Coal mine hard top rock burst prevention and control method based on deep learning

Abstract

The invention relates to the technical field of rock burst prevention and control, and discloses a coal mine hard top rock burst prevention and control method based on deep learning, which comprises the steps of data monitoring, data processing, data analysis, prevention and control pressure relief and result output. In the monitoring link, stress, displacement and microseismic multi-dimensional monitoring is carried out, and the hard top state is comprehensively mastered. During data processing, a feature extraction method is used to mine the deep value of the data; in the analysis stage, single-type data and multi-dimensional comprehensive analysis are combined, and the rock burst occurrence possibility is accurately evaluated. And a control pressure relief step of formulating targeted measures according to the risk level to reduce the disaster risk. In addition, results are displayed visually, so that management personnel can make decisions conveniently, data are stored in a database, and support is provided for subsequent optimization. In general, the method effectively improves the prediction accuracy and prevention and control efficiency of the coal mine rock burst, guarantees the safety and stability of coal mining, and reduces the loss of personnel and properties caused by disasters.

Description

Technical Field

[0001] The present invention relates to the technical field of rock burst prevention and control, and particularly to a method for preventing and controlling hard roof rock bursts in coal mines based on deep learning. Background Art

[0002] Rock burst is a complex mine dynamic phenomenon. Due to its high strength and stiffness, the hard roof in coal mines accumulates a large amount of elastic energy during the mining process. When the stress conditions reach the critical value, the hard roof instantaneously releases energy, leading to the occurrence of rock bursts. Such disasters can not only cause equipment damage and roadway deformation, but also trigger secondary disasters such as gas explosion and water inrush, seriously threatening the lives of coal mine workers and also bringing huge economic losses to coal mine enterprises.

[0003] Traditional prevention and control technologies for hard roof rock bursts in coal mines mainly rely on empirical judgment and simple monitoring means. In the monitoring link, most only rely on a single type of sensor, making it difficult to comprehensively perceive the stress, displacement, rock fracture and other aspects of the hard roof. For example, only monitoring the stress change through stress sensors cannot obtain information on the displacement of the hard roof and the internal rock fracture, resulting in one-sided prediction of rock bursts. In terms of data processing and analysis, traditional methods lack systematicness and scientificity, mostly being simple data statistics, unable to deeply explore the laws behind the data and difficult to accurately judge the possibility of rock burst occurrence. When facing complex geological conditions and mining environments, traditional prevention and control technologies often cannot take targeted measures in a timely and effective manner, and the prevention and control effects are poor.

[0004] In view of this, it is of great practical significance to develop a comprehensive, scientific and efficient method for preventing and controlling hard roof rock bursts in coal mines to improve the safety production level of coal mines and reduce the losses caused by rock burst disasters. Summary of the Invention

[0005] The purpose of the present invention is to provide a method for preventing and controlling hard roof rock bursts in coal mines based on deep learning, which solves the technical problems raised in the background art.

[0006] The purpose of the present invention can be achieved by the following technical solutions:

[0007] A method for preventing and controlling hard roof rock bursts in coal mines based on deep learning includes the following steps:

[0008] Data monitoring step: Real-time collect stress data, displacement data and microseismic data of the hard roof area through stress sensors, laser displacement sensors and microseismic monitors;

[0009] Data processing step: Extract features from the collected stress data, displacement data and microseismic event data to generate stress change rate, displacement accumulation, displacement change rate, microseismic frequency and magnitude characteristic parameters;

[0010] Data analysis steps: Analyze the hard roof stress concentration trend, deformation state, rock fracture degree through the characteristic parameters, and calculate the possibility determination value of rock burst occurrence. Then, determine that the possibility of rock burst occurrence is low according to the possibility determination value.

[0011] Prevention and control pressure relief steps: Trigger the prevention and control pressure relief measures according to the classification of the determination results corresponding to the low possibility of rock burst occurrence.

[0012] Result output steps: Visualize and store the data analysis results and prevention and control pressure relief results in the database.

[0013] As a further solution of the present invention: The stress data includes the vertical stress and horizontal stress borne by the hard roof monitored in real time through stress sensors; the displacement data includes the vertical displacement and horizontal displacement of the hard roof in the vertical and horizontal directions measured through laser displacement sensors; the microseismic data includes the microseismic events corresponding to the microseismic signals generated inside the hard roof due to rock fracture monitored by a microseismic monitor; specifically collect the occurrence time, magnitude and epicenter location of the microseismic events.

[0014] As a further solution of the present invention: The extraction and processing method of the stress change rate characteristics is as follows:

[0015] Within the specified observation period, divide it into several time nodes according to the preset time interval t;

[0016] Then mark the vertical stress and horizontal stress obtained at each time node as YC i and YS i , where i = 1, 2,... n, and n represents the number of time nodes within the observation period;

[0017] Then, record the time period between two adjacent time nodes as the monitoring time period. At the same time, record each monitoring time period as k, where k is a variable value. According to the time trend, k = 1, 2,... n - 1, and through:

[0018] Calculate the change rate BC of the vertical stress within the corresponding monitoring time period k ;

[0019] At the same time, through:

[0020] Calculate the change rate BS of the horizontal stress within the corresponding monitoring time period k .

[0021] As a further solution of the present invention, the extraction and processing method of the displacement cumulative amount and displacement change rate is as follows:

[0022] Mark the vertical displacement and horizontal displacement obtained at each time node as WC i and WS i ;

[0023] Then, through: LC k∈[i-1，i] = WC i - WC i-1 ;

[0024] Calculate the cumulative amount LC of the vertical displacement within the corresponding monitoring time period k ;

[0025] Then, through: LS k∈[i-1，i] = WS i - WS i-1 ;

[0026] Calculate the cumulative amount LS of the horizontal displacement within the corresponding monitoring time period k ;

[0027] Subsequently, through:

[0028] Calculate the change rate VC of the vertical displacement within the corresponding monitoring time period k ;

[0029] Meanwhile, through:

[0030] Calculate the change rate VS of the horizontal displacement within the corresponding monitoring time period k .

[0031] As a further solution of the present invention: The extraction and processing method of the microseismic frequency and magnitude characteristic parameters is as follows:

[0032] During the observation period, count the number of microseismic events occurring at the same seismic source position within each monitoring time period and the magnitude of each event, and mark them as C j，k and M c，j，k ;

[0033] where j represents the serial number index of different seismic source positions.

[0034] As a further solution of the present invention: Data analysis steps:

[0035] The analysis method for the tendency of stress concentration is as follows:

[0036] When BC1 < BC2 <... BC n-1 or BS1 < BS2 <... BS n-1 , it indicates that the vertical stress change rate continues to increase or the horizontal stress change rate continues to increase, and it is determined that there is a tendency of hard roof stress concentration.

[0037] As a further solution of the present invention, the analysis method of the deformed state is as follows:

[0038] Obtain the cumulative amount LC1 of the vertical displacement in the first monitoring time period and the cumulative amount LC of the vertical displacement in the (n - 1)-th monitoring time period n-1 ;

[0039] Then, through: LCC = LC n-1 - LC1;

[0040] Calculate the cumulative difference LCC of the vertical displacement cumulative amount within the observation period;

[0041] Then compare LCC with the preset cumulative difference threshold LCCy:

[0042] When LCC > LCCy, it is determined that the cumulative amount of vertical displacement within the observation period increases rapidly;

[0043] Meanwhile, obtain the cumulative amount LS1 of the horizontal displacement in the first monitoring time period and the cumulative amount LS of the horizontal displacement in the (n - 1)-th monitoring time period n-1 ;

[0044] Then, through: LSC = LS n-1 - LS1;

[0045] Calculate the cumulative difference LSC of the horizontal displacement cumulative amount within the observation period;

[0046] Then compare LSC with the preset cumulative difference threshold LSCy:

[0047] When LSC > LSCy, it is determined that the cumulative amount of horizontal displacement within the observation period increases rapidly;

[0048] Subsequently, obtain the change rate VC1 of the vertical displacement in the first monitoring time period and the change rate VC of the vertical displacement in the (n - 1)-th monitoring time period n-1 ;

[0049] Then, through: VCC = VC n-1 - VC1;

[0050] Calculate the change rate difference VCC of the vertical displacement cumulative amount within the observation period;

[0051] Then compare VCC with the preset change rate threshold VCCy:

[0052] When VCC > VCCy, it is determined that a large deformation occurs in the vertical direction of the hard top within the observation period;

[0053] Simultaneously obtain the change rate VS1 of the horizontal displacement amount within the first monitoring time period and the change rate VS of the horizontal displacement amount within the (n - 1)-th monitoring time period n-1 ;

[0054] Then through: VSC = VS n-1 - VS1;

[0055] Calculate the change rate difference VSC of the cumulative horizontal displacement amount within the observation period;

[0056] Then compare VSC with the preset change rate threshold VSCy:

[0057] When VSC > VSCy, it is determined that there is a large deformation of the hard roof in the horizontal direction within the observation period.

[0058] As a further solution of the present invention: The analysis method of the rock fracture degree is as follows:

[0059] Through:

[0060] Calculate the change value PB of the occurrence frequency of microseismic events at the corresponding source location j ;

[0061] At the same time through:

[0062] Calculate the average magnitude MP of microseismic events at the corresponding source location j ;

[0063] Compare the average magnitude MP j with the preset magnitude threshold MPy:

[0064] When MP j > MPy and PB j > 0, it indicates that the magnitude of microseismic events gradually increases, the occurrence frequency of microseismic events significantly increases, and then it is judged that the rock fracture inside the hard roof intensifies, and the possibility of rock burst occurrence increases.

[0065] As a further solution of the present invention: The analysis method of the possibility determination value of rock burst occurrence is as follows:

[0066] Through:

[0067] Respectively calculate the determination index values YC0, YS0, WC0, WS0, C0, M0 of vertical stress, horizontal stress, vertical displacement amount, horizontal displacement amount, number of microseismic events, and magnitude corresponding to microseismic events j ;

[0068] Then through:

[0069] Rj = α1 × YC0 + α2 × YS0 + α3 × WC0 + α4 × WS0 + α5 × C0 + α6 × M0;

[0070] Calculate the possibility determination value R of the occurrence of hard roof rock burst at the corresponding seismic source position j ;

[0071] Wherein, α1, α2, α3, α4, α5, and α6 are pre-set proportionality coefficients;

[0072] Then compare the possibility determination value R at the corresponding seismic source position j with the pre-set possibility determination thresholds R1y and R2y:

[0073] When R j < R1y, it indicates that the possibility of rock burst occurrence is low;

[0074] When R1y ≤ R j < R2y, it indicates that the possibility of rock burst occurrence is medium;

[0075] When R j ≥ R1y, it indicates that the possibility of rock burst occurrence is high.

[0076] As a further solution of the present invention: the pressure relief prevention and control measures are as follows:

[0077] When the possibility of rock burst occurrence is low, regularly review the monitoring data, maintain the normal coal mining progress; at the same time, maintain and calibrate the monitoring equipment; and strengthen the safety education and training of coal mine workers;

[0078] When the possibility of rock burst occurrence is medium, drill the hard roof for pressure relief, and at the same time strengthen the support strength;

[0079] When the possibility of rock burst occurrence is high, immediately stop the coal mining operation in the relevant area, organize the workers to evacuate to a safe area; at the same time, start the emergency plan and notify the relevant departments and personnel to make preparations for response.

[0080] Advantages of the present invention:

[0081] Multi-dimensional data monitoring to achieve comprehensive perception:

[0082] This method constructs an all-round monitoring system covering stress, displacement and microseismicity. By arranging multiple stress sensors in the hard roof area of the coal mine, the vertical stress and horizontal stress are monitored in real time, and the dynamic changes of the stress borne by the hard roof are accurately captured; laser displacement sensors are installed at the interface between the hard roof and the coal seam and at key positions to measure the displacement in the vertical and horizontal directions, providing intuitive data for judging the deformation of the hard roof; microseismic monitors are arranged around the coal mining area to record the occurrence time, magnitude and epicenter location of microseismic events, so as to master the fracture information of the rock inside the hard roof. The collaborative work of multiple types of sensors ensures a comprehensive and accurate perception of the state of the hard roof in the coal mine, avoiding the limitations of a single monitoring method.

[0083] Scientific data processing to improve prediction accuracy:

[0084] In the data processing link, corresponding feature extraction methods are adopted for different types of data. For stress data, the change rates of vertical and horizontal stresses are calculated, which can clearly reflect the change trend of stress over time; for displacement data, the cumulative amount and change rate are calculated to quantify the deformation degree and speed of the hard roof; in the microseismic data processing, the occurrence times and magnitudes of microseismic events are counted, and then the changes in magnitude and occurrence frequency are obtained. These scientific data processing methods transform the original monitoring data into characteristic parameters with clear physical meanings, providing strong support for subsequent data analysis and rockburst prediction, and greatly improving the accuracy and reliability of prediction.

[0085] In-depth data analysis to achieve risk early warning:

[0086] In the data analysis stage, not only is a single type of data analyzed, but also a quantitative assessment of the possibility of rockburst occurrence is achieved through comprehensive analysis. By analyzing the characteristics of various types of data, abnormal situations such as stress concentration in the hard roof, increased displacement deformation, and intensified rock fracture can be detected in a timely manner. By calculating the determination value of the possibility of rockburst occurrence, the risk level is divided into three levels: low, medium and high, realizing the accurate early warning of rockburst risk. This multi-dimensional data analysis method provides a scientific basis for coal mine safety management, enabling managers to take corresponding prevention and control measures according to different risk levels.

[0087] Targeted prevention and control measures to ensure production safety:

[0088] Based on the evaluation results of the likelihood of rock burst occurrence, targeted prevention and control pressure relief strategies have been formulated. When the likelihood of rock burst is low, measures such as regularly reviewing monitoring data, maintaining monitoring equipment, and strengthening safety education and training are taken to ensure the normal progress of coal mine mining; when the likelihood is medium, borehole pressure relief is carried out on the hard roof and the support strength is enhanced to effectively reduce the risk of rock burst occurrence; when the likelihood is high, mining operations are immediately stopped, personnel are organized to evacuate, and the emergency plan is activated to maximize the protection of personnel's lives and the safety and stability of coal mine production. This hierarchical and targeted prevention and control measure improves the coal mine's ability to respond to rock burst and reduces the losses caused by disasters.

[0089] Result visualization and storage to assist management decision-making:

[0090] The analysis results of various types of data are presented to relevant management personnel and stored in a pre-established database. The visualized result presentation enables management personnel to intuitively understand the state of the hard roof of the coal mine and the risk level of rock burst, providing convenience for decision-making; the storage of data provides rich historical data for subsequent data analysis, model optimization, and experience summary, helping to continuously improve the prevention and control methods of hard roof rock burst in coal mines and enhancing the level of coal mine safety management. Brief Description of the Drawings

[0091] The present invention will be further described below in conjunction with the accompanying drawings.

[0092] Figure 1 It is a system block diagram of a method for preventing and controlling hard roof rock burst in coal mines based on deep learning according to the present invention.

[0093] Figure 2 It is a schematic flow chart of the data processing step in a method for preventing and controlling hard roof rock burst in coal mines based on deep learning according to the present invention.

[0094] Figure 3 It is a schematic flow chart of the data analysis step in a method for preventing and controlling hard roof rock burst in coal mines based on deep learning according to the present invention. Detailed Embodiments

[0095] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0096] Embodiment 1

[0097] Please refer to Figure 1 and Figure 2As shown in the figure, the present invention is a method for preventing and controlling hard roof rock bursts in coal mines based on deep learning, comprising the following steps:

[0098] Data monitoring step:

[0099] Stress monitoring: In the hard roof area of the coal mine, a plurality of stress sensors are arranged at a specified spacing, and the vertical stress and horizontal stress borne by the hard roof are monitored in real time through the stress sensors;

[0100] Displacement monitoring: Laser displacement sensors are installed at the interface between the hard roof and the coal seam and at key parts of the hard roof, and the vertical displacement and horizontal displacement amounts of the hard roof in the vertical and horizontal directions are measured through the laser displacement sensors;

[0101] Microseismic monitoring: Microseismic monitors are arranged around the coal mining area to monitor microseismic events corresponding to microseismic signals generated due to rock fractures inside the hard roof; specifically, the occurrence time, magnitude, and epicenter location of the microseismic events are collected;

[0102] Data processing step:

[0103] Stress data processing: Feature extraction processing is performed on the collected vertical stress and horizontal stress; the specific method is as follows:

[0104] Within a specified observation period, it is divided into several time nodes at a preset time interval t;

[0105] Subsequently, the vertical stress and horizontal stress obtained at each time node are respectively marked as YC i and YS i , where i = 1, 2,..., n, and n represents the number of time nodes within the observation period;

[0106] Then, the time period between two adjacent time nodes is recorded as the monitoring time period. At the same time, each monitoring time period is recorded as k, where k is a variable value. According to the time trend, k = 1, 2,..., n - 1, and through:

[0107] The change rate BC of the vertical stress within the corresponding monitoring time period is calculated k ;

[0108] At the same time, through:

[0109] The change rate BS of the horizontal stress within the corresponding monitoring time period is calculated k ;

[0110] Displacement data processing: Feature extraction processing is performed on the collected vertical displacement and horizontal displacement amounts; the specific method is as follows:

[0111] Mark the vertical displacement and horizontal displacement obtained at each time node as WC i and WS i ;

[0112] Then, through: LC k∈[i-1，i] =WC i -WC i-1 ;

[0113] Calculate the cumulative amount LC of the vertical displacement within the corresponding monitoring time period k ;

[0114] Then, through: LS k∈[i-1，i] =WS i -WS i-1 ;

[0115] Calculate the cumulative amount LS of the horizontal displacement within the corresponding monitoring time period k ;

[0116] Subsequently, through:

[0117] Calculate the change rate VC of the vertical displacement within the corresponding monitoring time period k ;

[0118] Meanwhile, through:

[0119] Calculate the change rate VS of the horizontal displacement within the corresponding monitoring time period k ;

[0120] Microseismic data processing: Extract the characteristics of the magnitude and occurrence frequency of microseismic events; The specific method is as follows:

[0121] During the observation period, count the number of occurrences of microseismic events at the same seismic source location within each monitoring time period and the magnitude of each time, and mark them as C j，k and M c，j，k ;

[0122] Among them, j represents the serial number index of different seismic source locations;

[0123] Data analysis steps:

[0124] Data comprehensive analysis:

[0125] Through: Calculate the determination index value YC0 of the vertical stress;

[0126] Through: Calculate the determination index value YS0 of the horizontal stress;

[0127] Through: Calculate the determination index value WC0 of the vertical displacement

[0128] By: Calculate the determination index value WS0 of the horizontal displacement

[0129] By: Calculate the determination index value C0 of the number of microseismic events

[0130] By: Calculate the determination index value M0 of the magnitude corresponding to the microseismic event j ;

[0131] By: R j = α1×YC0 + α2×YS0 + α3×WC0 + α4×WS0 + α5×C0 + α6×M0 j Wherein, α1, α2, α3, α4, α5, α6 are preset proportionality coefficients, and α1 + α2 + α3 + α4 + α5 + α6 = 1;

[0132] Calculate the possibility determination value R of the occurrence of hard roof rock burst at the corresponding focal position j ;

[0133] Then compare the possibility determination value R at the corresponding focal position j with the preset possibility determination thresholds R1y and R2y:

[0134] When R j < R1y, it indicates that the possibility of rock burst occurrence is low;

[0135] When R1y ≤ R j < R2y, it indicates that the possibility of rock burst occurrence is medium;

[0136] When R j ≥ R1y, it indicates that the possibility of rock burst occurrence is high;

[0137] Prevention and control pressure relief steps:

[0138] When the possibility of rock burst occurrence is low, regularly review the monitoring data and maintain the normal coal mining progress;

[0139] At the same time, maintain and calibrate the monitoring equipment to ensure the accuracy of the data; and strengthen the safety education and training of coal mine workers to improve their awareness and prevention awareness of rock burst;

[0140] When the possibility of rock burst occurrence is medium, drill holes for pressure relief on the hard roof and strengthen the support strength at the same time;

[0141] Among them, borehole pressure relief is to determine the position, diameter, and depth of the borehole according to the geological conditions and stress distribution of the hard roof; strengthening the support strength is to increase the number of bolts and cables or improve their specifications on the basis of the original support, or increase the length and pre-tightening force of the cables; at the same time, reinforce the roadway.

[0142] In this embodiment, through borehole pressure relief, part of the stress in the hard roof is released, and the degree of stress concentration is reduced; additional shed support or shotcrete is provided to improve the bearing capacity of the roadway.

[0143] When the possibility of rock burst is high, immediately stop the coal mining operation in the relevant area, organize the staff to evacuate to a safe area; at the same time, start the emergency plan and notify the relevant departments and personnel to make preparations for response.

[0144] Among them, for the preparation for response, the method of blasting pressure relief is adopted to conduct large-scale pressure relief treatment on the hard roof.

[0145] Blasting pressure relief is to determine the position, depth, spacing, and charge amount of the blasting holes according to the rock mechanical properties and stress state of the hard roof.

[0146] After blasting pressure relief, conduct a comprehensive inspection and repair on the hard roof and the roadway.

[0147] Specifically, check whether the roadway support is damaged, and if there is damage, repair and reinforce it in time; re-monitor the stress, displacement, and microseismic conditions of the hard roof, and gradually resume the coal mining operation after each index returns to the safe range.

[0148] This embodiment constructs a systematic prevention and control system for rock burst in coal mine hard roofs. Through stress, displacement, and microseismic monitoring, data is collected comprehensively, and various types of data are subjected to feature extraction and processing, and key parameters such as stress change rate, displacement accumulation, and change rate are calculated. With the help of a comprehensive analysis model, the judgment value is compared with the preset threshold to accurately judge the possibility of rock burst. According to different possibilities, targeted prevention and control pressure relief strategies are adopted, such as regular re-inspection, borehole pressure relief, and blasting pressure relief, etc. This not only ensures the safety of personnel during the mining process, but also reduces the impact of rock burst on coal mining through reasonable prevention and control measures, ensuring the continuity and safety of coal mining operations.

[0149] Embodiment 2

[0150] As Embodiment 2 of the present invention, when this application is specifically implemented, compared with Embodiment 1, the difference in the technical solution of this embodiment from that of Embodiment 1 is only that in this embodiment, the data analysis steps are as follows:

[0151] Stress data analysis:

[0152] By analyzing the magnitude and trend of the stress change rate, judge the change of the hard roof stress state;

[0153] During the observation period, if the vertical stress change rate or the horizontal stress change rate continues to increase, it indicates that the hard roof stress has a concentration trend;

[0154] Specifically:

[0155] When BC1 < BC2 <... BC n-1 or BS1 < BS2 <... BS n-1 , it means that the vertical stress change rate continues to increase or the horizontal stress change rate continues to increase, and it is determined that the hard roof stress has a concentration trend;

[0156] Displacement data analysis:

[0157] By analyzing the magnitude and trend of the displacement cumulative amount and the displacement change rate, judge the change of the hard roof displacement state;

[0158] During the observation period, when the displacement cumulative amount increases rapidly, or the displacement change rate increases significantly, it indicates that the hard roof may have undergone large deformation;

[0159] Specifically:

[0160] Obtain the cumulative amount LC1 of the vertical displacement in the first monitoring time period and the cumulative amount LC of the vertical displacement in the n-1th monitoring time period n-1 ;

[0161] Then through: LCC = LC n-1 -LC1;

[0162] Calculate the cumulative difference LCC of the vertical displacement cumulative amount during the observation period;

[0163] Then compare LCC with the preset cumulative difference threshold LCCy:

[0164] When LCC > LCCy, it is determined that the vertical displacement cumulative amount increases rapidly during the observation period;

[0165] At the same time, obtain the cumulative amount LS of the horizontal displacement in the first monitoring time period and the cumulative amount LS of the horizontal displacement in the n-1th monitoring time period n-1 ;

[0166] Then through: LSC = LS n-1 -LS1;

[0167] Calculate the cumulative difference LSC of the horizontal displacement cumulative amount during the observation period;

[0168] Then compare LSC with the preset cumulative difference threshold LSCy:

[0169] When LSC > LSCy, it is determined that the cumulative horizontal displacement increases rapidly during the observation period;

[0170] Subsequently, obtain the change rate VC1 of the vertical displacement during the first monitoring time period and the change rate VC of the vertical displacement during the (n - 1)-th monitoring time period n-1 ;

[0171] Then, through: VCC = VC n-1 - VC1;

[0172] Calculate the change rate difference VCC of the cumulative vertical displacement during the observation period;

[0173] Then compare VCC with the preset change rate threshold VCCy:

[0174] When VCC > VCCy, it is determined that there is a large deformation of the hard roof in the vertical direction during the observation period;

[0175] Meanwhile, obtain the change rate VS1 of the horizontal displacement during the first monitoring time period and the change rate VS of the horizontal displacement during the (n - 1)-th monitoring time period n-1 ;

[0176] Then, through: VSC = VS n-1 - VS1;

[0177] Calculate the change rate difference VSC of the cumulative horizontal displacement during the observation period;

[0178] Then compare VSC with the preset change rate threshold VSCy:

[0179] When VSC > VSCy, it is determined that there is a large deformation of the hard roof in the horizontal direction during the observation period;

[0180] Microseismic data analysis:

[0181] During the observation period, when the magnitude M of the microseismic event gradually increases, or the microseismic occurrence frequency significantly increases, it indicates that the rock fracture inside the hard roof intensifies, and the possibility of rock burst increases;

[0182] Specifically:

[0183] Through:

[0184] Calculate the change value PB of the microseismic event occurrence frequency at the corresponding seismic source location j ;

[0185] Meanwhile, through:

[0186] Calculate the average magnitude MP of microseismic events at the corresponding hypocenter positions j ;

[0187] Compare the average magnitude MP j with a pre-set magnitude threshold MPy:

[0188] When MP j > MPy and PB j > 0, it indicates that the magnitude of microseismic events gradually increases, the occurrence frequency of microseismic events significantly increases, and then it is judged that the rock fracture inside the hard roof intensifies, and the possibility of rock burst occurrence increases;

[0189] Meanwhile, this embodiment also includes a result output step:

[0190] The result output step is to display the stress data analysis, displacement data analysis, microseismic data analysis, and data comprehensive analysis results to relevant management personnel, and at the same time store them in the corresponding pre-established database;

[0191] Based on Embodiment 1, this embodiment further enriches the data analysis dimension. By analyzing the stress change rate, displacement accumulation and change rate, as well as the magnitude and occurrence frequency trend of microseismic events, the stress and displacement states of the hard roof, and the possibility of rock burst occurrence are judged more deeply. In addition, the result output step displays various analysis results to management personnel and stores them in the database, facilitating subsequent query and analysis. This helps management personnel more comprehensively and timely grasp the situation of the coal mine hard roof, make more scientific decisions, and further improve the accuracy and timeliness of the coal mine rock burst prevention and control work.

[0192] Embodiment 3

[0193] As Embodiment 3 of the present invention, when this application is specifically implemented, compared with Embodiment 1 and Embodiment 2, the technical solution of this embodiment is to combine and implement the solutions of the above Embodiment 1 and Embodiment 2.

[0194] This embodiment integrates the technical solutions of Embodiment 1 and Embodiment 2, having both the prevention and control system of Embodiment 1 and the rich data analysis dimension and result output function of Embodiment 2. On the one hand, it can effectively respond to rock burst through comprehensive monitoring and targeted prevention and control measures; on the other hand, with the help of multi-dimensional data analysis and result output, it provides more comprehensive information support for management personnel. This comprehensive solution realizes the multi-level optimization of the coal mine hard roof rock burst prevention and control work, and maximally ensures the safe and efficient progress of coal mine mining.

[0195] The above formulas are all dimensionless and take their numerical calculations. The formulas are obtained by collecting a large amount of data for software simulation to get a formula closest to the actual situation. The preset parameters and threshold selection in the formulas are set by those skilled in the art according to the actual situation.

[0196] As described above, the above are only specific embodiments of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present application can easily think of changes or substitutions, which should all be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for preventing and controlling hard roof rock bursts in coal mines based on deep learning, characterized in that, It includes the following steps: Data monitoring step: The stress data, displacement data and microseismic data of the hard roof area are collected in real time through stress sensors, laser displacement sensors and microseismic monitors; Data processing step: Feature extraction is performed on the collected stress data, displacement data and microseismic event data to generate stress change rate, displacement accumulation, displacement change rate, microseismic frequency and magnitude characteristic parameters; Data analysis step: The stress concentration trend, deformation state, rock fracture degree of the hard roof are analyzed through the characteristic parameters, and the possibility determination value of rock burst occurrence is calculated. Then, according to the possibility determination value, it is determined that the possibility of rock burst occurrence is low; Prevention and control pressure relief step: The prevention and control pressure relief measures are triggered in grading according to the determination result corresponding to the low possibility of rock burst occurrence; Result output step: The data analysis results and prevention and control pressure relief results are visualized and stored in the database.

2. The coal mine hard roof rock burst prevention and control method based on deep learning according to claim 1, wherein, The stress data includes the vertical stress and horizontal stress borne by the hard roof monitored in real time through stress sensors; the displacement data includes the vertical displacement and horizontal displacement of the hard roof measured in the vertical and horizontal directions through laser displacement sensors; the microseismic data includes microseismic events corresponding to microseismic signals generated inside the hard roof due to rock fracture monitored by microseismic monitors. Specifically, the occurrence time, magnitude and epicenter location of microseismic events are collected.

3. The method for preventing and controlling hard roof rock burst in coal mines based on deep learning according to claim 2, characterized in that, The extraction and processing method of the stress change rate feature is as follows: Within a specified observation period, it is divided into several time nodes according to a preset time interval t; The vertical stress and horizontal stress obtained at each time node are respectively marked as YC i and YS i , where i = 1, 2, …… n, and n represents the number of time nodes within the observation period; Then, the time period between two adjacent time nodes is recorded as the monitoring time period. At the same time, each monitoring time period is recorded as k, where k is a variable value. According to the time trend, k = 1, 2,... n - 1; During the monitoring time period k, the vertical stress YC at the current time node i is subtracted from the vertical stress YC at the previous time node i-1 , and the result is divided by the time interval t to obtain the change rate of the vertical stress within the corresponding monitoring time period, which is labeled as BC k ; Meanwhile, within the monitoring time period k, the horizontal stress YS at the current time node i is subtracted from the horizontal stress YS at the previous time node i-1 . The difference is then divided by the time interval t to obtain the change rate of the horizontal stress within the corresponding monitoring time period, which is labeled as BS k .

4. The prevention and control method for hard roof rock burst in coal mines based on deep learning according to claim 3, characterized in that The extraction and processing method of the displacement accumulation and displacement change rate is as follows: Mark the vertical displacement and horizontal displacement obtained at each time node as WC i and WS i ; During the monitoring time period k, calculate the vertical displacement WC of the current time node i and the vertical displacement WC of the previous time node i-1 to obtain the cumulative amount of the vertical displacement within the corresponding monitoring time period, and mark it as LC k ; Meanwhile, within the monitoring time period k, calculate the horizontal displacement WS at the current time node i and the horizontal displacement WS at the previous time node i-1 to obtain the cumulative amount of the horizontal displacement within the corresponding monitoring time period, and label it as LS k ; During the monitoring time period k, divide the cumulative vertical displacement LC k by the time interval t to obtain the change rate of the vertical displacement amount within the corresponding monitoring time period, and label it as VC k ; During the monitoring time period k, the cumulative horizontal displacement LS k is divided by the time interval t to obtain the change rate of the vertical displacement amount within the corresponding monitoring time period, and it is denoted as VC k .

5. The prevention and control method for hard roof rock burst in coal mines based on deep learning according to claim 4, characterized in that, The extraction and processing method of the microseismic frequency and magnitude characteristic parameters is as follows: During the observation period, count the number of microseismic events occurring at the same source location in each monitoring time period and the magnitude of each event, and label them as C j，k and M c，j，k ; Among them, j represents the serial number index of different epicenter positions.

6. The coal mine hard roof rock burst prevention and control method based on deep learning according to claim 5, characterized in that, Data analysis step: The analysis method of the stress concentration trend is as follows: When BC1 < BC2 < …… BC n-1 or BS1 < BS2 < …… BS n-1 , it indicates that the vertical stress change rate continues to increase or the horizontal stress change rate continues to increase, and it is determined that there is a tendency for hard roof stress to concentrate.

7. The coal mine hard roof rock burst prevention and control method based on deep learning according to claim 6, characterized in that, The analysis method of the deformation state is as follows: Obtain the cumulative amount LC1 of the vertical displacement in the first monitoring period and the cumulative amount LC of the vertical displacement in the (n - 1)-th monitoring period n-1 ; Followed by: LCC = LC n-1 - LC1; Calculate the cumulative difference LCC of the vertical displacement accumulation within the observation period; Then compare LCC with the preset cumulative difference threshold LCCy: When LCC > LCCy, it is determined that the vertical displacement accumulation within the observation period increases rapidly; Simultaneously obtain the cumulative amount LS1 of the horizontal displacement in the first monitoring time period and the cumulative amount LS of the horizontal displacement in the (n - 1)-th monitoring time period n-1 ; Followed by: LSC = LS n-1 -LS1; Calculate the cumulative difference LSC of the horizontal displacement accumulation within the observation period; Then compare LSC with the preset cumulative difference threshold LSCy: When LSC > LSCy, it is determined that the horizontal displacement accumulation within the observation period increases rapidly; Subsequently, obtain the change rate VC1 of the vertical displacement amount within the first monitoring time period and the change rate VC of the vertical displacement amount within the (n - 1)-th monitoring time period n-1 ; Next, through: VCC = VC n-1 - VC1; Calculate the change rate difference VCC of the vertical displacement accumulation within the observation period; Then compare VCC with the preset change rate threshold VCCy: When VCC > VCCy, it is determined that there is a large deformation of the hard roof in the vertical direction within the observation period; Simultaneously obtain the change rate VS1 of the horizontal displacement amount in the first monitoring time period and the change rate VS of the horizontal displacement amount in the (n - 1)-th monitoring time period n-1 ; Then, through: VSC = VS n-1 - VS1; Calculate the change rate difference VSC of the horizontal displacement accumulation within the observation period; Then compare VSC with the preset change rate threshold VSCy: When VSC > VSCy, it is determined that there is a large deformation of the hard roof in the horizontal direction within the observation period.

8. The prevention and control method for hard roof rock burst in coal mines based on deep learning according to claim 7, characterized in that, The analysis method of the rock fracture degree is as follows: Adopted by: Calculate the change value PB of the microseismic event occurrence frequency at the corresponding focal position j ; At the same time, divide the sum of the magnitudes M of the microseismic events at the same seismic source location within all monitoring time periods by the total number of microseismic events to obtain the average magnitude MP of the microseismic events at the corresponding seismic source location c，j，k ; j ; Compare the average magnitude MP j with a pre-set magnitude threshold MPy: When MP j > MPy and PB j > 0, it indicates that the magnitude of the microseismic event gradually increases, the occurrence frequency of the microseismic event significantly increases, and then it is judged that the rock fracture inside the hard roof intensifies, and the possibility of rock burst occurrence increases.

9. The coal mine hard roof rock burst prevention and control method based on deep learning according to claim 8, characterized in that, The analysis method of the possibility determination value of rock burst occurrence is as follows: Adopted by: Calculate the determination index values YC0, YS0, WC0, WS0, C0, and M0 of the vertical stress, horizontal stress, vertical displacement, horizontal displacement, number of microseismic events, and magnitude corresponding to the microseismic events respectively j ; Then through: R j = α1 × YC0 + α2 × YS0 + α3 × WC0 + α4 × WS0 + α5 × C0 + α6 × M0; Calculate the possibility determination value R of hard roof rockburst occurring at the corresponding seismic source location j ; Wherein, α1, α2, α3, α4, α5, and α6 are preset proportionality coefficients; Subsequently, the likelihood determination value R at the corresponding seismic source position j is compared with the preset likelihood determination thresholds R1y and R2y: When R j <R1y, it indicates a low possibility of rock burst occurrence; When R1y ≤ R j < R2y, it indicates that the possibility of rock burst occurrence is medium; When R j ≥ R1y, it indicates a high possibility of rock burst occurrence.

10. A method for preventing and controlling hard roof rock bursts in coal mines based on deep learning according to claim 9, characterized in that, The pressure relief prevention and control measures are as follows: When the possibility of rock burst is low, the monitoring data shall be regularly rechecked to maintain the normal coal mining progress; meanwhile, the monitoring equipment shall be maintained and calibrated; and the safety education and training of coal mine workers shall be strengthened; When the possibility of rock burst is medium, the hard roof shall be drilled for pressure relief, and the support strength shall be strengthened at the same time; When the possibility of rock burst is high, the coal mining operations in the relevant area shall be immediately stopped, and the staff shall be organized to evacuate to the safe area; meanwhile, the emergency plan shall be activated, and relevant departments and personnel shall be notified to make preparations for response.

Images