A mine pressure data intelligent analysis system and method for close-range multi-coal seam combined mining

Abstract

The application discloses a mine pressure data intelligent analysis system and method for close-range multi-coal seam combined mining, and particularly relates to the technical field of mine pressure monitoring and safety control, and is used for solving the problem that the existing method is difficult to distinguish the independent contribution of different disturbance sources under multi-source coupled disturbance, leading to the problem that the support strategy and equipment operation cannot accurately match the rock stratum control requirement; by acquiring and preprocessing the multi-source mine pressure monitoring data of the target lower working face, the feature analysis is performed on the standardized time series data to identify the data feature mode; the transmission impedance index reflecting the disturbance transmission attenuation characteristics is determined in combination with the geological lithology parameters, the amplitude-frequency correction is performed on each data feature mode; the dynamic change trend and stability of the energy representation proportion of each disturbance source after correction are analyzed under different time scales to determine the dominant state and state transfer risk, and the support strategy and equipment operation parameters of the working face are dynamically adjusted according to the determination result; and the accurate analysis and quantitative evaluation of different disturbance sources are realized.

Description

Technical Field

[0001] This invention relates to the field of mine pressure monitoring and safety control technology, and more specifically, to an intelligent analysis system and method for mine pressure data in close-range multi-coal seam joint mining. Background Technology

[0002] In the field of underground coal mining, especially under the condition of close-range multi-coal seam joint mining, effective monitoring and analysis of mine pressure is the key to ensuring the safe advancement of the working face. Existing technologies mainly rely on deploying sensor networks in roadways and working faces to collect multi-source data such as support resistance, roof subsidence, and roadway convergence. Based on historical data analysis and engineering experience models, roof pressure prediction and support strength assessment are carried out. The aim is to judge the rock strata movement state through monitoring data and provide a basis for safety decision-making. This falls into the typical category of mine safety devices and methods.

[0003] However, in environments with repeated mining of multiple coal seams in close proximity, the stress and deformation fields of the surrounding rock of the lower working face are a composite result of the combined effects of the residual structure of the overlying coal seam, inter-layer rock pillars, and the mining of this layer. Existing analytical methods struggle to effectively distinguish and quantify the independent contributions of different sources of disturbance when processing mine pressure data generated by such multi-source coupled disturbances. This leads to inherent biases in the early warning models and thresholds established based on mixed data, making it impossible for support strategies and equipment scheduling schemes developed based on such analyses to accurately match the actual rock strata control requirements. Often, conservative strategies with general reinforcement are adopted for safety redundancy considerations, resulting in significant waste of support materials and equipment operating energy consumption. There is a prominent contradiction between improving safety and efficient resource utilization. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, the present invention provides an intelligent analysis system and method for mine pressure data in close-range multi-coal seam joint mining to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: A method for intelligent analysis of mine pressure data in close-range multi-coal-seam joint mining includes the following steps: S1. Obtain multi-source mine pressure monitoring data of the target lower working face during the mining process; S2. Preprocess the multi-source mine pressure monitoring data to obtain a standardized time-series data sequence; S3. Perform feature analysis on the standardized time series data to identify data feature patterns that characterize disturbances of the overlying coal seam's residual structure, inter-layer rock pillars, and mining disturbances within the same layer. S4. Based on the characteristic patterns of each data and combined with the geological and lithological parameters of the target area, determine the transmission impedance index that reflects the attenuation characteristics of the rock strata to the disturbance transmission between different disturbance sources and monitoring points. S5. Based on the transmission impedance index, the amplitude and frequency of each data characteristic mode are corrected to obtain the corrected energy characterization of each disturbance source. At different time scales, the dynamic change trend and stability of the proportion of the corrected energy characterization of each disturbance source are analyzed to determine the dominant state and state transition risk of the overlying coal seam residual structure disturbance, interlayer rock column disturbance and mining disturbance in this layer in the multi-source mine pressure monitoring data. S6. Based on the dominant state and state transition risk, dynamically adjust the current support strategy and equipment operating parameters for the target lower working face.

[0006] Furthermore, S1 includes: Multi-source mine pressure monitoring data includes time-series data of support resistance of hydraulic supports in the target lower working face collected by pressure sensors, time-series data of the convergence of top and bottom plates between the top and bottom plates in the target lower working face collected by displacement sensors, and time-series data of roadway surface displacement of adjacent roadways in the target lower working face collected by convergence measuring points.

[0007] Furthermore, S2 includes: Outlier removal and interpolation repair were performed on the time series data of support resistance, roof and floor approach, and roadway surface displacement, respectively. The time-series data of the repaired support resistance, the time-series data of the roof and floor convergence, and the time-series data of the roadway surface displacement are aligned with the timestamps of the unified sampling interval. The time-series data of support resistance, roof and floor approach, and roadway surface displacement after timestamp alignment are normalized to obtain a standardized time-series data sequence.

[0008] Furthermore, S3 includes: Time-domain statistical feature extraction and frequency-domain spectral feature extraction are performed on standardized time-series data sequences; Based on the extracted time-domain statistical features and frequency-domain spectral features, combined with the long-period characteristics of the overlying coal seam residual structure disturbance, the medium-period and cooperative deformation characteristics of the inter-layer rock pillar disturbance, and the short-period and strong dynamic characteristics of the mining disturbance of the current layer, data feature patterns that characterize the overlying coal seam residual structure disturbance, the inter-layer rock pillar disturbance, and the mining disturbance of the current layer are separated and identified from the standardized time-series data sequence.

[0009] Furthermore, S4 includes: Geological and lithological parameters include the thickness, elastic modulus, and Poisson's ratio of each stratum between the overlying mined coal seam and the target lower working face; Determining the transmission impedance index involves calculating the equivalent stiffness and damping coefficient of the rock strata structure between different disturbance sources and monitoring points based on the disturbance transmission path and geological lithology parameters corresponding to each data characteristic pattern, and determining the transmission impedance index that reflects the corresponding disturbance transmission attenuation characteristics based on the equivalent stiffness and damping coefficient.

[0010] Furthermore, geological lithological parameters are obtained by analyzing geological exploration reports and borehole columnar sections of the target area and correlated with the spatial location of the target underlying working face.

[0011] Furthermore, S5 includes: The amplitude and frequency response of the data characteristic mode characterizing the disturbance source are compensated by using the transmission impedance index of the corresponding disturbance source, so as to obtain the corrected energy characterization of each disturbance source after eliminating the influence of transmission path attenuation. The analysis includes calculating the real-time proportion of the corrected energy representations of the overlying coal seam residual structure disturbance, inter-layer rock pillar disturbance, and mining disturbance in the current layer in the total energy at long-period, medium-period, and short-period time scales, and statistically analyzing the fluctuation amplitude and rate of change of the real-time proportions at each time scale. The determination includes determining that when the corrected energy representation ratio of a certain disturbance source is consistently higher than a set threshold and the fluctuation amplitude is lower than a set threshold at the corresponding time scale, the disturbance source is determined to be in a dominant state. When the corrected energy representation ratios of different disturbance sources are detected to change rapidly with one increasing and the rate of change exceeds a set threshold, a state transition risk is determined to exist.

[0012] Furthermore, the long-cycle timescale is determined based on the total interval from the end of mining of the overlying coal seam to the current mining time of the target lower working face, the medium-cycle timescale is determined based on the stress redistribution cycle of the interlayer rock pillar, and the short-cycle timescale is determined based on the cyclic propulsion cycle of the hydraulic support of the target lower working face.

[0013] Furthermore, S6 includes: When it is determined that the disturbance of the overlying coal seam's residual structure is in a dominant state, the preset value of the support strength for the periodic pressure on the roof of the target lower working face is reduced and the existing advance speed of the hydraulic support is maintained. When rock pillar disturbance is dominant during the judgment period, increase the support density of the advance support section of the roadway and reduce the average coal cutting speed of the coal mining machine. When it is determined that the disturbance in this layer of mining is in a dominant state, restore the preset value of the support strength for the periodic pressure on the roof of the target lower working face and adjust the advance speed of the hydraulic support according to the standard procedure. When a state transition risk is identified, an enhanced monitoring mode is activated and the current support strategy and equipment operating parameters are temporarily maintained unchanged.

[0014] On the other hand, the present invention provides an intelligent analysis system for mine pressure data in close-range multi-coal seam joint mining, comprising the following modules: The data acquisition module is used to acquire multi-source mine pressure monitoring data of the target lower working face during the mining process; The sequence generation module is used to preprocess multi-source mine pressure monitoring data to obtain standardized time-series data sequences; The pattern recognition module is used to perform feature analysis on standardized time-series data sequences and identify data feature patterns that characterize disturbances of the overlying coal seam's residual structure, inter-layer rock pillars, and mining disturbances within the same layer. The indicator determination module is used to determine the transmission impedance index, which reflects the attenuation characteristics of the rock strata to the disturbance transmission between different disturbance sources and monitoring points, based on the characteristic patterns of each data and the geological lithology parameters of the target area. The risk assessment module is used to perform amplitude-frequency correction on each data characteristic mode based on the transmission impedance index to obtain the corrected energy characterization of each disturbance source. At different time scales, it analyzes the dynamic change trend and stability of the proportion of the corrected energy characterization of each disturbance source to determine the dominant state and state transition risk of disturbances from the overlying coal seam residual structure, interlayer rock pillars, and mining disturbances in the current layer in the multi-source mine pressure monitoring data. The parameter adjustment module is used to dynamically adjust the current support strategy and equipment operating parameters for the target lower working face based on the dominant state and state transition risk.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. By deeply processing multi-source mine pressure monitoring data and accurately analyzing disturbance sources, the accuracy and reliability of mine pressure analysis under near-distance multi-coal seam mining conditions are effectively improved. Standardized preprocessing and feature analysis of multi-source data enable the separation and identification of characteristic patterns of different disturbance sources, such as overlying residual structures, inter-layer rock pillars, and mining within the same layer, from mixed monitoring signals. Furthermore, by combining geological and lithological parameters to determine the transmission impedance index, amplitude and frequency correction is performed on the energy characterization of each disturbance source to restore the true disturbance intensity. The dominant state and transfer risk of each disturbance source are dynamically assessed at different time scales, so that mine pressure analysis no longer relies on empirical interpretation of mixed data, but instead constructs an analytical framework based on the physical mechanism of disturbance transmission. This allows for a more accurate revelation of the composite effect of the surrounding rock stress field, providing a more refined and reliable basis for roof pressure prediction and stratum movement judgment, and significantly enhancing the safety management capability of the working face in complex mining environments.

[0016] 2. Based on the accurate determination of the dominant state of disturbance sources, dynamic adaptive adjustment of support strategies and equipment operating parameters is realized. When a specific disturbance source is identified as dominant, the system can adjust key parameters such as support strength and advance speed in a targeted manner. For example, when the overlying residual structure dominates, the preset support value can be appropriately reduced, and when the interlayer rock pillar dominates, the roadway support can be strengthened and the coal mining machine speed can be controlled. This on-demand allocation strategy avoids the conservative support and operation mode with high strength and high energy consumption that is commonly used in traditional methods due to the inability to distinguish disturbance sources. It makes the resource input such as support materials and equipment power consumption match the actual rock strata control needs, reduces resource waste under the premise of ensuring safety, and realizes the efficient use of mine equipment and energy, which is in line with the technical direction of equipment energy-saving transformation. Attached Figure Description

[0017] Figure 1 This is a flowchart of an intelligent analysis method for mine pressure data in close-range multi-coal-seam joint mining according to the present invention; Figure 2 This is a schematic diagram of the structure of an intelligent analysis system for mine pressure data in close-range multi-coal seam joint mining according to the present invention. Detailed Implementation

[0018] 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 embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0019] Example 1: Figure 1 This invention presents an intelligent analysis method for mine pressure data in close-range multi-coal-seam joint mining, which includes the following steps: S1. Obtain multi-source mine pressure monitoring data of the target lower working face during the mining process; S2. Preprocess the multi-source mine pressure monitoring data to obtain a standardized time-series data sequence; S3. Perform feature analysis on the standardized time series data to identify data feature patterns that characterize disturbances of the overlying coal seam's residual structure, inter-layer rock pillars, and mining disturbances within the same layer. S4. Based on the characteristic patterns of each data and combined with the geological and lithological parameters of the target area, determine the transmission impedance index that reflects the attenuation characteristics of the rock strata to the disturbance transmission between different disturbance sources and monitoring points. S5. Based on the transmission impedance index, the amplitude and frequency of each data characteristic mode are corrected to obtain the corrected energy characterization of each disturbance source. At different time scales, the dynamic change trend and stability of the proportion of the corrected energy characterization of each disturbance source are analyzed to determine the dominant state and state transition risk of the overlying coal seam residual structure disturbance, interlayer rock column disturbance and mining disturbance in this layer in the multi-source mine pressure monitoring data. S6. Based on the dominant state and state transition risk, dynamically adjust the current support strategy and equipment operating parameters for the target lower working face.

[0020] S1. Obtain multi-source mine pressure monitoring data of the target lower working face during the mining process. The specific implementation is as follows: In close-range multi-coal-seam joint mining, the target lower working face refers to the working face of the currently being mined lower coal seam. The roof strata of this working face are affected by the residual structures of the overlying coal seam, inter-layer rock pillars, and the mining activities within this seam. To effectively monitor mine pressure, it is necessary to acquire multi-source mine pressure monitoring data reflecting the stress and deformation state of the surrounding rock. This data originates from various sensors and measuring points deployed in the working face and adjacent roadways, and is collected continuously or periodically to form time-series data. Specifically, the multi-source mine pressure monitoring data includes time-series data on the support resistance of the hydraulic supports of the target lower working face collected by pressure sensors, time-series data on the convergence of the roof and floor between the roof and floor of the target lower working face collected by displacement sensors, and time-series data on the surface displacement of adjacent roadways of the target lower working face collected by convergence measuring points.

[0021] The acquisition of support resistance time-series data is achieved through pressure sensors installed on the hydraulic supports. The pressure sensors are either resistance strain gauge pressure sensors or hydraulic pressure sensors, with their range determined based on the rated working resistance of the hydraulic support at the working face. For example, when the rated working resistance of the hydraulic support is 12,000 kN, the pressure sensor range is selected from 0 to 15,000 kN to ensure coverage of possible pressure fluctuations. The pressure sensors are specifically installed at the oil circuit interface in the lower cavity of the hydraulic support column or at the oil pressure measurement point of the balancing jack, connected to the hydraulic pipeline via threaded connections or quick-connect interfaces to directly sense changes in fluid pressure during the support process. One pressure sensor is installed on each critical column of each hydraulic support near the goaf and coal face side to obtain spatial distribution information of the support force. Data acquisition is achieved by connecting the sensor's built-in signal cable to a distributed data acquisition unit deployed in the roadway. The acquisition unit records pressure readings at fixed time intervals, called the sampling interval. The sampling interval is set based on the working face advance speed and roof movement patterns, for example, set to collect data every 30 seconds or every 0.5 meters the hydraulic support moves. The sampling interval must be set to ensure that it can capture the sudden changes in resistance during the periodic pressure of the roof. This is typically determined initially by analyzing the historical pressure step distance of working faces with similar geological conditions. For example, if the historical pressure step distance is 15 meters and the average daily advance of the working face is 5 meters, the sampling interval is set to be correlated with the support advance, sampling once every 0.3 to 0.5 meters, and dynamically fine-tuned based on actual data during mining. The analog voltage signal output by the pressure sensor is converted into a digital signal by the analog-to-digital converter inside the data acquisition instrument and stored together with a timestamp generated by the internal clock of the instrument. The timestamp accuracy is 1% of a second, and the recording format is year, month, day, hour, minute, second, millisecond, e.g., 2023-10-27 14:30:25.123. The support resistance time-series data consists of a series of pressure values ​​arranged in chronological order. Each data point includes the sampling time and the corresponding support resistance value, with the resistance value unit being kilonewtons.

[0022] The acquisition of timing data on the convergence of the top and bottom plates is achieved through displacement sensors installed within the working face. Laser rangefinders or ultrasonic displacement sensors are selected, and their measurement range is determined based on the expected maximum convergence of the top and bottom plates. For example, when the expected total convergence is 800 mm, the sensor's measurement range is selected to be 0 to 1000 mm. The displacement sensors are installed between the top beam and base of the hydraulic supports, fixed to stable parts of the supports with metal clamps and locking bolts. Their measuring heads are vertically aligned with pre-set reflective targets on the lower surface of the top beam and the upper surface of the base, directly measuring the relative distance change between them. Along the length of the working face, a displacement monitoring point is placed every 10 hydraulic supports, with one displacement sensor installed at each monitoring point. Data acquisition is connected to a data acquisition instrument via the digital output interface of the displacement sensors. The sampling interval of the acquisition instrument is synchronized with the support resistance data acquisition, i.e., recording is performed using the same time interval or event trigger mode, for example, data is collected every 30 seconds or every 0.5 meters the support moves. The uniformity of the sampling interval ensures the alignment of data from different sources on the time axis. The raw signals collected by the displacement sensors are processed by the built-in processor to calculate the displacement value. This displacement value is the change in the reference distance calibrated at the initial installation moment; a positive value indicates a decrease in the distance between the top and bottom plates. Data recording includes a timestamp synchronized with the support resistance data to ensure time consistency. The time-series data of the top and bottom plate convergence is a set of displacement values ​​sequenced over time, with the unit of displacement being millimeters. It reflects the dynamic process of the relative convergence of the top and bottom plates caused by the combined effects of roof subsidence and floor bulging during mining.

[0023] The acquisition of time-series displacement data on the roadway surface is achieved by setting convergence measuring points in adjacent roadways below the target working face and performing measurements. Convergence measuring points are physical markers fixed at specific locations on the roadway surface. These are typically stainless steel measuring nails with external threads or metal discs with cross-shaped markings, securely installed on the rock surfaces of the roadway roof, floor, and sidewalls by drilling and injecting anchoring agent. The measuring point layout follows the roadway convergence monitoring section design. Within a 100-meter radius of the working face, a monitoring section is set every 5 meters along the roadway direction. Each monitoring section has one measuring point installed at the center of the roadway roof, the center of the floor, and the center of each of the left and right sidewalls, forming a closed measurement loop. The measuring tools are steel tape convergence meters or electronic convergence meters. Steel tape convergence meters are manually tensioned and the length of the steel tape between measuring points is read, while electronic convergence meters automatically record the distance using a built-in electronic encoder. Data acquisition is accomplished through manual periodic measurements or automated monitoring systems. During manual measurements, surveyors use convergence meters to sequentially measure the horizontal and vertical distances between measuring points within the same monitoring section, recording the measurement time and distance readings. The measurement frequency is determined based on the working face's advancing speed; for example, a comprehensive measurement is performed every 2 meters of advancement. Automated monitoring systems install miniature displacement sensors at each measuring point and network them, transmitting real-time distance data wirelessly to the acquisition terminal. The sampling interval can be set to once per hour. The displacement values ​​in the tunnel surface displacement time-series data can be the absolute value of the distance between two adjacent measuring points or the change in distance relative to the initial baseline, i.e., the convergence value. All measurement data are accompanied by precise timestamps, derived from the internal clock of the measuring instrument or manually recorded standard time. These data are organized according to monitoring section and measuring point numbers, forming multiple time-series data sequences that collectively constitute the tunnel surface displacement time-series data, with displacement values ​​in millimeters.

[0024] During the acquisition of the aforementioned multi-source mine pressure monitoring data, to ensure the spatiotemporal consistency of the data, the time reference of all data acquisition devices is synchronized with the master clock server located in the roadway chamber via an industrial Ethernet or fieldbus network. The master clock server receives timing signals from the BeiDou Navigation Satellite System or the Global Positioning System to calibrate the time, with synchronization accuracy required to be at the millisecond level. After acquisition, the data is stored in its original format in a relational database or time-series database, with a joint index established according to data type, sensor identifier, and timestamp during storage. The data acquisition stage also includes real-time diagnosis of sensor operating status. For example, faults are diagnosed by monitoring whether the zero-point drift of the pressure sensor exceeds 0.5% of the full scale or whether the displacement sensor signal remains unchanged for multiple consecutive sampling cycles. Suspicious data segments are marked by adding predefined status codes to the data stream, but the original measurement values ​​are not modified. Through the above specific implementation methods, comprehensive time-series data of support resistance, roof and floor approach, and roadway surface displacement during the mining process of the target lower working face are acquired. These data provide a complete, synchronous, and directly processable raw data foundation for subsequent preprocessing and feature analysis.

[0025] S2. Preprocess the multi-source mine pressure monitoring data to obtain a standardized time-series data sequence. The specific implementation is as follows: Outlier removal and interpolation repair were performed on the time-series data of support resistance, roof and floor approach, and roadway surface displacement. Outlier removal aims to identify and remove obvious erroneous data points caused by sensor momentary failures, transmission interference, or non-rock activity. For support resistance time-series data, the first method for outlier identification is based on the statistical distribution characteristics of the data. Specifically, the moving average and moving standard deviation of the data within a recent data window of the sensor are calculated. The moving average reflects the central trend of the data, and the moving standard deviation reflects the dispersion of the data. The length of the data window is set according to the data sampling frequency and the basic cycle of roof activity. For example, if the sampling interval is 30 seconds, the data window length can be set to the number of data points corresponding to 2 hours, i.e., 240 data points. The moving average and moving standard deviation are dynamically updated as new data are added. The threshold for identifying outliers is set as the moving average plus or minus K times the moving standard deviation. The value of K is determined based on the statistical characteristics of historical normal data fluctuations. For example, by analyzing data from a period without equipment failure in the past week, the relationship between the standard deviation multiple and the false alarm rate of outliers is calculated, and K is set to 3. This means that when the resistance value of a data point is higher than the moving average plus 3 times the moving standard deviation or lower than the moving average minus 3 times the moving standard deviation, that point is considered an outlier. A second method for identifying outliers is based on physical rationality. For example, absolute upper and lower limits are set for the resistance value of the hydraulic support. The lower limit is the minimum supporting force of the hydraulic support in its unloaded state, such as 100 kN, and the upper limit is 1.2 times the rated working resistance of the hydraulic support, such as 14400 kN. Data points exceeding this absolute range are directly identified as outliers. A third method for identifying outliers is to identify continuous constant values, i.e., output that remains unchanged due to sensor failure. For example, when the resistance values ​​of 10 or more consecutive data points are exactly the same, this sequence is considered an anomaly. For time-series data on the convergence of roof and floor plates, outlier detection also employs methods based on moving averages and moving standard deviations. However, the physical upper and lower limits are set according to the maximum permissible convergence of the roof and floor plates; for example, the displacement value should not be negative and should not exceed the sensor's range of 1000 mm. For time-series data on roadway surface displacement, outlier detection mainly considers the reasonableness of changes between two adjacent measurements. For example, points with a single-day convergence change exceeding 20 mm and significantly inconsistent with the trend of changes at surrounding measuring points are considered outliers. Interpolation repair is used to fill data gaps caused by outlier removal or missing original data. Interpolation methods include linear interpolation or interpolation based on time series trends. Linear interpolation is suitable for situations where data changes are stable before and after a gap. Specifically, at the gap time point, the estimated value of the point is calculated according to the values ​​and times of the previous and subsequent valid data points, proportional to time. Interpolation based on time series trends is suitable for situations where the data has obvious periodicity or trends. Specifically, a local trend line or curve is fitted using multiple valid data points before and after the gap, and then this trend model is used to predict the value of the gap.For example, for support resistance data, linear interpolation can be used during the stable phase without pressure, while interpolation repair can be performed using quadratic curves fitted from multiple points before and after the pressure change phase. After interpolation repair, abnormal or missing positions in the original data sequence are replaced with reasonable estimates, forming repaired support resistance time series data, repaired roof and floor approach time series data, and repaired roadway surface displacement time series data.

[0026] The time-series data of repaired support resistance, repaired roof and floor approach, and repaired roadway surface displacement are aligned using a unified sampling interval and timestamp alignment. The purpose of timestamp alignment is to enable data sequences from different sensors with different original sampling times or frequencies to be compared and fused at a unified and consistent point in time. First, the target sampling interval is determined. The selection of the target sampling interval is based on the highest sampling frequency in the data source and the analysis requirements. For example, if the original sampling interval for support resistance and roof and floor approach time-series data is 30 seconds, while the roadway surface displacement time-series data is measured manually once a day, the target sampling interval is usually set to an interval no lower than the highest frequency data, such as 30 seconds. Data with lower frequencies need to be aligned to this interval using interpolation methods. For data sequences with an original sampling frequency higher than the target sampling interval, for example, if the original sampling interval for a sensor is 10 seconds and the target interval is 30 seconds, a resampling method is used for alignment. Specifically, at each target time point, such as a full 30 seconds, one or more of the nearest original data points before and after that time are selected, and their arithmetic mean is calculated as the aligned data value for that target time. For data sequences whose original sampling frequency is lower than the target sampling interval, such as time series data of roadway surface displacement, interpolation methods are required to align the data to the target time point. The interpolation method is similar to the aforementioned interpolation repair method; for example, piecewise linear interpolation is used, calculating an interpolation estimate for the target time point based on two adjacent actual measurements and their timestamps. During timestamp alignment, it is essential to ensure that the time base of all data sequences is completely consistent, meaning all timestamps are based on the same synchronized clock source, which has been uniformly calibrated by the master clock server during the data acquisition step. The alignment process generates a set of data with identical time point sequences; each target time point corresponds to a support resistance value, a roof and floor approach value, and a roadway surface displacement value, thus forming timestamp-aligned support resistance time series data, timestamp-aligned roof and floor approach time series data, and timestamp-aligned roadway surface displacement time series data. These data achieve strict synchronization in the time dimension, laying the foundation for subsequent joint analysis.

[0027] The time-series data of support resistance, roof and floor approach, and roadway surface displacement, all aligned with timestamps, were normalized to obtain a standardized time-series data sequence. The purpose of normalization is to eliminate the influence of differences in dimensions and numerical ranges between different monitored physical quantities, enabling subsequent feature analysis to be performed at the same scale. The normalization process employed a minimum-maximum normalization method. For the time-series data of support resistance aligned with timestamps, the baseline parameters for normalization, namely the minimum and maximum values, were first calculated. The minimum and maximum values ​​were determined based on a stable data segment representing a normal mining phase, such as a one-week data segment without large roof pressure events. The minimum and maximum values ​​of all support resistance values ​​within this data segment were calculated and denoted as the minimum baseline resistance value and the maximum baseline resistance value, respectively. The minimum baseline resistance value is typically close to the initial support force range of the hydraulic support, while the maximum baseline resistance value is typically close to the resistance peak during the normal period of pressure on the working face. Then, for each resistance value in the time series data of support resistance after time-stamp alignment, a normalization calculation is performed using the formula: the normalized resistance value equals the resistance value minus the minimum reference resistance value, and the difference is then divided by the difference between the maximum and minimum reference resistance values. This calculation maps the original resistance value to a numerical range between zero and one. If subsequent extreme values ​​exceeding the original reference range occur, such as abnormally high resistance greater than the maximum reference resistance value, the normalized value may be greater than one, which will be retained as valid feature information. For the time series data of roof and floor convergence after time-stamp alignment, the same normalization method is used, i.e., the minimum and maximum reference convergence are calculated by selecting stable data segments, and then normalization is performed on each convergence value. For the time series data of roadway surface displacement after time-stamp alignment, since it may be displacement values ​​from multiple measuring points in multiple directions, it is necessary to perform independent normalization processing on the data sequence of each measuring point in each direction, and calculate their respective minimum and maximum reference displacement values ​​separately. After the above normalization calculation, the values ​​of all data sequences are transformed to the same dimensionless scaling range, thus obtaining a standardized time series data sequence. The standardized time series data sequence is a multivariate time series dataset, in which each time point contains a normalized support resistance value, a normalized roof and floor approach value, and multiple normalized roadway surface displacement values. These values ​​fluctuate around the zero-one interval, are comparable, and facilitate the subsequent extraction of statistical and frequency domain features at a uniform scale.

[0028] S3. Perform feature analysis on the standardized time-series data sequences to identify data feature patterns representing disturbances from the overlying coal seam's residual structure, inter-layer rock pillars, and mining disturbances within the seam. Specifically, this is implemented as follows: Time-domain statistical feature extraction and frequency-domain spectral feature extraction are performed on standardized time-series data sequences. Time-domain statistical feature extraction aims to directly quantify the statistical regularity and fluctuation patterns of the data sequence from a time dimension. For each data variable in the standardized time-series data sequence, such as the normalized support resistance value sequence, a set of time-domain statistical feature values ​​are calculated within a sliding analysis window. The length of the analysis window is set according to the physical cycle of data change. For support resistance data, the window length typically corresponds to the number of data points covered by several cycles of hydraulic support advancement at the working face. For example, if the average cycle advancement distance of the hydraulic support is 0.8 meters, and the daily advancement of the working face is 5 meters, then the window length can be selected to correspond to a data time range of 10 to 15 meters of advancement. Within the set window, the calculated time-domain statistical features include mean, standard deviation, skewness, kurtosis, root mean square value, and kurtosis factor. The mean is the arithmetic mean of all data points within the window, reflecting the average level of the data. The standard deviation is the square root of the sum of the squares of the deviations of each data point from the mean, reflecting the dispersion of the data. Skewness measures the asymmetry of data distribution, calculated as the ratio of the third central moment to the cube of the standard deviation. A skewness greater than zero indicates a long tail in the positive direction. Kurtosis measures the sharpness of data distribution, calculated as the ratio of the fourth central moment to the fourth power of the standard deviation minus 3. A kurtosis greater than zero indicates a sharper distribution than a normal distribution. The root mean square (RMS) value is calculated as the square root of the arithmetic mean of the squares of the data points within the window, reflecting the energy level of the data. The kurtosis factor is calculated as the ratio of the maximum absolute value of the data within the window to the RMS value, reflecting the strength of the impact component in the data. For time series data on roof and floor approach and tunnel surface displacement, the same window sliding method is used to calculate their respective time-domain statistical characteristic sequences. Frequency domain spectral feature extraction aims to convert time-domain data to the frequency domain and analyze its frequency components and energy distribution. The Discrete Fourier Transform (DFT) is used to convert each variable in the standardized time-series data from the time domain to the frequency domain. First, the input parameters of the DFT, i.e., a data segment for analysis, need to be determined. The analysis segment should be long enough to encompass the complete period of the lowest frequency to be analyzed, typically a continuous data segment equal to or longer than the analysis window length. A Discrete Fourier Transform is performed on the data within the analysis segment. The result is a series of complex numbers, each corresponding to a specific frequency component. The frequency value of this component is equal to the sampling frequency divided by the analysis segment length and then multiplied by the component's index. The sampling frequency is the reciprocal of the data sampling interval; for example, if the sampling interval is 30 seconds, the sampling frequency is one-sixtieth of a Hertz. The power spectral density value of each frequency component is obtained by calculating the square of its modulus. The power spectral density values ​​of all frequency components constitute the power spectrum of the analysis segment. Frequency domain spectral features extracted from the power spectrum include the dominant frequency, band energy, and spectral entropy.The dominant frequency refers to the frequency value corresponding to the frequency component with the largest power spectral density value within a defined effective frequency range. The upper limit of the effective frequency range is the Nyquist frequency, which is half the sampling frequency, while the lower limit is determined by the length of the analysis segment. For example, if the segment length is one day's worth of data, the lowest analyzable frequency is approximately 1 / 100,000th of a Hz. Band energy refers to dividing the entire effective frequency range into several continuous bands and calculating the sum of the power spectral density values ​​of all frequency components within each band. For example, bands could be divided into low-frequency, mid-low-frequency, mid-high-frequency, and high-frequency bands. The boundary frequencies for these band divisions need to be pre-set based on the theoretical frequency characteristics of different disturbance sources. Spectral entropy measures the uniformity of spectral energy distribution. It is calculated by first normalizing the power spectral density values ​​across the entire effective frequency range until their sum is one, then treating these normalized values ​​as a probability distribution and calculating their information entropy. A larger spectral entropy value indicates a more dispersed spectral energy distribution. By performing a sliding-window discrete Fourier transform on a standardized time-series data sequence, a frequency domain spectral characteristic sequence varying over time can be obtained.

[0029] Based on the extracted time-domain statistical features and frequency-domain spectral features, combined with the long-period characteristics of the overlying coal seam residual structure disturbance, the medium-period and cooperative deformation characteristics of the inter-layer rock pillar disturbance, and the short-period and strong dynamic characteristics of the mining disturbance of the current layer, data feature patterns that characterize the overlying coal seam residual structure disturbance, the inter-layer rock pillar disturbance, and the mining disturbance of the current layer are separated and identified from the standardized time-series data sequence. Long-period characteristics refer to the residual structures of overlying coal seams, such as residual coal pillars, goaf caving zones, and fracture zones. Their stress adjustment and transmission processes are slow, and their impact on the lower working face is manifested as long-lasting and slowly changing loads or deformations. In the time domain, this is reflected in the slow trend changes of the mean and standard deviation on a scale of several days or even several weeks. In the frequency domain, this is reflected in the extremely low dominant frequency, with the main energy concentrated in the ultra-low frequency band close to zero Hertz and the low frequency band below a specific threshold. For example, this threshold can be estimated based on the basic period of movement of the overlying strata. If the basic period is one month, the corresponding frequency is approximately 1 / 386,000 Hertz. Medium-period and coordinated deformation characteristics refer to rock pillars located between upper and lower coal seams, whose disturbance transmission period lies between that of legacy structures and current mining. Due to the integrity of the rock pillars, their deformation often exhibits high spatiotemporal correlation among different measuring points on the roadway surface displacement. In the time domain, this is reflected in the moderate fluctuation period of roof and floor approach and roadway displacement, such as several changes within hours to a day. In the frequency domain, the dominant frequency is in the low to mid-frequency band, and the spectra of roadway displacement sequences at different locations show significant coherence in specific low to mid-frequency bands. Short-period and strong dynamic characteristics refer to disturbances in this mining layer directly caused by processes such as face advancement and roof fracturing. These disturbances change rapidly and are strongly correlated with the production cycle. In the time domain, this is reflected in the data sequence showing sharp peaks and high volatility, with peak factor and kurtosis values ​​significantly increasing at specific times. In the frequency domain, the dominant frequency is high, with energy mainly distributed in the mid to high-frequency band, and it is correlated with the frequency components of the hydraulic support propulsion cycle. The separation and identification process specifically involves first setting frequency thresholds and correlation coefficient thresholds to distinguish different disturbance modes. The frequency thresholds include a first frequency threshold separating long-term and medium-term periods, and a second frequency threshold separating medium-term and short-term periods. The first frequency threshold is determined based on the statistical median of the theoretical lowest frequency influenced by the overlying strata's residual structure and the highest possible frequency of inter-stratum rock column disturbance. For example, it can be set by analyzing the inflection point of the spectral energy distribution during periods without mining in the current stratum in historical data. The second frequency threshold is determined based on the statistical median of the basic frequency of the hydraulic support cycle operation during mining in the current stratum and the highest frequency of inter-stratum rock column deformation. For example, it can be set by statistically analyzing the frequency range corresponding to the main peak of the spectral energy in data from normal mining periods. The correlation coefficient threshold is used to determine the characteristics of coordinated deformation. A threshold, such as 0.7, is set by calculating the time-domain correlation coefficient or frequency-domain coherence coefficient of displacement sequences at different roadway measuring points within a sliding window. When the correlation coefficient is consistently higher than this threshold, coordinated deformation is considered to exist.The identification process is based on the aforementioned characteristics and thresholds: When the frequency domain spectral features extracted from the standardized time-series data show that, within an analysis window exceeding one day, the dominant frequency remains consistently below the first frequency threshold, and the proportion of low-frequency band energy to total energy exceeds a set threshold, such as 60%, the data pattern for that period is determined to be a data characteristic pattern representing disturbances in the overlying coal seam's legacy structure. When the frequency domain spectral features show that the dominant frequency is between the first and second frequency thresholds, and the time-domain correlation coefficient or coherence coefficient of a specific mid-to-low frequency band in the displacement sequences of at least two different roadway measuring points remains consistently above the correlation coefficient threshold, the data pattern for that period is determined to be a data characteristic pattern representing disturbances in the interlayer rock pillar. When the peak factor and kurtosis value in the time domain statistical features significantly increase within a short window, such as one hour, and exceed twice the standard deviation of their historical background values, and the frequency domain spectral features show that the dominant frequency is above the second frequency threshold, and the proportion of mid-to-high frequency band energy significantly increases, the data pattern for that period is determined to be a data characteristic pattern representing disturbances in the current layer's mining operations. By continuously applying the above-mentioned judgment logic throughout the entire mining timeline, data feature patterns that may alternate or overlap in time and belong to different disturbance sources can be separated and identified from continuous, mixed, standardized time-series data sequences.

[0030] S4. Based on the characteristic patterns of each data point and combined with the geological lithology parameters of the target area, determine the transmission impedance index that reflects the attenuation characteristics of disturbance transmission from different disturbance sources to monitoring points. The specific implementation is as follows: Geological lithological parameters include the thickness, elastic modulus, and Poisson's ratio of each stratum between the overlying coal seam and the target lower working face. These parameters are obtained by analyzing the geological exploration report and borehole columnar sections of the target area and correlated with the spatial location of the target lower working face. The geological exploration report is a comprehensive technical document generated after geological surveys of the target mining area before mining commence. It contains information on the distribution, occurrence, and physical and mechanical properties of strata obtained through surface drilling, geophysical exploration, and other methods. A borehole columnar section is a specific graphical form within the geological exploration report. It visually displays, in depth order, the lithological names, thicknesses, depths, and rock mechanical parameters obtained through core laboratory testing of each stratum traversed by a borehole. The method for obtaining thickness parameters from the borehole columnar section involves identifying all independent strata within a vertical range from the bottom of the overlying coal seam to the top of the target lower working face, and then reading the vertical thickness of each stratum marked on the columnar section, typically in meters. The method for obtaining the elastic modulus and Poisson's ratio parameters is to consult the rock mechanics test report accompanying the borehole columnar section. This report presents uniaxial or triaxial compression tests on core samples taken from rock strata at different depths, thus obtaining the static elastic modulus and Poisson's ratio values ​​for the corresponding rock strata. If the report lacks direct test data for a particular rock stratum, the lithological description of that stratum, such as sandstone or mudstone, is referenced. The average value of the empirical range of rock mechanics parameters for similar lithologies is used as the elastic modulus and Poisson's ratio for that stratum. The empirical range is derived from publicly available rock mechanics handbooks or historical test databases for the mining area. Since a target area typically has multiple exploration boreholes, it is necessary to compare and synthesize the strata information revealed by all boreholes. A standard strata columnar section of the target area is established using stratigraphic correlation methods. This columnar section reflects the typical spatial sequence and thickness of the strata between the overlying mined coal seam and the target lower working face. Associating with the spatial location of the target lower working face means overlaying the spatial range of the target lower working face in the three-dimensional coordinate system of the mining area with the standard rock stratum columnar section to determine the rock stratum sequence and thickness corresponding to different positions directly above the working face. In particular, for the time of occurrence of the data feature patterns of different disturbance sources identified and the working face advance position, the specific rock stratum path from the monitoring point directly above the working face to the disturbance source at that time is determined, thereby achieving a precise correlation between geological lithology parameters and dynamic mining position.

[0031] Determining the transmission impedance index involves calculating the equivalent stiffness and damping coefficient of the rock strata structure between different disturbance sources and monitoring points based on the disturbance transmission path and geological lithology parameters corresponding to each data feature pattern. The transmission impedance index reflecting the corresponding disturbance transmission attenuation characteristics is then determined based on the equivalent stiffness and damping coefficient. The disturbance transmission path corresponding to each data feature pattern needs to be defined geometrically and mechanically according to the disturbance source type and monitoring point location. For data feature patterns characterizing disturbances caused by residual structures in the overlying coal seam, the disturbance source is located as a specific residual structure in the overlying mined coal seam, such as the boundary of a residual coal pillar or the middle of a goaf. The disturbance transmission path is defined as the rock mass path starting from the bottom point of this residual structure, penetrating vertically downwards or along the direction of maximum principal stress through the interlayer rock strata, and finally reaching the monitoring point at the hydraulic support of the target lower working face. This path is simplified into a series of mechanical transmission chains composed of interconnected rock strata of different lithologies. For data characteristic patterns representing disturbances in inter-layer rock pillars, the disturbance source is located as an inter-layer rock pillar in the plastic deformation zone around the working face. The disturbance transmission path is defined as the rock mass path starting from a point on the sidewall or bottom of the rock pillar, passing through part of the rock pillar and surrounding rock, and reaching the displacement monitoring point on the roadway surface in an approximately horizontal or inclined direction. For data characteristic patterns representing disturbances in the current working face, the disturbance source is located as the fracture line of the coal wall or the roof of the goaf in the current working face. The disturbance transmission path is defined as the rock mass path starting from the source point, passing through the immediate roof or the main roof strata, and reaching the support resistance or roof and floor approach measurement monitoring point in a shorter path. When calculating the equivalent stiffness, each disturbance transmission path segment is considered as a combination of different rock strata segments connected in series. For a path segment transmitting along the normal direction of the rock strata, its axial stiffness is equal to the sum of the stiffnesses of all the rock strata traversed by that path in series. The stiffness of a single rock stratum is calculated by multiplying its elastic modulus by its cross-sectional area and then dividing by its thickness in the transmission direction. The cross-sectional area is taken as a representative unit area, such as one square meter. For path segments along the bedding plane or obliquely traversing the rock strata, the elastic modulus needs to be directionally corrected according to the angle between the path and the bedding plane. For example, when the path is parallel to the bedding plane, the elastic modulus of the rock in the direction parallel to the bedding plane is used. This value can be estimated by consulting the mechanical parameter table of anisotropic rocks or by using empirical formulas based on the elastic modulus perpendicular to the bedding plane. The total equivalent stiffness of the entire transmission path is calculated by combining the stiffnesses of each path segment in series or parallel relationships. In series relationships, the reciprocal of the total equivalent stiffness is equal to the sum of the reciprocals of the stiffnesses of each series segment; in parallel relationships, the total equivalent stiffness is equal to the sum of the stiffnesses of each parallel segment.When calculating the damping coefficient, a path model similar to that used for stiffness calculation is employed, but the elastic modulus is replaced with an equivalent viscous damping coefficient characterizing the effects of internal friction and fractures within the rock strata. This damping coefficient is related to the elastic modulus, Poisson's ratio, and rock mass integrity coefficient of the rock strata and can be obtained through empirical relationships. For example, the equivalent viscous damping coefficient equals the elastic modulus of the rock strata multiplied by a dimensionless damping ratio. The damping ratio is selected empirically within the range of 0.01 to 0.05 based on the lithology, with lower values ​​for hard rocks with high integrity coefficients and higher values ​​for fractured soft rocks. The transmission impedance index is determined based on the equivalent stiffness and damping coefficient. The transmission impedance index is a physical quantity that comprehensively reflects the resistance of the rock strata structure to wave disturbances. Its calculation can be expressed as the product of the equivalent stiffness and a factor related to the damping coefficient and the characteristic frequency of the disturbance. Specifically, for harmonic disturbances, the transmission impedance index can be calculated as the square of the equivalent stiffness plus the square of the damping coefficient multiplied by the angular frequency, and then the square root of the sum, where the angular frequency is equal to two times pi multiplied by the dominant frequency corresponding to the characteristic pattern of the disturbance data. For complex, non-harmonic disturbances, an approximate calculation can be performed based on its dominant frequency component, or the weighted geometric mean of the equivalent stiffness and damping coefficient can be used as the transmission impedance index. The weights are determined according to the distribution of disturbance energy in different frequency bands; for example, if the disturbance energy is mainly concentrated in low frequencies, the equivalent stiffness is given a higher weight. The final calculated transmission impedance index is a physical quantity with units, its dimension being the same as the dimension of force divided by the dimension of displacement, for example, kilonewtons per millimeter. The larger this value, the stronger the attenuation effect of the rock strata structure on the transmission of the disturbance from the disturbance source to the monitoring point, that is, the higher the proportion of energy lost during the transmission process. By repeating the above path definition, parameter acquisition, and calculation process for each identified data feature pattern, based on its corresponding disturbance source type and the working surface location at the time of occurrence, the specific transmission impedance index value reflecting the transmission attenuation characteristics of different disturbances can be determined.

[0032] S5. Based on the transmission impedance index, amplitude-frequency correction is performed on each data characteristic mode to obtain the corrected energy characterization of each disturbance source. At different time scales, the dynamic change trend and stability of the proportion of the corrected energy characterization of each disturbance source are analyzed to determine the dominant state and state transition risk of disturbances from the overlying coal seam's residual structure, inter-layer rock pillars, and mining disturbances in the current layer in the multi-source mine pressure monitoring data. The specific implementation is as follows: The amplitude and frequency response of the data characteristic patterns representing the disturbance source are compensated using the transmission impedance index of the corresponding disturbance source, resulting in the corrected energy characterization of each disturbance source after eliminating the attenuation effect of the transmission path. The compensation calculation is based on the physical principle of wave propagation in rock strata, with the core assumption that the amplitude attenuation and phase change of the signal received by the monitoring point are mainly caused by the impedance characteristics of the transmission path. For each identified data characteristic pattern representing a specific disturbance source, its original amplitude information comes from the time-domain or frequency-domain characteristic values ​​corresponding to that pattern. For example, for a harmonic disturbance, its amplitude can be directly obtained from the spectral amplitude corresponding to the dominant frequency of the pattern; for a non-harmonic complex pattern, its amplitude can be characterized by the root mean square value of the time-domain data within the pattern's time period. Frequency response compensation mainly targets phase delay, but in this energy correction, the focus is on the recovery of amplitude energy. The specific implementation of the amplitude compensation calculation involves multiplying the original amplitude received by the monitoring point by a compensation coefficient, which is a function of the transmission impedance index. The compensation coefficient can be determined based on theoretical models or empirical calibration. One approach based on theoretical models simplifies the transmission path to a single-degree-of-freedom spring-damped system, where the reciprocal of the gain of its transfer function at a specific frequency is the compensation coefficient. This gain calculation involves the transmission impedance index, the disturbance frequency, and a reference impedance. The reference impedance is defined as the impedance of an ideal, lossless path, calculated using the stiffness of a standard path in a perfectly elastic medium over an extremely short distance. The compensation coefficient is equal to the ratio of the transmission impedance index to the reference impedance. Another approach based on empirical calibration selects a period in the early stages of mining where a single disturbance source is dominant and the transmission path is relatively clear, such as the initial pressure stage affected only by disturbances from the current mining layer. Through inversion calculations, the corrected energy characterization is matched to the theoretical activity intensity trend of the disturbance source, thus determining the empirical relationship between the compensation coefficient and the transmission impedance index. For example, the compensation coefficient is equal to the transmission impedance index raised to the power of 0.5 multiplied by an empirical constant. Through the above compensation calculations, the amplitude of the original data feature patterns can be amplified. The degree of amplification is positively correlated with the transmission impedance index value; that is, the path with higher impedance and stronger attenuation will have its signal compensated more, thus theoretically restoring the original disturbance energy level at the disturbance source. Squaring the compensated amplitude, or squaring the compensated root mean square value, and then multiplying it by a time factor proportional to the pattern duration, yields the corrected energy characterization of the disturbance source within that time period. This energy characterization is a dimensionless or dimensionless value reflecting the magnitude of energy released by the disturbance source. By performing the above correction on all identified data feature patterns according to the disturbance source type, the corrected energy characterization sequences over time for disturbances caused by overlying coal seam residual structures, inter-layer rock pillars, and mining disturbances within the same layer are obtained.

[0033] The analysis includes calculating the real-time proportion of corrected energy representations of overlying coal seam residual structural disturbances, inter-layer rock pillar disturbances, and current mining disturbances in the total energy at long-period, medium-period, and short-period time scales, respectively, and statistically analyzing the fluctuation amplitude and rate of change of these real-time proportions at each time scale. The long-period time scale is determined based on the total interval from the end of mining in the overlying coal seam to the current mining time in the target lower working face. Specifically, the date of the end of mining in the overlying coal seam is verified from its historical mining record, and the total number of days elapsed from that date to the current analysis date is calculated. This total number of days is used as the baseline value for the long-period time scale. The actual time window length used for analysis can be between one-half and one-quarter of this baseline value; for example, if the total interval is 300 days, the long-period analysis window length can be set to 100 days. The medium-period time scale is determined based on the stress redistribution period of the inter-layer rock pillars. The stress redistribution cycle refers to the time required for the internal stress of an interlayer rock column to adjust from a non-equilibrium state to a new relative equilibrium state after being affected by mining. This cycle can be estimated through numerical simulation or statistically analyzed by monitoring the time it takes for the roadway deformation rate to stabilize. For example, based on historical monitoring data, the stage of severe roadway deformation usually lasts five to fifteen days, followed by a period of gradual deformation. Therefore, the medium-cycle time scale is set to ten days, and the corresponding analysis window length can be set to five to ten days. The short-cycle time scale is determined based on the cyclic propulsion cycle of the hydraulic support in the target lower working face. The cyclic propulsion cycle refers to the average time required for the hydraulic support to complete one complete cycle of lowering, moving, raising, and supporting. This can be obtained from the working face production log. For example, if one cycle is completed on average every twenty minutes, the short-cycle time scale can be set to eight hours per production shift, and the analysis window length can be set to two to four hours. After determining each time scale and its analysis window length, real-time percentage calculations are performed. For any given analysis time, on a long-period timescale, a time period of backtracking from the current time is taken, spanning a long-period window. The energy characteristics of the accumulated disturbances from the overlying coal seam's residual structure, after correction, are summed within this time period to obtain the long-period accumulated overlying energy value. Similarly, the long-period accumulated energy values ​​of inter-layer rock pillars and the long-period accumulated energy values ​​of mining within the current layer are obtained. These three accumulated values ​​are summed to obtain the total long-period accumulated energy value. Then, the percentage of the long-period accumulated energy values ​​of the disturbances from the overlying coal seam's residual structure, inter-layer rock pillars, and mining within the current layer relative to the total long-period accumulated energy value is calculated. This represents their real-time proportion on the long-period timescale. The calculation of the real-time proportion on medium-period and short-period timescales uses the same logic, except that the backtracking window length corresponds to the medium-period window length and the short-period window length, respectively. Subsequently, the fluctuation amplitude and rate of change of the real-time proportion at each timescale are statistically analyzed. The fluctuation amplitude is calculated by taking the difference between the maximum and minimum real-time proportion of a certain disturbance source within a statistical period, such as the most recent ten analysis window periods, as its fluctuation amplitude within that statistical period.The statistical method for the rate of change is to calculate the difference between the real-time proportion of a certain disturbance source between two consecutive analysis times, and then divide it by the time interval between the two analysis times to obtain the instantaneous rate of change. Then, the average of the absolute values ​​of these instantaneous rates of change within a statistical period is calculated as the average rate of change within that statistical period.

[0034] The determination process includes: when the corrected energy representation percentage of a disturbance source consistently exceeds a set threshold at the corresponding time scale, but the fluctuation amplitude is below the set threshold, the disturbance source is determined to be in a dominant state. When rapid fluctuations in the corrected energy representation percentages of different disturbance sources are detected, exceeding the set threshold, a state transition risk is determined. The set thresholds include a dominance percentage threshold, a dominance fluctuation amplitude threshold, and a state transition rate threshold. The dominance percentage threshold is used to determine whether the relative energy contribution of a disturbance source is large enough to become a dominant factor. This threshold is obtained based on historical data analysis, statistically analyzing the energy percentage distribution of a disturbance source at the corresponding time scale during previously recognized periods of single-source dominance. The lower quartile of this distribution is taken as the dominance percentage threshold. For example, by analyzing the three past periods of dominance by overlying legacy structure disturbances, it was found that the long-period energy percentage was not less than 65%, so the dominance percentage threshold at the long-period time scale can be set to 60%. The dominance fluctuation amplitude threshold is used to determine the stability of the dominant state. The threshold is also set based on the proportion fluctuation characteristics of the historical stable dominant phase. For example, the standard deviation of the proportion fluctuation in the above stable phase is calculated, and the dominant fluctuation amplitude threshold is set to twice the standard deviation. If the fluctuation amplitude is lower than this threshold, the state is considered stable. The state transition rate threshold is used to identify the risk of rapid shift in the dominant position of different disturbance sources. This threshold is determined by analyzing the abrupt change points of the energy proportion change rate of different disturbance sources in historical data before a significant mineral pressure manifestation event. For example, the maximum value of the proportion change rate in the 24 hours before the occurrence of multiple risk events is statistically analyzed, and the average value is taken as a reference for the state transition rate threshold, for example, set to a proportion change of 5% per hour. The specific application of the judgment logic is: on a long-term time scale, if the corrected energy characterization proportion of the overlying coal seam residual structure disturbance exceeds its dominant proportion threshold of 60% for more than 5 consecutive long-term analysis windows (i.e., continuously meeting the time condition), and its maximum fluctuation amplitude within these 5 windows is lower than the dominant fluctuation amplitude threshold of 8%, then it is determined that the overlying coal seam residual structure disturbance is currently in a long-term dominant state. At a medium-period timescale, if the corrected energy characterization ratio of inter-layer rock column disturbance is above 50% of its threshold for three consecutive analysis windows and the fluctuation amplitude is below 10% of the threshold, then the inter-layer rock column disturbance is determined to be in a medium-period dominant state. At a short-period timescale, if the corrected energy characterization ratio of mining disturbance in this layer is above 70% of its threshold for three consecutive analysis windows and the fluctuation amplitude is below 15% of the threshold, then the mining disturbance in this layer is determined to be in a short-period dominant state. The determination of state transition risk involves real-time monitoring of the rate of change of the corrected energy characterization ratio of any two disturbance sources at the same timescale.For example, on a short-period scale, if the rate of change of the proportion of mining disturbance in the current layer exceeds positive 5% (rapid growth) for two consecutive analytical intervals, while the rate of change of the proportion of inter-layer rock pillar disturbance exceeds negative 5% (rapid decline), and the absolute values ​​of both rates of change exceed the state transition rate threshold of 5% per hour, then it is determined that there is a risk of a shift from inter-layer rock pillar disturbance dominance to current layer mining disturbance dominance. This risk assessment is an early warning, indicating that the dominant structure of the disturbance may be undergoing dynamic changes.

[0035] S6. Based on the dominant state and state transition risk, dynamically adjust the current support strategy and equipment operating parameters for the target lower working face, specifically as follows: When it is determined that the disturbance from the overlying coal seam's residual structure is in a dominant state, the preset value of the support strength for the periodic pressure on the roof of the target lower working face is reduced while maintaining the existing advance speed of the hydraulic supports. The preset value of the support strength for the periodic pressure on the roof of the target lower working face is a target resistance value pre-set in the working face mining operation procedures to guide the hydraulic supports during the expected pressure period. This preset value is determined comprehensively based on the roof lithology, mining height, and mine pressure observation data of similar working faces. For example, the initial preset value may be set at 85% of the rated working resistance. The specific operation of reducing this preset support strength value is to calculate an adjustment coefficient based on the degree to which the corrected energy characterization remains higher than the set threshold under the dominant state of the overlying coal seam's residual structure disturbance. The adjustment coefficient is determined based on the statistical analysis of the ratio of the actual monitored roof pressure intensity to the standard pressure intensity calculated based on the theory of intact basic roof and rock beams during the dominant period of the overlying residual structure disturbance in historical data. For example, if the average ratio is 0.9 after analyzing three historical periods, the adjustment coefficient is set to 0.9. The new preset support strength is equal to the original preset value multiplied by this adjustment coefficient; for example, the original preset value of 10200 kN multiplied by 0.9 equals 9180 kN. This adjustment is achieved by inputting the new preset parameters through the human-machine interface of the working face hydraulic support electro-hydraulic control system. Maintaining the existing advance speed of the hydraulic support means not changing the reference speed value of the hydraulic support following the coal cutting machine, which is currently determined by the working face production shift plan. This reference speed is usually measured in meters per minute; for example, the current advance speed is 3 meters per minute. Maintaining operation means not issuing instructions to the control system to modify this speed reference value; the hydraulic support continues to follow the machine-following movement operation according to the procedure set in the previous shift.

[0036] When inter-layer rock pillar disturbance is determined to be dominant, the support density of the advance support section of the roadway is increased and the average coal cutting speed of the mining machine is reduced. The advance support section refers to a section of the roadway located within a certain range in front of the working face in the target lower working face, which requires enhanced support due to mining disturbance. This range is typically 20 to 50 meters in front of the working face. Support density refers to the number of active support structures arranged per unit length within this roadway section, such as the number of individual hydraulic props per meter or the number of anchor bolt sets per square meter. Specific implementation methods for increasing support density include increasing the number of rows of individual hydraulic props or reducing the spacing between rows and props. For example, the original support design was a single row of props with a prop spacing of 1 meter and a row spacing of 1.5 meters; after increasing the density, it is changed to double rows of props with a prop spacing of 0.8 meters and a row spacing of 1 meter. The new support density value is determined based on the convergence rate and deformation of the roadway surface displacement under the dominant inter-layer rock pillar disturbance state. First, the rate of change of the current roadway surface displacement is calculated using monitoring data. Then, based on a pre-established "displacement rate - support density increment" comparison table established according to the engineering experience of this mining area, the percentage increase in support density required is determined. For example, if the displacement rate exceeds 1 mm per hour, the support density needs to be increased by 30%. Reducing the average coal cutting speed of the coal mining machine refers to reducing the travel speed of the coal mining machine cutting the coal wall per unit time. The initial value of the average coal cutting speed is set according to the coal wall hardness, coal mining machine power, and production capacity. For example, the initial speed is 8 meters per minute. The reduction operation is determined based on the intensity and trend of the energy characterization after the inter-layer rock pillar disturbance is corrected. Specifically, a speed adjustment coefficient is set, which is inversely proportional to the energy characterization value of the inter-layer rock pillar disturbance. For example, when the energy characterization value exceeds a certain threshold, the adjustment coefficient is equal to the base coefficient divided by the normalized value of the energy characterization value. A new recommended coal cutting speed value is calculated. For example, the original speed of 8 meters per minute multiplied by the adjustment coefficient of 0.75 results in 6 meters per minute. The recommended value is set through the coal mining machine control system or can be achieved by the on-site operator manually adjusting the parameters of the coal mining machine's traction unit.

[0037] When it is determined that the disturbance in this layer of mining is dominant, the preset value of the support strength for the periodic pressure on the roof of the target lower working face is restored, and the advance speed of the hydraulic support is adjusted according to the standard procedures. Restoring the preset value of the support strength means resetting the preset value, which may have been lowered due to other dominant disturbances, back to the original standard value specified in the working face operation procedures for normal periodic pressure, or the benchmark value after correction based on the mining conditions of this layer. The determination of this benchmark value is strictly based on parameters such as the roof classification, mining height, and advance speed of the target working face, calculated through empirical formulas or mine pressure theory, and clearly listed in the procedures. For example, the standard value is 90% of the rated working resistance, i.e., 10800 kN. The restoration operation is achieved by manually or automatically modifying the preset parameter of the support strength to this standard value in the hydraulic support control system. Adjusting the advance speed of the hydraulic support according to the standard procedures refers to the provisions on advance speed and roof management in the "Target Lower Working Face Mining Operation Procedures". The adjustments are based on the identified roof activity cycle and pressure signs under the dominant mining disturbance conditions of this layer. Specifically, the timing of the pressure-inducing phase is determined by real-time monitoring of the support resistance change curve. When the support resistance continues to rise and exceeds the pressure warning threshold, the hydraulic support advance speed is appropriately reduced during the pressure-inducing period, for example, from the normal 3 meters per minute to 2 meters per minute, in accordance with the regulations, to slow down the roof exposure and subsidence. Once the pressure-inducing period ends and the resistance returns to the normal range, the advance speed is restored to the normal value according to the regulations. The speed adjustment is accomplished by modifying the advance speed parameters in the electro-hydraulic control system linked to the coal mining machine.

[0038] When a state transition risk is identified, an enhanced monitoring mode is activated while the current support strategy and equipment operating parameters are temporarily maintained. Activating the enhanced monitoring mode means temporarily increasing the frequency, density, or items of data collection on top of existing routine monitoring. Specific measures include shortening the sampling interval of key sensors, for example, adjusting the sampling interval for support resistance and roof / floor convergence from 30 seconds to 10 seconds; increasing the monitoring frequency of surface displacement in specific roadway areas, for example, changing from once per shift to once per hour; and temporarily adding several portable mine pressure monitoring instruments to areas indicated by risk warnings, such as adding borehole stress gauges in the middle of the working face or at points of abnormal roadway deformation. The enhanced monitoring mode must last for at least one typical cycle that could complete a state transition, for example, 24 hours or until the state transition risk assessment conditions are lifted. Temporarily maintaining the current support strategy and equipment operating parameters means that during enhanced monitoring, no adjustment instructions based on any single dominant state assessment are executed, and all support parameters and equipment speeds remain at the settings established before the risk assessment. For example, if the pre-set support strength is 9000 kN and the advance speed is 3 meters per minute, these parameters will remain unchanged during the risk monitoring period. This strategy aims to avoid insufficient support or production disruptions that may result from hasty parameter adjustments when the dominant pattern of the disturbance source is not yet clear and is undergoing rapid dynamic changes. Maintenance is implemented through system logic; when a risk flag is triggered, the system automatically locks the current parameter settings, suspends the execution of other automatic adjustment commands, records only data, and issues an alarm to prompt manual attention.

[0039] Example 2: Figure 2 A schematic diagram of the intelligent analysis system for mine pressure data in close-range multi-coal seam joint mining according to the present invention is provided. The intelligent analysis system for mine pressure data in close-range multi-coal seam joint mining includes the following modules: The data acquisition module is used to acquire multi-source mine pressure monitoring data of the target lower working face during the mining process; The sequence generation module is used to preprocess multi-source mine pressure monitoring data to obtain standardized time-series data sequences; The pattern recognition module is used to perform feature analysis on standardized time-series data sequences and identify data feature patterns that characterize disturbances of the overlying coal seam's residual structure, inter-layer rock pillars, and mining disturbances within the same layer. The indicator determination module is used to determine the transmission impedance index, which reflects the attenuation characteristics of the rock strata to the disturbance transmission between different disturbance sources and monitoring points, based on the characteristic patterns of each data and the geological lithology parameters of the target area. The risk assessment module is used to perform amplitude-frequency correction on each data characteristic mode based on the transmission impedance index to obtain the corrected energy characterization of each disturbance source. At different time scales, it analyzes the dynamic change trend and stability of the proportion of the corrected energy characterization of each disturbance source to determine the dominant state and state transition risk of disturbances from the overlying coal seam residual structure, interlayer rock pillars, and mining disturbances in the current layer in the multi-source mine pressure monitoring data. The parameter adjustment module is used to dynamically adjust the current support strategy and equipment operating parameters for the target lower working face based on the dominant state and state transition risk.

[0040] All calculations involved in the embodiments are dimensionless numerical calculations, and the preset parameters and thresholds in the calculations are set by those skilled in the art according to the actual situation.

[0041] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0042] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and inventive constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0043] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0044] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or modules may be electrical, mechanical, or other forms.

[0045] 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.

[0046] In conclusion, the above description is only a preferred 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 method for intelligent analysis of mine pressure data in close-range multi-coal seam joint mining, characterized in that, Includes the following steps: S1. Obtain multi-source mine pressure monitoring data of the target lower working face during the mining process; S2. Preprocess the multi-source mine pressure monitoring data to obtain a standardized time-series data sequence; S3. Perform feature analysis on the standardized time series data to identify data feature patterns that characterize disturbances of the overlying coal seam's residual structure, inter-layer rock pillars, and mining disturbances within the same layer. S4. Based on the characteristic patterns of each data and combined with the geological and lithological parameters of the target area, determine the transmission impedance index that reflects the attenuation characteristics of the rock strata to the disturbance transmission between different disturbance sources and monitoring points. S5. Based on the transmission impedance index, the amplitude and frequency of each data characteristic mode are corrected to obtain the corrected energy characterization of each disturbance source. At different time scales, the dynamic change trend and stability of the proportion of the corrected energy characterization of each disturbance source are analyzed to determine the dominant state and state transition risk of the overlying coal seam residual structure disturbance, interlayer rock column disturbance and mining disturbance in this layer in the multi-source mine pressure monitoring data. S6. Based on the dominant state and state transition risk, dynamically adjust the current support strategy and equipment operating parameters for the target lower working face.

2. The intelligent analysis method for mine pressure data in close-range multi-coal seam joint mining according to claim 1, characterized in that, S1 includes: Multi-source mine pressure monitoring data includes time-series data of support resistance of hydraulic supports in the target lower working face collected by pressure sensors, time-series data of the convergence of top and bottom plates between the top and bottom plates in the target lower working face collected by displacement sensors, and time-series data of roadway surface displacement of adjacent roadways in the target lower working face collected by convergence measuring points.

3. The intelligent analysis method for mine pressure data in close-range multi-coal seam joint mining according to claim 1, characterized in that, S2 include: Outlier removal and interpolation repair were performed on the time series data of support resistance, roof and floor approach, and roadway surface displacement, respectively. The time-series data of the repaired support resistance, the time-series data of the roof and floor convergence, and the time-series data of the roadway surface displacement are aligned with the timestamps of the unified sampling interval. The time-series data of support resistance, roof and floor approach, and roadway surface displacement after timestamp alignment are normalized to obtain a standardized time-series data sequence.

4. The intelligent analysis method for mine pressure data in close-range multi-coal seam joint mining according to claim 1, characterized in that, S3 includes: Time-domain statistical feature extraction and frequency-domain spectral feature extraction are performed on standardized time-series data sequences; Based on the extracted time-domain statistical features and frequency-domain spectral features, combined with the long-period characteristics of the overlying coal seam residual structure disturbance, the medium-period and cooperative deformation characteristics of the inter-layer rock pillar disturbance, and the short-period and strong dynamic characteristics of the mining disturbance of the current layer, data feature patterns that characterize the overlying coal seam residual structure disturbance, the inter-layer rock pillar disturbance, and the mining disturbance of the current layer are separated and identified from the standardized time-series data sequence.

5. The intelligent analysis method for mine pressure data in close-range multi-coal seam joint mining according to claim 1, characterized in that, S4 include: Geological and lithological parameters include the thickness, elastic modulus, and Poisson's ratio of each stratum between the overlying mined coal seam and the target lower working face; Determining the transmission impedance index involves calculating the equivalent stiffness and damping coefficient of the rock structure between different disturbance sources and monitoring points based on the disturbance transmission path and geological lithology parameters corresponding to each data characteristic pattern, and determining the transmission impedance index that reflects the corresponding disturbance transmission attenuation characteristics based on the equivalent stiffness and damping coefficient.

6. The intelligent analysis method for mine pressure data in close-range multi-coal seam joint mining according to claim 5, characterized in that, Geological lithology parameters are obtained by analyzing geological exploration reports and borehole columnar sections of the target area and correlated with the spatial location of the target underlying working face.

7. The intelligent analysis method for mine pressure data in close-range multi-coal seam joint mining according to claim 1, characterized in that, S5 include: The amplitude and frequency response of the data characteristic mode characterizing the disturbance source are compensated by using the transmission impedance index of the corresponding disturbance source, so as to obtain the corrected energy characterization of each disturbance source after eliminating the influence of transmission path attenuation. The analysis includes calculating the real-time proportion of the corrected energy representations of the overlying coal seam residual structure disturbance, inter-layer rock pillar disturbance, and mining disturbance in the current layer in the total energy at long-period, medium-period, and short-period time scales, and statistically analyzing the fluctuation amplitude and rate of change of the real-time proportions at each time scale. The determination includes determining that when the corrected energy representation ratio of a certain disturbance source is consistently higher than a set threshold and the fluctuation amplitude is lower than a set threshold at the corresponding time scale, the disturbance source is determined to be in a dominant state. When the corrected energy representation ratios of different disturbance sources are detected to change rapidly with one increasing and the rate of change exceeds a set threshold, a state transition risk is determined to exist.

8. The intelligent analysis method for mine pressure data in close-range multi-coal seam joint mining according to claim 7, characterized in that, The long-cycle timescale is determined based on the total interval from the end of mining of the overlying coal seam to the current mining time of the target lower working face. The medium-cycle timescale is determined based on the stress redistribution cycle of the interlayer rock pillar. The short-cycle timescale is determined based on the cyclic propulsion cycle of the hydraulic support of the target lower working face.

9. The intelligent analysis method for mine pressure data in close-range multi-coal seam joint mining according to claim 1, characterized in that, S6 include: When it is determined that the disturbance of the overlying coal seam's residual structure is in a dominant state, the preset value of the support strength for the periodic pressure on the roof of the target lower working face is reduced and the existing advance speed of the hydraulic support is maintained. When rock pillar disturbance is dominant during the judgment period, increase the support density of the advance support section of the roadway and reduce the average coal cutting speed of the coal mining machine. When it is determined that the disturbance in this layer of mining is in a dominant state, restore the preset value of the support strength for the periodic pressure on the roof of the target lower working face and adjust the advance speed of the hydraulic support according to the standard procedure. When a state transition risk is identified, an enhanced monitoring mode is activated and the current support strategy and equipment operating parameters are temporarily maintained unchanged.

10. A smart analysis system for mine pressure data in close-range multi-coal-seam joint mining, used to implement the smart analysis method for mine pressure data in close-range multi-coal-seam joint mining as described in any one of claims 1-9, characterized in that, Includes the following modules: The data acquisition module is used to acquire multi-source mine pressure monitoring data of the target lower working face during the mining process; The sequence generation module is used to preprocess multi-source mine pressure monitoring data to obtain standardized time-series data sequences; The pattern recognition module is used to perform feature analysis on standardized time-series data sequences and identify data feature patterns that characterize disturbances of the overlying coal seam's residual structure, inter-layer rock pillars, and mining disturbances within the same layer. The indicator determination module is used to determine the transmission impedance index, which reflects the attenuation characteristics of the rock strata to the disturbance transmission between different disturbance sources and monitoring points, based on the characteristic patterns of each data and the geological lithology parameters of the target area. The risk assessment module is used to perform amplitude-frequency correction on each data characteristic mode based on the transmission impedance index to obtain the corrected energy characterization of each disturbance source. At different time scales, it analyzes the dynamic change trend and stability of the proportion of the corrected energy characterization of each disturbance source to determine the dominant state and state transition risk of disturbances from the overlying coal seam residual structure, interlayer rock pillars, and mining disturbances in the current layer in the multi-source mine pressure monitoring data. The parameter adjustment module is used to dynamically adjust the current support strategy and equipment operating parameters for the target lower working face based on the dominant state and state transition risk.

Images