Roof separation online monitoring method

Abstract

The invention discloses a roof separation online monitoring method, and relates to the technical field of roof separation monitoring. Comprising the following steps that a plurality of sensors are arranged at different layers of a mine roof, different types of sensor arrangement schemes are selected according to geological conditions so as to comprehensively collect rock stratum change data, wide coverage and comprehensive data collection are ensured, and through multi-layer sensor arrangement, dynamic sampling frequency adjustment and multi-dimensional data fusion analysis, the rock stratum change data can be obtained. The system improves the accuracy of roof separation monitoring, realizes self-adaptive early warning for different geological environments in combination with an environment feature model, and reduces the risk of false alarm and missing alarm. The system can dynamically adjust the sampling frequency, capture the key change process, and continuously optimize the early warning parameters to avoid the problem of'monitoring blind areas' or strategy outdated, thereby improving the real-time performance, flexibility and stability of mine monitoring, and remarkably improving the mine safety management efficiency and intelligent level.

Description

Technical Field

[0001] The present invention relates to the technical field of roof separation monitoring, and in particular to an online monitoring method for roof separation. Background Art

[0002] Online roof delamination monitoring is a technical approach that provides real-time monitoring of delamination in the roof rock strata of mines or underground tunnels during mining or support operations. Delamination occurs when the adhesion between rock strata weakens or breaks due to stress changes, geological structures, or mining activities, resulting in gaps or separations. The online monitoring system uses sensors, displacement meters, and other equipment to continuously collect information on the displacement, pressure, and delamination of the roof rock strata. This data is transmitted in real time to a monitoring platform to analyze roof stability and provide timely warnings of potential roof collapse risks. This technology is widely used in underground mining sites such as coal and metal mines to improve mine safety and operational efficiency.

[0003] The existing technology has the following deficiencies:

[0004] In existing online monitoring of roof delamination, inaccurate interpretation of monitoring data is an easily overlooked issue that can have serious consequences. In complex geological environments, irregularities such as crack expansion and interlayer slip can occur when roof rock strata are subjected to stress changes. These changes are nonlinear and random. If the monitoring system analyzes data according to a fixed pattern, it may not be able to identify early signals of abnormal changes. For example, in soft rock or multi-fault strata, tiny crack expansion may be misinterpreted as normal changes, when in fact it is a precursor to roof instability. If the system ignores these signals, it may miss critical early warning opportunities, leading to sudden roof collapse and a major safety accident in the mine. Therefore, it is particularly important to improve the accuracy of data interpretation, especially to conduct targeted analysis of the changing patterns of different geological conditions.

[0005] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not form the prior art that is already known to a person of ordinary skill in the art. Summary of the Invention

[0006] The purpose of the present invention is to provide an online monitoring method for roof abscission. Through multi-level sensor arrangement, dynamic sampling frequency adjustment and multi-dimensional data fusion analysis, the system improves the accuracy of roof abscission monitoring. Combined with the environmental characteristic model, it realizes adaptive early warning for different geological environments, reducing the risk of false alarms and missed reports. The system can dynamically adjust the sampling frequency to capture key change processes, and by continuously optimizing the early warning parameters, it avoids "monitoring blind spots" or strategy outdated problems, improves the real-time, flexibility and stability of mine monitoring, and significantly improves the efficiency and intelligence level of mine safety management to solve the problems in the above-mentioned background technology.

[0007] In order to achieve the above object, the present invention provides the following technical solution: a method for online monitoring of roof separation, comprising the following steps:

[0008] Multiple sensors are deployed at different levels of the mine roof, and different types of sensor deployment schemes are selected based on geological conditions to comprehensively collect rock layer change data, ensuring wide coverage and comprehensive data collection;

[0009] Set the dynamic sampling frequency of the sensor and automatically adjust the sampling interval according to the severity of the roof rock changes. When a sudden displacement change is detected, the sampling frequency is automatically increased to ensure that key changes are captured in real time.

[0010] Pre-process the collected raw monitoring data, combine it with the mine geological characteristic parameters, incorporate the differences in geological environment into the monitoring analysis, and generate an environmental characteristic model;

[0011] Perform multi-dimensional data fusion on displacement, pressure, and geological parameters, match different types of data in time and space, generate overall rock formation change trend maps, and identify potential separation locations and change patterns;

[0012] A nonlinear trend recognition algorithm is used to analyze the multi-dimensional data fusion results. By detecting the rate of change, acceleration, and fluctuation amplitude of the data, early abnormal changes in the rock formation are identified, and normal changes are distinguished from potential roof instability risks.

[0013] Based on the collected historical data and real-time data, the early warning model is continuously optimized using adaptive learning algorithms, so that it can automatically adjust parameters according to new data patterns, improve its adaptability to different geological environments, and thus enhance the accuracy and real-time nature of the early warning.

[0014] Preferably, multiple sensors are arranged at different levels of the mine roof, and different types of sensor arrangement schemes are selected according to geological conditions to comprehensively collect rock layer change data. The specific steps to ensure wide coverage and comprehensive data collection are as follows:

[0015] By comprehensively analyzing the mine's geological structure and identifying key monitoring areas, we ensure the accuracy and targeting of sensor placement;

[0016] Select appropriate sensor types and layout schemes according to different geological environments to cover key areas of displacement and stress changes;

[0017] Three-dimensional sensors are installed at different levels of the mine roof to form a multi-level monitoring network to ensure comprehensive and reliable data collection;

[0018] By optimizing the sensor layout and introducing a redundant data acquisition mechanism, the stability of the monitoring system and the continuity of data acquisition are improved.

[0019] Preferably, the dynamic sampling frequency of the sensor is set to automatically adjust the sampling interval according to the severity of the roof rock change. When a sudden change in displacement is detected, the sampling frequency is automatically increased to ensure that the key change process is captured in real time. The specific steps are as follows:

[0020] Set the initial sampling frequency based on geological conditions and historical data to ensure that basic data collection covers early trends in roof changes;

[0021] Identify abnormal change signals in data through real-time fluctuation monitoring mechanism to improve the accuracy of identifying key changes;

[0022] When an abnormal signal is detected, the sampling frequency is automatically increased to ensure that the key change process is fully captured;

[0023] The sampling strategy is dynamically adjusted through a multi-layer feedback mechanism to ensure flexible and adaptive sampling frequency in different changing scenarios.

[0024] Preferably, the collected raw monitoring data is preprocessed, and the differences in geological environment are incorporated into the monitoring analysis by combining the mine geological characteristic parameters. The specific steps for generating the environmental characteristic model are as follows:

[0025] Collect mine geological characteristic parameters and establish a geological information database to provide comprehensive environmental basic information for subsequent data analysis;

[0026] Correlate and match monitoring data with geological characteristic parameters to improve the ability to interpret data fluctuations and the accuracy of analysis;

[0027] Ensure the authenticity and reliability of monitoring data and reduce interference factors through data cleaning, denoising and outlier identification;

[0028] Generate environmental characteristic models, dynamically adjust sampling and early warning strategies based on geological differences, and improve the accuracy and adaptability of the monitoring system.

[0029] Preferably, the specific steps of dynamically matching displacement, pressure and geological parameters in the time dimension to generate a change trend sequence and identify the time pattern of rock formation changes and potential risk signals are as follows:

[0030] Synchronize timestamps and establish a unified time series to ensure that different types of data can be compared and analyzed on the same timeline;

[0031] By analyzing the changing trends of time series data, we can identify the fluctuation patterns of rock displacement and pressure and make a preliminary assessment of the risk of separation.

[0032] Establish a time correlation model to analyze the mutual influence and temporal relationship between displacement, pressure and geological parameters;

[0033] Dynamically update trend sequences and adjust monitoring strategies in real time to ensure rapid response to changes in rock formations.

[0034] Preferably, the specific steps of integrating multi-layer bit data in the spatial dimension and generating an overall rock formation change trend map using three-dimensional modeling to intuitively identify key change areas and potential separation locations are as follows:

[0035] Spatial matching of sensor data from different layers generates a multi-layered map of mine roof changes, improving the comprehensiveness of data analysis.

[0036] Use spatial matching data to generate a three-dimensional model to visually present rock formation changes and identify key change areas;

[0037] By integrating time and space data, an overall change trend diagram is generated to show the change pattern of the rock layer and the potential separation location;

[0038] Mark high-risk areas based on the changing trend chart and dynamically adjust warning parameters to improve the system's warning accuracy and flexibility.

[0039] Preferably, a nonlinear trend recognition algorithm is used to analyze the multidimensional data fusion results. By detecting the rate of change, acceleration, and fluctuation amplitude of the data, the early abnormal change signals of the rock formation are identified, and the specific steps for distinguishing normal changes from potential roof instability risks are as follows:

[0040] When analyzing the results of multidimensional data fusion, we first construct a multidimensional change rate matrix and calculate the time derivative of the collected displacement, pressure, and geological parameter data to obtain the change rate of each type of data. The change rate calculation formula is as follows:

[0041] Where M rate (i, t) is the change rate matrix element of the i-th sensor at time t, D i (t) is the original monitoring data of the i-th sensor at time t, Δt is the time interval, D i (t-Δt) is the original monitoring data of the i-th sensor at time t-Δt, that is, the original monitoring data of the previous moment;

[0042] Based on the change rate matrix, the change acceleration matrix is ​​further calculated to measure the severity and mutation of data changes. The calculation expression of the change acceleration is as follows:

[0043] Where M accel(i, t) is the acceleration matrix element of the i-th sensor at time t, M rate (i, t-Δt) is the change rate matrix element of the i-th sensor at time t-Δt, that is, the change rate matrix element at the previous moment.

[0044] Preferably, based on the rate of change and acceleration matrix, the fluctuation amplitude analysis parameter is calculated to identify abnormal fluctuations in the data. The fluctuation amplitude calculation formula is as follows:

[0045] Where A wave (i) is the fluctuation amplitude analysis parameter of the i-th sensor, is the average rate of change of the i-th sensor, T is the total number of time points;

[0046] Binding change rate M rate (i, t), acceleration M accel (i, t) and the fluctuation amplitude parameter A wave (i) Calculate the nonlinear risk assessment index to ultimately distinguish between normal changes and potential roof instability risks. The calculation formula is as follows:

[0047] R risk (i) = α·M rate (i, t)+β·M accel (i, t)+γ·A wave (i), where R risk (i) is the nonlinear risk assessment index of the i-th sensor, α, β and γ are weight parameters, and α is used to adjust the change rate M rate (i, t) has an impact on the overall risk assessment, and β is used to adjust the acceleration M accel The influence of (i, t) parameters, γ is used to adjust the fluctuation amplitude parameter A wave (i) Impact.

[0048] Preferably, based on the collected historical data and real-time data, the early warning model is continuously optimized using an adaptive learning algorithm, so that it can automatically adjust parameters according to new data patterns, improve its adaptability to different geological environments, and thus enhance the accuracy and real-time nature of the early warning. The specific steps are as follows:

[0049] First, a basic early warning model is constructed based on the collected historical data and real-time data, and key parameters are defined, including the displacement change rate, pressure change rate, and environmental characteristic parameter weights. In order to comprehensively consider the impact of multidimensional data, a data weight matrix is ​​used to represent the contribution of each parameter to the early warning model. The formula is as follows:

[0050] Where M0 is the basic warning weight matrix, which represents the weight distribution of each parameter of the initial model, w1, w2, ..., w9 are the weights of different data types at different time points;

[0051] During real-time monitoring, the abnormal deviation rate between the actual data and the model prediction value is calculated based on the current displacement change rate and pressure change rate. The abnormal deviation rate calculation formula is as follows:

[0052] Where D t is the abnormal deviation rate, V t is the real-time displacement change rate, is the displacement change rate predicted by the model, P t is the real-time pressure change rate, is the rate of change of pressure predicted by the model.

[0053] Preferably, according to the calculated abnormal deviation rate D t , dynamically adjust the weight values ​​of the basic weight matrix M0 to generate a new weight matrix. The generation formula is as follows: M t =M0+ω·D t I, where M t is the weight matrix after real-time optimization, ω is the learning rate, and I is the identity matrix;

[0054] According to the optimized weight matrix M t , recalculate the warning threshold to dynamically adjust the warning strategy for different areas. The calculation formula of the warning threshold is as follows:

[0055] Where, T t is the real-time warning threshold, M t [p] is the pth weight value in the optimized weight matrix, F p It is the influencing factor of geological characteristic parameters.

[0056] In the above technical solution, the technical effects and advantages provided by the present invention are:

[0057] Through multi-level sensor arrangement, dynamic sampling frequency adjustment and multi-dimensional data fusion analysis, the present invention effectively improves the interpretation accuracy of roof separation monitoring data and significantly reduces the risk of missed reports and false alarms. The environmental characteristic model is generated in combination with the geological characteristic parameters of the mine, so that the monitoring process can be adaptively adjusted according to different geological environments, thereby more accurately identifying abnormal signals such as rock displacement, stress changes and crack expansion. Especially in soft rock layers or areas with dense faults, the system can capture small but continuous change trends in advance and issue early warnings in time, avoiding the delayed warnings or misjudgments caused by fixed thresholds and single data source analysis in traditional monitoring systems. In addition, the system can automatically adjust the sampling frequency when an abnormal signal is detected to ensure that detailed data of key change processes are captured, thereby greatly improving the real-time and reliability of monitoring.

[0058] The present invention introduces an adaptive early warning mechanism of the environmental characteristic model, which enables the monitoring system to dynamically optimize the early warning parameters and sampling strategies according to the real-time changes in the mine, thereby achieving more flexible monitoring and early warning management. For high-risk areas such as soft rock formations, the system will automatically lower the early warning threshold and increase the sampling frequency, while in more stable areas, it can reduce unnecessary high-frequency monitoring and improve the efficiency of system resource utilization. At the same time, the adaptive early warning mechanism has the ability to continuously optimize. The system can continuously adjust the environmental characteristic model based on real-time feedback to enable it to maintain efficient operation for a long time and avoid the problem of "monitoring blind spots" or outdated strategies. This flexibility and continuous optimization capability not only reduces the false alarm rate and missed alarm rate, but also reduces the burden on mine management personnel and improves the safety, intelligence and stability of online monitoring of mine roof separation. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction to the drawings required for use in the embodiments will be given below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0060] Figure 1 The present invention is a flowchart of a method for online monitoring of roof separation. DETAILED DESCRIPTION

[0061] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, these example embodiments are provided so that the description of this disclosure will be thorough and complete and will fully convey the concepts of the example embodiments to those skilled in the art.

[0062] The present invention provides Figure 1 The method for online monitoring of roof separation shown includes the following steps:

[0063] Multiple sensors, including displacement sensors and pressure sensors, are deployed at different levels of the mine roof. Different sensor placement schemes are selected based on geological conditions to comprehensively collect data on stress, displacement, and crack changes in the rock formation, ensuring wide coverage and comprehensive data collection.

[0064] Multiple sensors, including displacement sensors and pressure sensors, are deployed at different levels of the mine roof. Different sensor placement schemes are selected based on geological conditions to comprehensively collect data on stress, displacement, and crack changes in the rock formation. The specific steps to ensure wide coverage and comprehensive data collection are as follows:

[0065] By comprehensively analyzing the mine's geological structure and identifying key monitoring areas, we ensure the accuracy and targeting of sensor placement;

[0066] Before deploying sensors, the geological structure of the mine roof must be thoroughly explored and analyzed to determine the rock strata type, thickness, fault distribution, joint development, and potential weak spots. Geological mapping, borehole exploration, and other methods are used to identify areas prone to delamination, slippage, and crack expansion. This step aims to precisely locate key monitoring areas, avoiding blind sensor placement and thus improving the relevance and effectiveness of monitoring. In particular, areas with multiple faults or soft rock formations should be prioritized for monitoring, as these areas are more susceptible to roof instability.

[0067] Select appropriate sensor types and layout schemes according to different geological environments to cover key areas of displacement and stress changes;

[0068] Select the appropriate sensor type based on the identified geological structure characteristics. For areas prone to large-scale displacement, such as soft rock layers and coal seam roofs, high-precision displacement sensors should be deployed first. For hard rock layers or areas with a higher risk of interlayer slippage, pressure sensors should be deployed to monitor stress changes. In addition, special sensors such as laser displacement meters and fiber optic sensors can be deployed to obtain more comprehensive monitoring data. The sensor layout plan should take into account the shape, depth, and spatial distribution of the mine tunnels to ensure that the sensor network has sufficient coverage and avoid monitoring blind spots.

[0069] Three-dimensional sensors are installed at different levels of the mine roof to form a multi-level monitoring network to ensure comprehensive and reliable data collection;

[0070] Sensors are installed in layers at different levels of the mine roof to form a three-dimensional monitoring network. For example, sensors can be placed on the tunnel's vault, sidewalls, major faults, and critical rock strata interfaces, covering different mechanical layers of the roof. The fixing method should be selected based on the geological stability of each layer. For example, expansion bolts may be used to secure sensors in relatively stable rock formations, while support devices may be used to protect sensors in slippery rock formations. Furthermore, sensor stability and data collection should be regularly checked to ensure long-term monitoring reliability.

[0071] Improve the stability of the monitoring system and the continuity of data acquisition by optimizing sensor layout and introducing redundant data collection mechanisms;

[0072] After sensor deployment is complete, the sensor layout should be further optimized based on the initial data collected. For areas experiencing significant data fluctuations or ineffective coverage, additional sensors should be added to increase monitoring density. Furthermore, a redundant data collection mechanism should be established, with backup sensors deployed at key monitoring points to ensure the system can still obtain data even if some sensors fail. The introduction of a redundant mechanism can effectively improve the stability of the monitoring system and prevent data interruptions or omissions due to equipment failures.

[0073] Set the dynamic sampling frequency of the sensor and automatically adjust the sampling interval according to the severity of the roof rock changes. When a sudden displacement change is detected, the sampling frequency is automatically increased to ensure that key changes are captured in real time.

[0074] The specific steps for setting the dynamic sampling frequency of the sensor, automatically adjusting the sampling interval according to the severity of the roof rock changes, and automatically increasing the sampling frequency when a sudden displacement change is detected to ensure that key changes are captured in real time are as follows:

[0075] Set the initial sampling frequency based on geological conditions and historical data to ensure that basic data collection covers early trends in roof changes;

[0076] After the sensors are deployed, the first step is to define sampling rules based on the mine's geological environment and historical data on roof changes, setting the initial sampling frequency for different situations. For example, in relatively stable hard rock areas, the initial sampling frequency can be set to once an hour; while in high-risk areas such as soft rock formations or fault intersections, the initial sampling frequency should be set to once every 10 minutes. When setting the initial frequency, it is important to comprehensively consider the rock formation type, the stability of the monitoring area, and the accuracy of the monitoring data to ensure that the basic data collected by the sensor network can fully cover early trends in roof changes.

[0077] Identify abnormal change signals in data through real-time fluctuation monitoring mechanism to improve the accuracy of identifying key changes;

[0078] Continuously collected monitoring data is analyzed in real time for displacement and stress changes in the roof rock formation, identifying abnormal signals of data fluctuation. For example, when the system detects a sudden change in displacement data from a sensor within a short period of time, the fluctuation monitoring mechanism is triggered, marking the data as "abnormally changing" and recording the amplitude and trend of the fluctuation. This mechanism effectively avoids false alarms caused by small-scale data fluctuations while ensuring that critical change signals are not overlooked. The identification of abnormal change signals should be combined with the previously deployed sensor network and historical data models to ensure accurate identification.

[0079] When an abnormal signal is detected, the sampling frequency is automatically increased to ensure that the key change process is fully captured;

[0080] Once the system identifies an abnormal change signal, it should immediately adjust the sensor's sampling frequency. Especially for sensors in areas experiencing significant change, the sampling frequency should be increased to seconds or minutes to ensure that critical changes are captured in real time. For example, under normal circumstances, the sampling frequency for a certain area might be once every 10 minutes. However, if the system detects a sudden change in displacement, the sampling frequency could be adjusted to once per minute or higher. Furthermore, upper and lower limits can be set for the sensor's sampling interval to ensure that the adjusted frequency does not significantly impact system stability.

[0081] Dynamically adjust the sampling strategy through a multi-layer feedback mechanism to ensure flexible and adaptive sampling frequency under different changing scenarios;

[0082] After adjusting the sampling frequency, the system must also conduct real-time feedback analysis on the collected high-frequency data, dynamically adjusting subsequent sampling strategies based on the roof's changing trends. For example, when high-frequency data indicates that roof changes are stabilizing, the system can gradually return to a lower sampling frequency to reduce equipment load. Conversely, if the trend continues to intensify, high-frequency sampling is maintained until the changes stabilize. Simultaneously, this multi-layered feedback mechanism should continuously learn from historical sampling data to optimize future dynamic sampling rules, ensuring that the system can adaptively adjust to different changing scenarios.

[0083] Pre-process the collected raw monitoring data and combine it with the mine geological characteristic parameters, including rock structure type, fault distribution and roof hardness, to incorporate the differences in geological environment into the monitoring analysis and generate an environmental characteristic model;

[0084] The collected raw monitoring data is pre-processed and combined with the mine geological characteristic parameters, including rock structure type, fault distribution, and roof hardness, to incorporate the differences in the geological environment into the monitoring analysis. The specific steps for generating the environmental characteristic model are as follows:

[0085] Collect mine geological characteristic parameters and establish a geological information database to provide comprehensive environmental basic information for subsequent data analysis;

[0086] Before preprocessing the raw monitoring data, key geological parameters of the mine must be collected. These include the rock layer structure, the distribution of faults and joints, rock hardness, and mining depth. This geological information is typically obtained through drilling sampling, geological mapping, and historical mine reports, and is categorized and entered into a geological information database. The mine is divided into multiple monitoring sub-areas based on different geological regions, and each area is labeled with geological characteristics. This process ensures that geological differences are taken into account in subsequent data analysis, avoiding biased data interpretation caused by ignoring environmental differences.

[0087] Correlate and match monitoring data with geological characteristic parameters to improve the ability to interpret data fluctuations and the accuracy of analysis;

[0088] Dynamically collected sensor data is correlated and matched with the geological characteristic parameters of the corresponding area. For example, displacement data collected by displacement sensors is combined with parameters such as rock layer structure and fault distribution to determine whether data fluctuations are related to geological characteristics. In areas with low roof hardness, detecting large displacement changes may indicate crack expansion or interlayer slip; in hard rock areas, it is more likely to be short-term elastic deformation caused by stress concentration. This correlation and matching process enables data analysis to be combined with geological context for comprehensive judgment, thereby improving the ability to interpret abnormal changes.

[0089] Ensure the authenticity and reliability of monitoring data and reduce interference factors through data cleaning, denoising and outlier identification;

[0090] The raw monitoring data collected by the sensors undergoes preprocessing, including data cleaning, denoising, and outlier identification. Due to the complex mine environment, the monitoring data may contain interfering signals, such as noise generated by equipment vibration, air flow, or human interaction. Therefore, a data filtering algorithm is applied to filter out the noise and identify and remove outliers. For example, if a sensor exhibits extreme value changes within a short period of time and is not verified by other sensors, it can be considered an outlier and excluded from analysis. This preprocessing ensures the authenticity and reliability of the data and prevents erroneous data from interfering with subsequent analysis.

[0091] Generate environmental characteristic models, dynamically adjust sampling and early warning strategies based on geological differences, and improve the accuracy and adaptability of monitoring systems;

[0092] After data preprocessing, an environmental characteristic model is generated by combining geological characteristic parameters with cleaned data. This model incorporates rock formation variation patterns, historical data trends, and potential risk points for each monitored area, and is dynamically updated. The environmental characteristic model can be used to guide the monitoring system's dynamic sampling strategy and early warning threshold setting. For example, in soft rock areas, the environmental model can automatically lower the early warning threshold and increase the sampling frequency; in more stable areas, the model can appropriately reduce the sampling density to conserve system resources. This dynamic adjustment based on the environmental characteristic model makes the monitoring system more adaptable and accurate.

[0093] Perform multi-dimensional data fusion on displacement, pressure, and geological parameters, match different types of data in time and space, generate overall rock formation change trend maps, and identify potential separation locations and change patterns;

[0094] By dynamically matching displacement, pressure, and geological parameters in the time dimension to generate a trend sequence, the specific steps for identifying the temporal patterns of rock formation changes and potential risk signals are as follows:

[0095] Synchronize timestamps and establish a unified time series to ensure that different types of data can be compared and analyzed on the same timeline;

[0096] First, the collected displacement, pressure, and geological parameter data are time-stamped and synchronized to ensure that different types of data can be compared and analyzed at the same time point. Because the sampling frequency and data generation time of sensors may vary, the data needs to be time-aligned. The real-time displacement and pressure monitoring data is time-stamped and matched with the preprocessed geological parameter data to form a unified time series. This step ensures that all types of data can display the dynamic process of rock formation changes on the same timeline, thus avoiding data analysis errors caused by time deviations.

[0097] By analyzing the changing trends of time series data, we can identify the fluctuation patterns of rock displacement and pressure and make a preliminary assessment of the risk of separation.

[0098] Trend analysis is performed on synchronized time series data to identify patterns in displacement and pressure data. For example, data characteristics such as stress increases and decreases, displacement amplitudes, and frequency over a specific time period are observed. The underlying causes of these changes are analyzed in conjunction with geological parameters. A sudden change in displacement accompanied by a sudden increase in pressure within a specific time period may indicate crack expansion or interlayer slip in the roof strata. By analyzing this trend data, a timeline of rock formation changes can be generated, providing a basis for subsequent early warning and analysis.

[0099] Establish a time correlation model to analyze the mutual influence and temporal relationship between displacement, pressure and geological parameters;

[0100] Based on the temporal variations of displacement, pressure, and geological parameter data, a correlation model is established between different data types to analyze their temporal interactions. For example, this study examines whether pressure changes precede displacement changes, or whether there is a more pronounced time delay in soft rock formations. This temporal correlation model can help monitoring systems more accurately predict rock formation trends, thereby improving the timeliness and accuracy of early warnings.

[0101] Dynamically update trend sequences and adjust monitoring strategies in real time to ensure rapid response to rock formation changes;

[0102] As new monitoring data is continuously collected, the system should dynamically update time series trends and adjust monitoring strategies in real time based on data changes. For example, if pressure data consistently rises and displacement data exhibits a sudden change, the system can automatically increase sampling frequency and mark the area as a high-risk area for intensive monitoring. This dynamic update mechanism ensures that the monitoring system can respond to rock formation changes in real time, avoiding missing critical changes.

[0103] The specific steps for integrating multi-layer data in the spatial dimension and using 3D modeling to generate a map of overall rock formation change trends and visually identify key change areas and potential separation locations are as follows:

[0104] Spatial matching of sensor data from different layers generates a multi-layered map of mine roof changes, improving the comprehensiveness of data analysis.

[0105] Based on the sensor placement, displacement and pressure data from different layers are spatially matched to create a multi-layered map of mine roof changes. For example, data from sensors installed at the tunnel's vault, sidewalls, and fault intersections can be used through a spatial matching algorithm to generate a snapshot of changes in these areas at the same point in time. This spatial matching helps the system identify overall trends in rock formation changes, avoiding the limitations of single-point data analysis and enabling more accurate determination of potential delamination locations.

[0106] Use spatial matching data to generate a three-dimensional model to visually present rock formation changes and identify key change areas;

[0107] The matched spatial data is imported into a 3D modeling system to generate a 3D trend map of the mine roof. In the 3D model, displacement and pressure change data can be visualized through color and shape changes, making it easier to identify areas experiencing drastic changes. For example, if the displacement curve in a certain area shows a downward concave trend accompanied by stress concentration, it may indicate a potential delamination area. 3D modeling not only visually demonstrates rock formation changes but also helps analyze the depth and extent of these changes.

[0108] By integrating time and space data, an overall change trend diagram is generated to show the change pattern of the rock layer and the potential separation location;

[0109] Based on 3D modeling, a trend chart of overall mine roof changes is generated, integrating temporal and spatial change data into a single graph. This trend chart clearly demonstrates the dynamics of rock formations, including crack propagation directions, interlayer slip locations, and areas of stress concentration. Analysis of this trend chart can identify early signs of rock delamination, providing timely warnings and preventing roof collapse accidents.

[0110] Mark high-risk areas based on trend charts and dynamically adjust warning parameters to improve the system's warning accuracy and flexibility;

[0111] Based on the analysis of the overall trend graph, high-risk areas are marked and the warning parameters for those areas are dynamically adjusted. For example, in areas experiencing rapid crack expansion, the warning threshold can be lowered and the data sampling frequency increased; whereas in areas experiencing stable changes, the warning threshold can be appropriately raised and the sampling frequency reduced. This dynamic adjustment mechanism ensures a more accurate and flexible warning strategy, reducing the risk of false alarms and missed alerts.

[0112] A nonlinear trend recognition algorithm is used to analyze the multi-dimensional data fusion results. By detecting the rate of change, acceleration, and fluctuation amplitude of the data, early abnormal changes in the rock formation are identified, and normal changes are distinguished from potential roof instability risks.

[0113] A nonlinear trend recognition algorithm is used to analyze the multidimensional data fusion results. By detecting the rate of change, acceleration, and fluctuation amplitude of the data, early abnormal changes in the rock formation are identified, and the specific steps to distinguish normal changes from potential roof instability risks are as follows:

[0114] When analyzing the results of multidimensional data fusion, we first construct a multidimensional change rate matrix and calculate the time derivative of the collected displacement, pressure, and geological parameter data to obtain the change rate of each type of data. The change rate calculation formula is as follows:

[0115] Where M rate (i, t) is the change rate matrix element of the i-th sensor at time t, D i (t) is the original monitoring data of the i-th sensor at time t, Δt is the time interval, D i (t-Δt) is the original monitoring data of the i-th sensor at time t-Δt, that is, the original monitoring data of the previous moment;

[0116] Change rate matrix M rateIt can intuitively reflect the changing rate of the roof rock layer at different time points and provide basic data for subsequent acceleration and fluctuation amplitude calculations.

[0117] Based on the change rate matrix, the change acceleration matrix is ​​further calculated to measure the severity and mutation of data changes. The calculation expression of the change acceleration is as follows:

[0118] Where M accel (i, t) is the acceleration matrix element of the i-th sensor at time t, M rate (i, t-Δt) is the change rate matrix element of the i-th sensor at time t-Δt, that is, the change rate matrix element at the previous moment;

[0119] Acceleration matrix M accel It can be used to identify the severity of rock formation changes. When the acceleration value continues to rise and exceeds a certain threshold, it may indicate that crack expansion or interlayer slippage is occurring in the rock formation.

[0120] Based on the rate of change and acceleration matrix, the fluctuation amplitude analysis parameters are calculated to identify abnormal fluctuations in the data. The fluctuation amplitude calculation formula is as follows:

[0121] Where A wave (i) is the fluctuation amplitude analysis parameter of the i-th sensor, is the average rate of change of the i-th sensor, T is the total number of time points;

[0122] Fluctuation amplitude analysis parameter A wave It can identify abnormal fluctuations in the readings of different sensors. When the fluctuation amplitude exceeds the set threshold, it indicates that the rock formation in the area where the sensor is located is abnormal and requires special attention.

[0123] Binding change rate M rate (i, t), acceleration M accel (i, t) and the fluctuation amplitude parameter A wave (i) Calculate the nonlinear risk assessment index to ultimately distinguish between normal changes and potential roof instability risks. The calculation formula is as follows:

[0124] R risk (i) = α·M rate (i, t)+β·M accel (i, t)+γ·A wave (i), where R risk (i) is the nonlinear risk assessment index of the i-th sensor, α, β and γ are weight parameters, and α is used to adjust the change rate M rate(i, t) has an impact on the overall risk assessment, and β is used to adjust the acceleration M accel The influence of (i, t) parameters, γ is used to adjust the fluctuation amplitude parameter A wave (i) Impact.

[0125] Nonlinear risk assessment index R risk The level of the index directly reflects the degree of roof instability. When the index exceeds a certain warning threshold, the system will automatically issue a warning signal, indicating that the area may be at risk of roof instability.

[0126] Based on collected historical and real-time data, the early warning model is continuously optimized using adaptive learning algorithms, enabling it to automatically adjust parameters according to new data patterns, improving its adaptability to different geological environments, and thus enhancing the accuracy and real-time nature of early warnings;

[0127] Based on the collected historical data and real-time data, the early warning model is continuously optimized using adaptive learning algorithms, enabling it to automatically adjust parameters according to new data patterns, improving its adaptability to different geological environments, and thus enhancing the accuracy and real-time nature of early warnings. The specific steps are as follows:

[0128] First, a basic early warning model is constructed based on the collected historical data and real-time data, and key parameters are defined, including the displacement change rate, pressure change rate, and environmental characteristic parameter weights. In order to comprehensively consider the impact of multidimensional data, a data weight matrix is ​​used to represent the contribution of each parameter to the early warning model. The formula is as follows:

[0129] Where M0 is the basic warning weight matrix, which represents the weight distribution of each parameter of the initial model. w1, w2, ..., w9 are the weights of different data types (displacement, pressure, geological characteristics) at different time points. These weight values ​​are obtained through regression analysis of historical data and will be dynamically updated in subsequent steps.

[0130] This step establishes the basic weight structure of the early warning model. The values ​​of the weight matrix M0 represent the degree of influence of different data on the early warning, providing a basis for subsequent dynamic adjustments.

[0131] During real-time monitoring, the abnormal deviation rate between the actual data and the model prediction value is calculated based on the current displacement change rate and pressure change rate. The abnormal deviation rate calculation formula is as follows:

[0132] Where D t is the abnormal deviation rate, which is used to measure the abnormality of the current data. t is the real-time displacement change rate, is the displacement change rate predicted by the model, P t is the real-time pressure change rate, is the rate of pressure change predicted by the model;

[0133] By calculating the abnormal deviation rate D t , identify the abnormality level of real-time data and provide input values ​​for dynamic optimization of the model.

[0134] According to the calculated abnormal deviation rate D t , dynamically adjust the weight values ​​of the basic weight matrix M0 to generate a new weight matrix. The generation formula is as follows: M t =M0+ω·D t I, where M t is the weight matrix after real-time optimization, ω is the learning rate, which controls the amplitude of weight adjustment, and I is the identity matrix, which is used to maintain the consistency of the direction of weight adjustment;

[0135] The adjusted weight matrix M t It can more accurately reflect the changes in the current geological environment and enable the early warning model to dynamically adapt to new data patterns. By adjusting the weight matrix M t , so that the early warning model can adapt to changes in the geological environment in real time and improve the ability to identify abnormal situations.

[0136] According to the optimized weight matrix M t , recalculate the warning threshold to dynamically adjust the warning strategy for different areas. The calculation formula of the warning threshold is as follows:

[0137] Where, T t is the real-time warning threshold, M t [p] is the pth weight value in the optimized weight matrix, F p It is the influencing factor of geological characteristic parameters.

[0138] New warning threshold T t It can automatically adjust according to the geological characteristics and real-time monitoring data of different regions to ensure the accuracy and real-time performance of the early warning system. t , dynamically optimize the early warning strategy, so that the system can flexibly adjust the warning level and sampling frequency according to real-time data.

[0139] Specific implementation method 1: To accurately monitor changes in mine roof rock formations, multi-layered sensor deployment and dynamic sampling frequency adjustment are key steps. The core goal of this implementation method is to ensure that the sensor layout covers different geological layers within the mine, thereby obtaining comprehensive rock formation displacement, stress changes, and crack expansion data. Dynamically adjusting the sampling frequency enables real-time monitoring and early warning.

[0140] First, based on the mine's geological conditions, displacement sensors and pressure sensors need to be placed at different levels of the roof. When placing these sensors, key areas such as the tunnel's vault, sidewalls, and fault intersections should be considered, as these are often areas of stress concentration and prone to crack expansion or interlayer slip. Sensor placement must not only consider coverage but also redundant layout to prevent data loss due to individual sensor failures. Furthermore, it is crucial to select different sensor types based on the geological environment. For example, laser displacement sensors can be used in soft rock areas, while high-sensitivity pressure sensors are more suitable for hard rock areas. This multi-level deployment strategy ensures that the system can obtain comprehensive rock formation change data within the mine, avoiding the limitations of single-point monitoring.

[0141] After the sensors are arranged, a dynamic sampling frequency adjustment mechanism needs to be set up to achieve real-time monitoring of rock formation changes. Normally, the system will collect data according to the set initial sampling frequency, such as once an hour. However, when the system detects abnormal changes in sensor data in a certain area, such as sudden displacement changes or sudden pressure increases, the dynamic sampling mechanism will be automatically triggered to immediately increase the sampling frequency of the area to minutes or seconds to capture key changes. The advantage of this dynamic adjustment mechanism is that it can effectively cope with the uncertainty of rock formation changes, especially in the early stages of roof instability, and can accurately record the change process to avoid data delays or losses caused by fixed sampling frequencies.

[0142] Dynamic sampling frequency adjustment relies not only on real-time sensor data feedback but also requires a comprehensive assessment based on historical data trends and geological parameters. For example, in areas with a history of fracture expansion, the system can set lower warning thresholds in advance to identify risk signals at an early stage. Linking the dynamic sampling mechanism with data analysis can significantly improve the system's monitoring accuracy, especially in complex geological environments, by avoiding misjudgments or omissions caused by data fluctuations. This linkage ensures that the system not only responds passively to data changes but also proactively adjusts its sampling strategy based on environmental characteristics.

[0143] The combination of a multi-layered sensor layout and dynamic sampling frequency adjustment effectively addresses issues such as data delays, blind spots, and high false alarm rates in online monitoring of mine roof separation. This approach enables the system to promptly respond to early abnormal changes in the mine roof, providing real-time data support and empowering mine safety managers with decision-making support. Future work will allow for further optimization of sensor sensitivity and data transmission rates, as well as the introduction of additional sensor types, such as fiber optic and acoustic sensors, to enhance the comprehensiveness and accuracy of data collection.

[0144] Specific implementation method 2: During online monitoring of mine roof delamination, relying solely on raw data collected by sensors cannot accurately reflect the actual changes in the rock formations. Therefore, by combining mine geological characteristic parameters, preprocessing the raw data, and then performing multi-dimensional data fusion analysis, the accuracy of data interpretation and the intelligence level of the monitoring system can be effectively improved.

[0145] First, comprehensive geological characteristic parameters of the mine need to be collected. These parameters include rock formation structure, the distribution of faults and joints, rock hardness, and the mining depth of the mine. This information is typically obtained through geological mapping, borehole sampling, and historical mine reports, and is categorized and entered into a geological information database. During this entry process, the geological characteristics of different areas of the mine need to be annotated, dividing the mine into several monitoring sub-areas. The geological characteristic labels of each sub-area serve as an important basis for subsequent data analysis, effectively avoiding data analysis bias caused by ignoring geological differences.

[0146] After collecting raw monitoring data, it needs to be preprocessed to ensure its authenticity and reliability. Preprocessing includes data cleaning, denoising, and outlier identification. Mining environments are complex, and data can contain a lot of noise interference, such as equipment vibration, air flow, and human intervention. Therefore, the system must apply filtering algorithms to remove these interfering signals and mark and remove outliers. For example, if a sensor experiences an extreme change in value within a short period of time, while other sensors do not detect similar changes, this data can be considered an outlier and removed to avoid misleading subsequent analysis.

[0147] After data preprocessing is complete, displacement, pressure, and geological parameter data need to be fused multidimensionally. In the temporal dimension, different types of data are synchronized with timestamps to ensure comparative analysis on the same timeline, identifying temporal trends in rock formation changes. In the spatial dimension, sensor data from different layers are matched, and 3D modeling is used to generate a comprehensive trend map of rock formation changes. This multidimensional fusion analysis enables the system to comprehensively understand rock formation changes from both temporal and spatial perspectives, identifying potential delamination risks such as crack propagation and interlayer slip.

[0148] After the fusion analysis is complete, the system generates a mine environmental characteristic model based on the data analysis results. This model includes the changing trends, risk levels, and geological characteristic labels for each monitored area. As new data is continuously collected, the system needs to dynamically update the environmental characteristic model to adapt to changes in different areas. This dynamic update of the environmental characteristic model makes the monitoring system more intelligent and adaptable, allowing it to adjust monitoring strategies and warning parameters according to different scenarios, significantly improving monitoring accuracy.

[0149] Specific implementation method 3: In the process of monitoring roof separation in mines, relying solely on fixed warning parameters often cannot adapt to complex geological environments. Therefore, by generating an environmental feature model and dynamically adjusting the warning parameters and sampling frequency for each area, the accuracy and flexibility of warnings can be significantly improved.

[0150] The construction of the environmental characteristic model requires a combination of historical mine monitoring data, real-time data, and geological characteristic parameters. The model divides the mine into different monitoring sub-areas and assigns corresponding change patterns and risk level labels to each area. The model includes elements such as rock formation type, stress concentration areas, historical separation locations, and change trends. The purpose of building the environmental characteristic model is to provide the system with a flexible monitoring and early warning strategy that can accurately adapt to different geological environments.

[0151] Guided by the environmental characteristic model, the system adaptively adjusts sampling frequency and warning thresholds based on the geological characteristics and changing trends of different regions. For example, in soft rock areas, the system can lower the warning threshold and increase the sampling frequency to detect crack expansion signals earlier; while in hard rock areas, the warning threshold can be appropriately increased to reduce unnecessary high-frequency monitoring. This adaptive adjustment mechanism significantly improves the system's monitoring efficiency and warning accuracy, avoiding false alarms or missed alarms caused by fixed parameter settings.

[0152] As new data is continuously collected, the environmental characteristic model requires real-time feedback and optimization. Based on data analysis, the system continuously adjusts the model's parameters to better reflect actual rock formation changes. This real-time optimization of the environmental characteristic model ensures that the system's monitoring strategy can be continuously updated as the mine environment changes, improving the long-term stability and reliability of monitoring.

[0153] Through multi-level sensor arrangement, dynamic sampling frequency adjustment and multi-dimensional data fusion analysis, the present invention effectively improves the interpretation accuracy of roof separation monitoring data and significantly reduces the risk of missed reports and false alarms. The environmental characteristic model is generated in combination with the geological characteristic parameters of the mine, so that the monitoring process can be adaptively adjusted according to different geological environments, thereby more accurately identifying abnormal signals such as rock displacement, stress changes and crack expansion. Especially in soft rock layers or areas with dense faults, the system can capture small but continuous change trends in advance and issue early warnings in time, avoiding the delayed warnings or misjudgments caused by fixed thresholds and single data source analysis in traditional monitoring systems. In addition, the system can automatically adjust the sampling frequency when an abnormal signal is detected to ensure that detailed data of key change processes are captured, thereby greatly improving the real-time and reliability of monitoring.

[0154] The present invention introduces an adaptive early warning mechanism of the environmental characteristic model, which enables the monitoring system to dynamically optimize the early warning parameters and sampling strategies according to the real-time changes in the mine, thereby achieving more flexible monitoring and early warning management. For high-risk areas such as soft rock formations, the system will automatically lower the early warning threshold and increase the sampling frequency, while in more stable areas, it can reduce unnecessary high-frequency monitoring and improve the efficiency of system resource utilization. At the same time, the adaptive early warning mechanism has the ability to continuously optimize. The system can continuously adjust the environmental characteristic model based on real-time feedback to enable it to maintain efficient operation for a long time and avoid the problem of "monitoring blind spots" or outdated strategies. This flexibility and continuous optimization capability not only reduces the false alarm rate and missed alarm rate, but also reduces the burden on mine management personnel and improves the safety, intelligence and stability of online monitoring of mine roof separation.

[0155] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formulas are set by technicians in this field according to actual conditions.

[0156] The above description is merely illustrative of certain exemplary embodiments of the present invention. It goes without saying that those skilled in the art will be able to modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and description are illustrative in nature and should not be construed as limiting the scope of protection of the claims.

[0157] It should be noted that, in this document, if there are relational terms such as first and second, etc., they are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "comprises", "comprising" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device that includes a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprising a ..." does not exclude the presence of other identical elements in the process, method, article or device that includes the element.

[0158] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0159] Those skilled in the art will appreciate that the units and algorithm steps of each example 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 performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel 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.

[0160] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0161] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0162] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0163] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

[0164] The above description is merely illustrative of certain exemplary embodiments of the present invention. It goes without saying that those skilled in the art will be able to modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and description are illustrative in nature and should not be construed as limiting the scope of protection of the claims.

Claims

1. A roof separation online monitoring method, characterized in that: The following steps are involved: Multiple sensors are deployed at different levels of the mine roof, and different types of sensor deployment schemes are selected based on geological conditions to comprehensively collect rock layer change data, ensuring wide coverage and comprehensive data collection; Set the dynamic sampling frequency of the sensor and automatically adjust the sampling interval according to the severity of the roof rock changes. When a sudden displacement change is detected, the sampling frequency is automatically increased to ensure that key changes are captured in real time. Pre-process the collected raw monitoring data, combine it with the mine geological characteristic parameters, incorporate the differences in geological environment into the monitoring analysis, and generate an environmental characteristic model; Perform multi-dimensional data fusion on displacement, pressure, and geological parameters, match different types of data in time and space, generate overall rock formation change trend maps, and identify potential separation locations and change patterns; A nonlinear trend recognition algorithm is used to analyze the multi-dimensional data fusion results. By detecting the rate of change, acceleration, and fluctuation amplitude of the data, early abnormal changes in the rock formation are identified, and normal changes are distinguished from potential roof instability risks. Based on the collected historical data and real-time data, the early warning model is continuously optimized using adaptive learning algorithms, so that it can automatically adjust parameters according to new data patterns, improve its adaptability to different geological environments, and thus enhance the accuracy and real-time nature of the early warning.

2. The method for online monitoring of roof separation according to claim 1, characterized in that: Multiple sensors are deployed at different levels of the mine roof, and different types of sensor deployment schemes are selected based on geological conditions to comprehensively collect rock layer change data. The specific steps to ensure wide coverage and comprehensive data collection are as follows: By comprehensively analyzing the mine's geological structure and identifying key monitoring areas, we ensure the accuracy and targeting of sensor placement; Select appropriate sensor types and layout schemes according to different geological environments to cover key areas of displacement and stress changes; Three-dimensional sensors are installed at different levels of the mine roof to form a multi-level monitoring network to ensure comprehensive and reliable data collection; By optimizing the sensor layout and introducing a redundant data acquisition mechanism, the stability of the monitoring system and the continuity of data acquisition are improved.

3. The method for online monitoring of roof separation according to claim 1, characterized in that: The specific steps for setting the dynamic sampling frequency of the sensor, automatically adjusting the sampling interval according to the severity of the roof rock changes, and automatically increasing the sampling frequency when a sudden displacement change is detected to ensure that key changes are captured in real time are as follows: Set the initial sampling frequency based on geological conditions and historical data to ensure that basic data collection covers early trends in roof changes; Identify abnormal change signals in data through real-time fluctuation monitoring mechanism to improve the accuracy of identifying key changes; When an abnormal signal is detected, the sampling frequency is automatically increased to ensure that the key change process is fully captured; The sampling strategy is dynamically adjusted through a multi-layer feedback mechanism to ensure flexible and adaptive sampling frequency in different changing scenarios.

4. The method for online monitoring of roof separation according to claim 1, characterized in that: The collected raw monitoring data is preprocessed, combined with the mine geological characteristic parameters, and the differences in the geological environment are incorporated into the monitoring analysis. The specific steps for generating the environmental characteristic model are as follows: Collect mine geological characteristic parameters and establish a geological information database to provide comprehensive environmental basic information for subsequent data analysis; Correlate and match monitoring data with geological characteristic parameters to improve the ability to interpret data fluctuations and the accuracy of analysis; Ensure the authenticity and reliability of monitoring data and reduce interference factors through data cleaning, denoising and outlier identification; Generate environmental characteristic models, dynamically adjust sampling and early warning strategies based on geological differences, and improve the accuracy and adaptability of the monitoring system.

5. The method for online monitoring of roof separation according to claim 1, characterized in that: By dynamically matching displacement, pressure, and geological parameters in the time dimension to generate a trend sequence, the specific steps for identifying the temporal patterns of rock formation changes and potential risk signals are as follows: Synchronize timestamps and establish a unified time series to ensure that different types of data can be compared and analyzed on the same timeline; By analyzing the changing trends of time series data, we can identify the fluctuation patterns of rock displacement and pressure and make a preliminary assessment of the risk of separation. Establish a time correlation model to analyze the mutual influence and temporal relationship between displacement, pressure and geological parameters; Dynamically update trend sequences and adjust monitoring strategies in real time to ensure rapid response to changes in rock formations.

6. The method for online monitoring of roof separation according to claim 1, characterized in that: The specific steps for integrating multi-layer data in the spatial dimension and using 3D modeling to generate a map of overall rock formation change trends and visually identify key change areas and potential separation locations are as follows: Spatial matching of sensor data from different layers generates a multi-layered map of mine roof changes, improving the comprehensiveness of data analysis. Use spatial matching data to generate a three-dimensional model to visually present rock formation changes and identify key change areas; By integrating time and space data, an overall change trend diagram is generated to show the change pattern of the rock layer and the potential separation location; Mark high-risk areas based on the changing trend chart and dynamically adjust warning parameters to improve the system's warning accuracy and flexibility.

7. The method for online monitoring of roof separation according to claim 1, characterized in that: A nonlinear trend recognition algorithm is used to analyze the multidimensional data fusion results. By detecting the rate of change, acceleration, and fluctuation amplitude of the data, early abnormal changes in the rock formation are identified, and the specific steps to distinguish normal changes from potential roof instability risks are as follows: When analyzing the results of multidimensional data fusion, we first construct a multidimensional change rate matrix and calculate the time derivative of the collected displacement, pressure, and geological parameter data to obtain the change rate of each type of data. The change rate calculation formula is as follows: Where M rate (i, t) is the change rate matrix element of the i-th sensor at time t, D i (t) is the original monitoring data of the i-th sensor at time t, Δt is the time interval, D i (t-Δt) is the original monitoring data of the i-th sensor at time t-Δt, that is, the original monitoring data of the previous moment; Based on the change rate matrix, the change acceleration matrix is ​​further calculated to measure the severity and mutation of data changes. The calculation expression of the change acceleration is as follows: Where M accel (i, t) is the acceleration matrix element of the i-th sensor at time t, M rate (i, t-Δt) is the change rate matrix element of the i-th sensor at time t-Δt, that is, the change rate matrix element at the previous moment.

8. The method for online monitoring of roof separation according to claim 7, characterized in that: Based on the rate of change and acceleration matrix, the fluctuation amplitude analysis parameters are calculated to identify abnormal fluctuations in the data. The fluctuation amplitude calculation formula is as follows: Where A wave (i) is the fluctuation amplitude analysis parameter of the i-th sensor, is the average rate of change of the i-th sensor, T is the total number of time points; Binding change rate M rate (i, t), acceleration M accel (i, t) and the fluctuation amplitude parameter A wave (i) Calculate the nonlinear risk assessment index to ultimately distinguish between normal changes and potential roof instability risks. The calculation formula is as follows: R risk (i) = α·M rate (i, t)+β·M accel (i, t)+γ·A wave (i), where R risk (i) is the nonlinear risk assessment index of the i-th sensor, α, β and γ are weight parameters, and α is used to adjust the change rate M rate (i, t) has an impact on the overall risk assessment, and β is used to adjust the acceleration M accel The influence of (i, t) parameters, γ is used to adjust the fluctuation amplitude parameter A wave (i) Impact.

9. The method for online monitoring of roof separation according to claim 1, characterized in that: Based on the collected historical data and real-time data, the early warning model is continuously optimized using adaptive learning algorithms, enabling it to automatically adjust parameters according to new data patterns, improving its adaptability to different geological environments, and thus enhancing the accuracy and real-time nature of early warnings. The specific steps are as follows: First, a basic early warning model is constructed based on the collected historical data and real-time data, and key parameters are defined, including the displacement change rate, pressure change rate, and environmental characteristic parameter weights. In order to comprehensively consider the impact of multidimensional data, a data weight matrix is ​​used to represent the contribution of each parameter to the early warning model. The formula is as follows: Where M0 is the basic warning weight matrix, which represents the weight distribution of each parameter of the initial model, w1, w2, ..., w9 are the weights of different data types at different time points; During real-time monitoring, the abnormal deviation rate between the actual data and the model prediction value is calculated based on the current displacement change rate and pressure change rate. The abnormal deviation rate calculation formula is as follows: Where D t is the abnormal deviation rate, V t is the real-time displacement change rate, is the displacement change rate predicted by the model, P t is the real-time pressure change rate, is the rate of change of pressure predicted by the model.

10. The method for online monitoring of roof separation according to claim 9, characterized in that: According to the calculated abnormal deviation rate D t , dynamically adjust the weight values ​​of the basic weight matrix M0 to generate a new weight matrix. The generation formula is as follows: M t =M0+ω·D t I, where M t is the weight matrix after real-time optimization, ω is the learning rate, and I is the identity matrix; According to the optimized weight matrix M t , recalculate the warning threshold to dynamically adjust the warning strategy for different areas. The calculation formula of the warning threshold is as follows: Where, T t is the real-time warning threshold, M t [p] is the pth weight value in the optimized weight matrix, F p It is the influencing factor of geological characteristic parameters.

Images