf areas that are difficult to detect.

[0033] In some embodiments of this example, the sensor network includes distributed optical fibers for acquiring vibration signals of the surrounding rock; the rules for inverting the stress distribution of the target goaf include: sequentially performing phase demodulation on the vibration signals in the fused signal and converting the phase of the vibration signal to the displacement caused by vibration at each sensor node of the distributed optical fiber, and performing time derivative on the converted displacement to obtain the vibration acceleration of each sensor node of the corresponding distributed optical fiber; based on the vibration acceleration of each sensor node of the distributed optical fiber, the elastic modulus and Poisson's ratio of the rock mass of the current target goaf, determining the stress value of each sensor node of the distributed optical fiber, and performing spatial interpolation on the determined stress value to obtain the stress distribution data of the current target goaf.

[0034] Specifically, the vibration signal in the fused signal is first demodulated to obtain the corresponding phase change sequence. This phase change sequence is then converted into a dynamic strain sequence along the optical fiber, with the conversion relationship: ε(t) = φ(t)·λ / (4π·n·L·K_ε), where φ(t) is the phase change sequence, K_ε is the optical fiber strain sensitivity coefficient, ε(t) is the dynamic strain sequence along the optical fiber, λ is the probe wavelength, n is the refractive index of the optical fiber core, and L is the sensing gauge length. The dynamic strain sequence is then integrated within the spatial sensing gauge to obtain the displacement sequence d(t) of each sensing node, with the formula: d(t) = ∫ε(t)· dl, where dl is the infinitesimal element along the integration path of the optical fiber. The time derivative of the displacement sequence of each sensing node is then calculated to obtain the vibration acceleration of each sensing node in the corresponding distributed optical fiber. Based on the elastic modulus E, Poisson's ratio μ, and vibration acceleration a obtained from the distributed vibration monitoring, the stress value σ of each sensing node is calculated using the formula σ=E·(a / g)·μ, where g is the gravitational acceleration. Spatial interpolation is then performed on the stress value of each sensing node to obtain the stress distribution throughout the target goaf.

[0035] Furthermore, stress inversion also includes: collecting the temperature of the target goaf and correcting the elastic modulus E based on the pre-stored temperature-elastic modulus relationship curve.

[0036] S130: Based on the boundary coordinates and stress distribution data of the current target goaf, conduct a multi-indicator comprehensive risk assessment of the current target goaf, match the risk level of the current target goaf based on the risk assessment results, and trigger the corresponding risk warning scheme according to the matched risk level; and adjust the deployed sensor network based on the risk assessment results.

[0037] Specifically, this method collects multi-dimensional signals of the surrounding rock in the target goaf through a sensor network deployed in the goaf, achieving synchronous acquisition of multi-source signals. The collected multi-dimensional signals are then fused to obtain a fused signal, avoiding the problem of single-dimensional monitoring. Based on the fused signal, boundary identification and stress distribution inversion of the current target goaf are performed sequentially to obtain the boundary coordinates and stress distribution data, effectively identifying hidden goaf areas such as those near faults and below the roof, avoiding the problems of low boundary positioning accuracy and lag in stress monitoring. Based on the boundary coordinates and stress distribution data of the current target goaf, a multi-indicator comprehensive risk assessment is conducted. Based on the risk assessment results, the risk level of the current target goaf is matched, and a corresponding risk warning scheme is triggered according to the matched risk level. Furthermore, based on the risk assessment results, the deployed sensor network is adjusted, thus ensuring not only the timeliness of risk warnings but also real-time adaptation of the sensor network to dynamic conditions such as deformation and stress changes in the surrounding rock of the goaf, eliminating monitoring blind spots and improving monitoring stability.

[0038] In the above implementation process, this method integrates goaf boundary positioning and stress state monitoring by constructing a sensor network for collaborative monitoring of fiber optic acoustic waves and distributed fiber optic vibration, simultaneously outputting goaf range and overall stress distribution data, thus solving the problem of single monitoring dimension. By executing a multi-dimensional signal fusion boundary identification process, the accuracy of goaf boundary positioning is improved, with a positioning error ≤0.5m and a concealed goaf identification accuracy ≥95%, solving the problem of low boundary positioning accuracy. By optimizing the distributed vibration stress inversion process, real-time monitoring of goaf stress state is achieved, with a stress change response delay ≤100ms and overall stress monitoring coverage ≥98%, solving the problem of stress monitoring lag. By improving the signal processing and fusion process, the data accuracy in complex underground environments is improved to over 92%, and the resistance to electromagnetic and vibration interference is significantly enhanced, solving the problem of weak anti-interference capability. By constructing an adaptive monitoring adjustment mechanism based on risk assessment results to adjust the sensor network, it adapts in real-time to dynamic working conditions such as goaf surrounding rock deformation and stress changes, eliminating monitoring blind spots and improving monitoring stability by over 30%, solving the problem of poor dynamic adaptability. Furthermore, this method only requires deploying a sensor network in the target goaf area, reducing the deployment cost per monitoring area by more than 40%, reducing equipment maintenance and labor costs, adapting to large-scale coal mine applications, and solving the problem of high monitoring costs.

[0039] Furthermore, this method addresses the issues of delayed risk warnings and inaccurate assessments by constructing a multi-dimensional assessment and tiered early warning system, specifically including: ① Risk assessment indicators: Establishing three major assessment indicators—boundary stability, stress concentration, and surrounding rock deformation rate—with the core assessment formula R=0.4R_b+0.35R_sigma+0.25R_d (where R is the risk assessment result integrating the three major indicators, R_b is the boundary stability coefficient, R_sigma is the stress concentration coefficient, and R_d is the surrounding rock deformation rate coefficient), thereby quantifying the safety risk level of the goaf through the value of the risk assessment result R; ② Tiered early warning mechanism: Based on the risk level (risk... The risk levels are categorized into low, medium, high, and extremely high, triggering four levels of early warning. Specifically, low risk (R < 0.3) corresponds to generating a monitoring report; medium risk (0.3 ≤ R < 0.6) corresponds to triggering an audible and visual alarm; high risk (0.6 ≤ R < 0.8) corresponds to pushing disposal suggestions and locking work permissions; and extremely high risk (R ≥ 0.8) corresponds to triggering an emergency evacuation warning. The early warning response delay is ≤ 200ms. ③ Early Warning Feedback: Early warning information is simultaneously pushed to the monitoring center, on-site terminals, and mobile terminals of management personnel, supporting early warning information traceability and disposal process recording, forming a closed-loop management system of "early warning-disposal-feedback." Through multi-dimensional risk assessment and graded early warning, early warnings are more accurate and safer, with an early warning response delay of ≤ 200ms. Risks such as roof collapse and stress exceeding limits can be predicted in advance, reducing the risk of safety accidents by more than 70%, providing strong protection for safe coal mine production.

[0040] In some embodiments of this example, the multiple indicators include at least a boundary stability indicator to characterize the severity and development trend of the convergence deformation of the surrounding rock in the target goaf, a stress concentration indicator to characterize the degree of stress accumulation and the degree of uneven stress distribution in the surrounding rock of the target goaf, and a surrounding rock deformation rate indicator to characterize the speed of deformation of the surrounding rock in the target goaf. A comprehensive risk assessment of the current target goaf using multiple indicators is performed, including: assessing the boundary stability indicator, stress concentration indicator, and surrounding rock deformation rate indicator for the current target goaf separately, and combining the results of the boundary stability indicator assessment, stress concentration indicator assessment, and surrounding rock deformation rate indicator assessment to obtain the risk assessment result for the current target goaf.

[0041] Specifically, the rules for assessing boundary stability include: calculating the rate of change of curvature and the total displacement of the current target goaf boundary contour; normalizing the rate of change of curvature and the total displacement and then weighted summing them to obtain the boundary stability coefficient R_b (i.e., the boundary stability assessment result); where the larger the rate of change of curvature and the larger the total displacement, the larger the boundary stability coefficient R_b, and the worse the boundary stability and the higher the risk. The rules for assessing stress concentration index include: based on the inverted global stress distribution data, statistically analyzing the proportion of areas in the current target goaf where stress exceeds the ultimate strength of the surrounding rock, and the ratio of maximum stress to average stress; normalizing these two parameters and then weighted summing them to obtain the stress concentration coefficient R_sigma (i.e., the stress concentration assessment result); where the larger the proportion of areas with stress exceeding the limit and the larger the ratio of maximum stress to average stress, the larger the stress concentration coefficient, and the higher the stress accumulation and the higher the risk. The rules for evaluating the surrounding rock deformation rate index include: based on the displacement data obtained from multiple rounds of continuous monitoring, calculating the average deformation rate and the rate of change of the deformation rate of the surrounding rock per unit time, normalizing the two parameters respectively, and then weighted summing them to obtain the surrounding rock deformation rate coefficient R_d (i.e., the evaluation result of the surrounding rock deformation rate index); among them, the larger the average deformation rate and the larger the rate of change of the deformation rate, the larger the surrounding rock deformation rate coefficient R_d, the faster the surrounding rock deformation develops, and the higher the risk.

[0042] In some embodiments of this example, the initial deployment rules of the sensor network include: determining the potential boundary region and potential instability region of the target goaf based on the initial surrounding rock information of the target goaf; wherein, the potential instability region includes at least the fault region and the region where the strength of the roof strata is lower than a preset strength threshold; deploying distributed optical fibers in the potential boundary region of the target goaf, and deploying fiber optic acoustic sensors in the potential instability region of the target goaf; and forming a sensor network based on the deployed distributed optical fibers and fiber optic acoustic sensors; wherein, both the distributed optical fibers and the fiber optic acoustic sensors are fixed to the surrounding rock of the target goaf using explosion-proof anchor bolts.

[0043] In some embodiments of this example, the rules for adjusting the deployed sensor network include: identifying boundary anomaly regions of the target goaf based on the boundary coordinates of the target goaf evolving over time; identifying stress gradient anomaly regions of the target goaf based on stress distribution data of the target goaf evolving over time; determining safety risk regions of the target goaf based on the latest risk assessment results of the target goaf; using the identified boundary anomaly regions, stress gradient anomaly regions, and safety risk regions of the target goaf as key monitoring areas; determining the coverage range of the current sensor network's monitoring spatial distribution over the key monitoring areas; determining the monitoring defect areas of the current sensor network based on the coverage range determination results; and generating and executing a sensor network adjustment strategy based on the monitoring defect areas of the current sensor network until the adjusted sensor network's coverage range over the key monitoring areas reaches a preset coverage standard.

[0044] Specifically, this method addresses the issues of poor dynamic adaptability and monitoring blind spots in sensor networks deployed in goaf areas by designing an adaptive fiber optic sensor network deployment scheme, which includes: ① Deployment method: A combination of distributed fiber optics and fixed-point acoustic sensors is used. Distributed fiber optics (monitoring distance 0-500m, spatial resolution 1m) are deployed in potential boundary areas of goaf areas, and fiber optic acoustic sensor nodes (sampling frequency 1000Hz) are added in key areas such as faults and weak roofs to form a sensor network with full coverage; ② Fixing method: Explosion-proof anchors are used to fix the fiber optics, combined with high-temperature resistant and tensile-resistant sheaths, to ensure that the fiber optics can still work stably when the surrounding rock deforms, adapting to dynamic working conditions where the surrounding rock displacement is ≤50cm; ③ Adaptive adjustment: Based on previous monitoring data and geological information, the fiber optic deployment position is adjusted by moving sensor nodes to eliminate monitoring blind spots, achieving a monitoring coverage rate of ≥98%; ④ Network self-test: A built-in fiber optic link self-test unit monitors the fiber optic connection status in real time, with a fault identification delay of ≤10ms, ensuring stable operation of the sensor network. This method deploys a sensor network that integrates fiber optic acoustic waves and distributed vibration monitoring, simultaneously achieving goaf boundary positioning and stress state monitoring with a monitoring coverage rate of ≥98%. It solves the problem of single monitoring dimensions in traditional technologies, eliminating the need for multiple device combinations and providing more comprehensive and integrated monitoring. Furthermore, it can adapt to dynamic conditions such as surrounding rock deformation and stress changes in the goaf in real time, optimizing the sensor network through data feedback, eliminating monitoring blind spots, and adapting to different types of goaf scenarios, resulting in better dynamic adaptability and no monitoring blind spots.

[0045] Please refer to Figure 2 , Figure 2This is a flowchart of a closed-loop monitoring system corresponding to a coal mine goaf monitoring method provided in one embodiment of the present invention. The closed-loop monitoring system corresponding to this coal mine goaf monitoring method adopts a hierarchical closed-loop architecture design, with each module operating independently and sharing data. It supports dual-mode operation at the edge and cloud, can cope with the instability of underground networks, and achieves real-time monitoring of different types of goaf areas in all scenarios. The method uses "fiber optic acoustic sensing + distributed vibration monitoring + multi-source signal fusion + adaptive early warning" as its core design concept to construct a five-level closed-loop monitoring system, adapting to the harsh underground environment and dynamic working conditions of goaf areas. The specific process is as follows: First, a distributed fiber optic sensor network is deployed around the goaf. Fiber optic acoustic sensors and distributed vibration monitoring units simultaneously collect acoustic and vibration signals from the surrounding rock, which are then transmitted to the edge terminal via dual-mode transmission. Second, the collected signals are preprocessed and fused to eliminate electromagnetic and vibration interference, outputting high-quality, effective signals. Then, through boundary identification and stress inversion processes, the goaf boundary is accurately located, and the stress distribution across the entire area is inverted. Subsequently, the risk level is quantified based on the monitoring data, triggering tiered early warnings. Finally, data feedback optimizes the sensor network deployment, forming a closed-loop monitoring system to ensure monitoring accuracy and stability. This method achieves real-time monitoring, risk warning, and dynamic optimization of the goaf boundary and stress state across all scenarios through a five-level closed-loop process. The specific steps are as follows: 1. Sensor network deployment: Adaptively adjust sensor nodes and optimize monitoring coverage to build a sensor network with full coverage. Specifically, based on geological exploration data and potential extent of the goaf, an adaptive fiber optic sensor network is deployed, including: ① Distributed fiber optic deployment: Distributed fiber optic cables are deployed along the surrounding roadways, roof, and floor of the goaf, fixed with explosion-proof anchor bolts, with a fiber spacing of 1-2m, covering the entire potential goaf area (0-500m); ② Fixed-point sensor deployment: Fiber optic acoustic wave sensor nodes are added in key areas such as faults and weak roofs, with a sampling frequency set to 1000Hz to ensure accurate signal acquisition in key areas; ③ Network self-test: The fiber optic link self-test unit is activated to detect the fiber optic continuity and sensor node operating status, with a self-test pass rate ≥99%, and faulty nodes are marked and their locations are fed back in real time to ensure stable operation of the sensor network; ④ Parameter calibration: Parameter calibration is performed on the sensors and transmission links to ensure the accuracy of acoustic wave and vibration signal acquisition and transmission reliability, with a calibration error ≤0.5%.

[0046] 2. Multi-source signal acquisition: Simultaneous acquisition of sound wave and vibration signals, dual-mode transmission; Specifically, the sensor network initiates a synchronous acquisition process, acquiring and transmitting multi-source signals, including: ① Signal acquisition: Fiber optic acoustic sensors acquire acoustic signals from the surrounding rock (10Hz-10kHz), and distributed vibration monitoring units acquire vibration signals (0.01-100m / s²) based on Φ-OTDR technology. Signals are acquired synchronously and timestamps are added, with an acquisition cycle of 10ms / time to ensure signal time synchronization; ② Data transmission: A dual-mode transmission method using fiber optics and 5G is adopted. Fiber optics transmits massive amounts of raw signals (rate ≥1Gbps), while 5G transmits simplified status data and early warning information. The transmission process uses AES-256 encryption, with a transmission delay ≤30ms and a data transmission success rate ≥98% in harsh environments; ③ Data caching: Edge terminals locally cache the acquired signals (cache capacity ≥100GB) to prevent data loss due to network interruption. Data is automatically retransmitted after network recovery.

[0047] 3. Signal preprocessing and fusion: noise reduction, interference removal, and collaborative fusion to output an effective signal; Specifically, the edge terminal initiates a signal processing and fusion process, including: ① Signal denoising: An improved VMD algorithm is used to denoise the acoustic and vibration signals, decomposing them into 3 to 8 modes to remove electromagnetic and vibration interference. The signal-to-noise ratio is improved by more than 60% after denoising; ② Interference removal: External interference signals such as mining operations and equipment operation are removed using the 3σ criterion combined with a wavelet threshold algorithm, achieving an interference removal accuracy of ≥95%; ③ Signal standardization: Z-Score standardization is used to unify the signal scale, making the data distribution tend to N(0,1), improving the efficiency of subsequent algorithm processing; ④ Multi-source fusion: Acoustic and vibration signals are fused based on an improved DS evidence theory to eliminate data conflicts and output a high-quality fused signal, with a data fusion accuracy of ≥92%, providing reliable data support for subsequent analysis.

[0048] 4. Boundary location and stress inversion: Accurately locate the boundary and invert the stress distribution across the entire domain; Specifically, based on fused signals, the goaf boundary location and stress state inversion are carried out, including: ① Boundary location: An improved time-reversal location algorithm is used, combined with the propagation time difference of sound waves and vibration signals, to calculate the goaf boundary coordinates. The location error is ≤0.5m, the accuracy of identifying concealed goaf areas is ≥95%, and the location delay is ≤80ms; ② Stress inversion: Through the correlation model between vibration acceleration and stress, the stress distribution of the entire goaf area is inverted. The stress monitoring accuracy is ±0.1MPa, and the stress change response delay is ≤100ms; ③ Data correlation: The boundary location data and stress distribution data are correlated and analyzed to mark stress concentration areas and unstable boundary areas, and a goaf state analysis report is generated; ④ Dynamic update: The boundary location and stress monitoring data are updated at a frequency of 10Hz to track the dynamic changes of the goaf in real time and ensure the timeliness of the monitoring data.

[0049] 5. Status Assessment and Early Warning: Quantify risk levels and trigger tiered early warnings; Specifically, risk assessment and graded early warning are carried out based on monitoring and analysis results, including: ① Risk quantification: Calculating three major indicators—boundary stability, stress concentration, and surrounding rock deformation rate—and quantifying the risk level (including low, medium, high, and extremely high) based on the calculation results of the three major indicators; ② Graded early warning: Triggering corresponding early warning mechanisms according to the risk level, with an early warning response delay of ≤200ms, and pushing early warning information to the monitoring center, field terminals, and mobile terminals simultaneously; ③ Emergency response: For high-risk and extremely high-risk situations, pushing specific response suggestions (such as strengthening support, evacuating personnel, and stopping operations), and recording the response process and results; ④ Data feedback: Feeding back monitoring data, assessment results, and early warning response records to the sensor network deployment module, optimizing the fiber optic deployment location and sensor parameters, eliminating monitoring blind spots, improving monitoring accuracy, and forming a complete closed loop of "deployment-collection-processing-analysis-early warning-optimization".

[0050] The above implementation process employs a five-level closed-loop step: sensor network deployment and self-testing, multi-source signal acquisition and transmission, signal preprocessing and fusion, boundary positioning and stress inversion, and state assessment and early warning. This achieves precise positioning of goaf boundaries, real-time monitoring of stress state, and risk-level early warning. This method is applicable to complex scenarios such as goaf areas behind fully mechanized coal mining faces, old goaf areas, and goaf areas near faults. It enables precise positioning of goaf boundaries, dynamic monitoring of stress state, and risk early warning, and is suitable for harsh working environments with high dust levels, strong electromagnetic interference, and limited space underground. Specifically, the following effects are achieved: (1) Multi-source sensing collaborative monitoring: Integrating fiber optic acoustic sensing and distributed vibration monitoring technologies, it realizes the integration of goaf boundary positioning and stress state monitoring, with a monitoring coverage rate of ≥98%, solving the problem of single monitoring dimension in traditional technologies; (2) High-precision boundary positioning: Based on the improved time reversal positioning algorithm and multi-source signal fusion, the positioning error of the goaf boundary is ≤0.5m, the accuracy of identifying hidden goaf is ≥95%, and the positioning delay is ≤80ms, which significantly improves the positioning accuracy; (3) Real-time stress state inversion: Construct a vibration-stress correlation inversion model, with stress monitoring accuracy of ±0.1MPa and stress change response delay ≤100ms, to realize real-time monitoring of stress distribution in the entire goaf area and solve the problem of stress monitoring lag. (4) Strong anti-interference signal processing: The improved VMD noise reduction and interference removal algorithm is adopted, combined with multi-source signal fusion technology. The data accuracy rate is ≥92% in complex downhole environments, and the anti-electromagnetic and vibration interference capabilities are significantly improved. (5) Adaptive closed-loop monitoring: Construct a closed-loop system of "monitoring-analysis-early warning-optimization", which can adapt to the dynamic working conditions of the goaf, adjust the sensor network and monitoring parameters in real time, eliminate monitoring blind spots, and improve monitoring stability; (6) Low cost and high efficiency operation and maintenance: The system deployment architecture is simplified, the deployment cost of a single monitoring area is reduced by more than 40%, the equipment failure rate is reduced by 60%, and no dedicated personnel are required to be on-site, which significantly reduces operation and maintenance and labor costs.

[0051] Figure 3 This is a block diagram of a coal mine goaf monitoring system provided in one embodiment of the present invention. Figure 3 As shown, an embodiment of the present invention provides a coal mine goaf monitoring system, comprising: The signal fusion module is used to collect multi-dimensional signals of the surrounding rock in the target goaf through a sensor network deployed in the target goaf, and to perform inference and fusion on the collected multi-dimensional signals of the surrounding rock to obtain a fused signal. The boundary identification module is used to perform boundary identification and stress distribution inversion of the current target goaf area one by one based on the fused signal, so as to obtain the boundary coordinates and stress distribution data of the current target goaf area. The risk warning module is used to conduct a multi-indicator comprehensive risk assessment of the current target goaf area based on the boundary coordinates and stress distribution data of the current target goaf area, match the risk level of the current target goaf area based on the risk assessment results, and trigger the corresponding risk warning scheme according to the matched risk level; and adjust the deployed sensor network based on the risk assessment results.

[0052] Specifically, the system collects multi-dimensional signals of the surrounding rock in the target goaf through a sensor network deployed in the goaf, achieving synchronous acquisition of multi-source signals. The collected multi-dimensional signals are then fused to obtain a fused signal, avoiding the problem of single-dimensional monitoring. Based on the fused signal, boundary identification and stress distribution inversion of the current target goaf are performed sequentially to obtain the boundary coordinates and stress distribution data, effectively identifying hidden goaf areas such as those near faults and below the roof, avoiding the problems of low boundary positioning accuracy and delayed stress monitoring. Based on the boundary coordinates and stress distribution data of the current target goaf, a comprehensive risk assessment is conducted. Based on the risk assessment results, the risk level of the current target goaf is matched, and a corresponding risk warning scheme is triggered according to the matched risk level. Furthermore, based on the risk assessment results, the deployed sensor network is adjusted, thus ensuring not only the timeliness of risk warnings but also real-time adaptation of the sensor network to dynamic conditions such as deformation and stress changes in the surrounding rock of the goaf, eliminating monitoring blind spots and improving monitoring stability.

[0053] The optional embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the embodiments of the present invention are not limited to the specific details in the above embodiments. Within the scope of the technical concept of the embodiments of the present invention, various simple modifications can be made to the technical solutions of the embodiments of the present invention, and these simple modifications all fall within the protection scope of the embodiments of the present invention.

[0054] It should also be noted that the various specific technical features described in the above embodiments can be combined in any suitable manner without contradiction. To avoid unnecessary repetition, the embodiments of the present invention will not describe the various possible combinations separately.

[0055] Those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. This program is stored in a storage medium and includes several instructions to cause a microcontroller, chip, or processor to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0056] Furthermore, various different implementations of the present invention can be combined arbitrarily, as long as they do not violate the spirit of the present invention, they should also be regarded as the content disclosed in the present invention.

Claims

1. A method for monitoring coal mine goaf areas, characterized in that, include: Multi-dimensional signals of the surrounding rock in the target goaf are collected by a sensor network deployed in the target goaf, and the collected multi-dimensional signals of the surrounding rock are inferred and fused to obtain a fused signal. Based on the fused signal, the boundary identification and stress distribution inversion of the current target goaf are performed one by one to obtain the boundary coordinates and stress distribution data of the current target goaf. Based on the boundary coordinates and stress distribution data of the current target goaf, a multi-indicator comprehensive risk assessment is conducted on the current target goaf. Based on the risk assessment results, the risk level of the current target goaf is matched, and the corresponding risk warning scheme is triggered according to the matched risk level. as well as, Based on the risk assessment results, the deployed sensor network will be adjusted.

2. The method for monitoring coal mine goaf areas according to claim 1, characterized in that, The initial deployment rules for the sensor network include: Based on the initial surrounding rock information of the target goaf, the potential boundary region and potential instability region of the target goaf are determined; wherein, the potential instability region includes at least the fault region and the region where the strength of the roof rock layer is lower than a preset strength threshold; Distributed optical fibers are deployed in the potential boundary area of ​​the target goaf, and optical fiber acoustic sensors are deployed in the potential instability area of ​​the target goaf. Based on the deployed distributed optical fibers and optical fiber acoustic sensors, a sensor network is formed. Both the distributed optical fiber and the optical fiber acoustic sensor are fixed to the surrounding rock of the target goaf using explosion-proof anchor bolts.

3. The method for monitoring coal mine goaf areas according to claim 1, characterized in that, The rules for adjusting the deployed sensor network include: Based on the boundary coordinates of the target goaf evolving over time, the abnormal boundary regions of the target goaf are identified; based on the stress distribution data of the target goaf evolving over time, the abnormal stress gradient regions of the target goaf are identified; and based on the latest risk assessment results of the target goaf, the safety risk regions of the target goaf are determined. The identified boundary anomaly areas, stress gradient anomaly areas, and safety risk areas of the target goaf are used as key monitoring areas. The coverage of the current sensor network's monitoring spatial distribution over the key monitoring areas is determined, and the monitoring defect areas of the current sensor network are determined based on the coverage determination results. Based on the current sensor network's monitoring defect areas, a sensor network adjustment strategy is generated and executed until the adjusted sensor network's coverage of the key monitoring areas reaches a preset coverage standard.

4. The method for monitoring coal mine goaf areas according to claim 1, characterized in that, The inference and fusion of multi-dimensional signals from the collected surrounding rock includes: The acquired multi-dimensional signals of the surrounding rock were sequentially subjected to signal denoising, interference removal, and signal fusion; among them, The rules for signal noise reduction include: Variational mode decomposition is performed on the collected signals of the same dimension according to the preset number of candidates, so as to obtain the modal components of the corresponding dimension signal under each number of candidates. Determine the minimum interval between the energy distribution entropy of each modal component and the center frequency of adjacent modal components for a signal of the same dimension with the same number of candidates, and select the candidate number with the highest comprehensive fitness, which includes the minimum interval between the energy distribution entropy and the center frequency, as the optimal number of modes for the current dimension signal. Under the optimal number of modes, each modal component that satisfies the effective modal component condition in the corresponding dimension signal is selected, and the selected modal components are linearly superimposed to obtain the denoised corresponding dimension signal. The rules for interference removal include: Abnormal signals in the same dimension of the denoised signal are identified based on the 3σ criterion and the identified abnormal signals are removed. At the same time, the wavelet thresholding algorithm is used to remove interference signals in the same dimension of the signal. The rules for signal fusion include: Based on the feature vector of the same dimension signal after interference removal, a basic probability assignment function for the corresponding dimension signal is constructed, and the signal quality correction factor determined by the signal-to-noise ratio and sampling rate of the corresponding dimension signal is used to perform weighted correction on the constructed basic probability assignment function. Based on the conflict coefficients between the modified basic probability assignment functions of the signals in each dimension, the signals in each dimension after interference removal are fused according to the Dempster synthesis rule.

5. The method for monitoring coal mine goaf areas according to claim 1, characterized in that, The rules for identifying the boundary of the target goaf include: The fused signal is time-reversed to obtain the signal reception time of each sensor node in the corresponding sensor network. Based on the signal reception time of each sensor node, the propagation time difference of the acoustic signal and vibration signal in the fused signal at the surrounding rock of the current target goaf is determined. The surrounding rock of the current target goaf is discretized into multiple candidate boundary points. For each candidate boundary point, the determined propagation time difference is used as the observation value. Within the corresponding preset position range, the optimal position of the current candidate boundary point is obtained by minimizing the error between the theoretical propagation time difference of the acoustic signal and vibration signal from the candidate boundary point to each sensing node and the observation value. The optimal position of each candidate boundary point is obtained by traversing each candidate boundary point. The optimal positions of each candidate boundary point are connected to form the boundary coordinates of the current target goaf.

6. The method for monitoring coal mine goaf areas according to claim 1, characterized in that, The sensor network includes distributed optical fibers for acquiring vibration signals from the surrounding rock. The rules for stress distribution inversion in the target goaf include: The vibration signal in the fused signal is sequentially subjected to phase demodulation and conversion from the phase of the vibration signal to the displacement caused by vibration at each sensing node of the distributed optical fiber. The time derivative of the converted displacement is then calculated to obtain the vibration acceleration of each sensing node of the corresponding distributed optical fiber. Based on the vibration acceleration of each sensing node of the distributed optical fiber, the elastic modulus and Poisson's ratio of the rock mass in the current target goaf, the stress value of each sensing node of the distributed optical fiber is determined, and the determined stress value is spatially interpolated to obtain the stress distribution data of the current target goaf.

7. The method for monitoring coal mine goaf areas according to claim 1, characterized in that, The multiple indicators include at least the boundary stability indicator for characterizing the intensity and development trend of the convergence deformation of the surrounding rock in the target goaf, the stress concentration indicator for characterizing the stress accumulation degree and the degree of uneven stress distribution of the surrounding rock in the target goaf, and the surrounding rock deformation rate indicator for characterizing the speed of deformation of the surrounding rock in the target goaf. The multi-indicator comprehensive risk assessment of the current target goaf area includes: The current target goaf area is assessed by boundary stability index, stress concentration index, and surrounding rock deformation rate index. The risk assessment result of the current target goaf area is obtained by combining the results of the boundary stability index assessment, stress concentration index assessment, and surrounding rock deformation rate index assessment.

8. A coal mine goaf monitoring system, characterized in that, include: The signal fusion module is used to collect multi-dimensional signals of the surrounding rock in the target goaf through a sensor network deployed in the target goaf, and to perform inference and fusion on the collected multi-dimensional signals of the surrounding rock to obtain a fused signal. The boundary identification module is used to perform boundary identification and stress distribution inversion of the current target goaf area one by one based on the fused signal, so as to obtain the boundary coordinates and stress distribution data of the current target goaf area. The risk warning module is used to conduct a multi-indicator comprehensive risk assessment of the current target goaf based on the boundary coordinates and stress distribution data of the current target goaf. Based on the risk assessment results, it matches the risk level of the current target goaf and triggers the corresponding risk warning scheme according to the matched risk level. Furthermore, based on the aforementioned risk assessment results, the deployed sensor network is adjusted.

9. A machine-readable storage medium storing instructions thereon, characterized in that, When executed by a processor, the instruction causes the processor to be configured to perform the coal mine goaf monitoring method according to any one of claims 1 to 7.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the coal mine goaf monitoring method according to any one of claims 1 to 7.

Images