Intelligent monitoring method and system for water disaster of top plate of steeply inclined working face

Abstract

The invention relates to an intelligent monitoring method and system for a top plate water disaster of a steeply inclined working face, and the method comprises the steps: collecting underground real-time monitoring data based on a preset sampling frequency; wherein the real-time monitoring data comprises micro-seismic sensing data, water level data, water pressure data and humidity data; inputting the real-time monitoring data into an early warning model to obtain a prediction result; wherein the prediction result comprises a disaster type, an occurrence area and time; the early warning model is obtained by training and adjusting a long short-term memory network model through historical monitoring data; and when the prediction result exceeds a preset early warning threshold value, triggering early warning. The invention aims to realize whole-process management from real-time monitoring to data analysis and early warning, and provides technical guarantee for safe, efficient and green mining of the roof water-rich and steeply inclined working face.

Description

Technical Field

[0001] The present invention relates to the technical field of mine safety monitoring, and particularly to an intelligent monitoring method and system for roof water disasters in steeply inclined working faces. Background Art

[0002] During the coal mining process, the roof water disaster in steeply inclined working faces is a serious safety hazard, which may cause various hazards. Steeply inclined coal seams are widely distributed in western mining areas of China. In these areas, the surface ecological environment is usually fragile, the burial depth of coal seams varies, and roof water not only exists as a resource but is also an important disaster-causing factor. The specific background is as follows:

[0003] Asymmetric and non-uniform damage: The dip angle and gravity effect of steeply inclined working faces cause asymmetric and non-uniform damage to surrounding rocks, and the self-stabilizing ability of mining equipment is poor. The control of surrounding rock stability becomes complex, and factors such as coal seam environment, coal seam occurrence depth, water, and gas further increase the difficulty of surrounding rock control. Dual nature of water: When the coal seam is shallowly buried, roof water is both a precious resource and a potential disaster source. Mining activities may cause roof water to flow out through fissures, damage the surface ecological environment, and pose a threat to production safety due to water disasters. Risk of deep-buried aquifers: In deep-buried coal seams, although it is difficult for aquifers to be replenished from surface water sources and rainfall, when water-conducting fissures penetrate the aquifer, the water inflow to the working face may be accompanied by quicksand and silt. Although this does not cause serious water disasters, it will flood equipment, increase the drainage workload, and affect the efficiency of mechanized mining. Changes in the physical and mechanical properties of surrounding rocks: After water seeps into the surrounding rocks, it will change their physical and mechanical properties, weaken the bearing capacity of the surrounding rocks, and increase production risks. Therefore, regardless of the conditions, the prevention and control of roof water disasters are the key to ensuring efficient mechanized production. After the key strata such as the main roof break in the roof of steeply inclined working faces, mining-induced fissures will form in the overlying strata. These fissures gradually converge and penetrate the aquifer to form a water-conducting channel, thereby triggering roof water disasters. The complexity of roof failure characteristics makes the prevention and control of roof water disasters more challenging.

[0004] The roof water disaster in steeply inclined working faces of mines is a major hidden danger in mine safety production. In the prior art, single monitoring means such as traditional hydrogeological monitoring and microseismic monitoring have problems such as limited monitoring range and low warning accuracy, and it is difficult to effectively predict and prevent roof water disasters in steeply inclined working faces. Therefore, there is an urgent need for an intelligent monitoring method that combines multiple monitoring means to improve the monitoring accuracy and warning ability of roof water disasters in steeply inclined working faces. Summary of the Invention

[0005] The object of the present invention is to provide an intelligent monitoring method and system for roof water disasters in steeply inclined working faces, which comprehensively utilize microseismic monitoring, sensor networks and artificial intelligence technologies. The method aims to achieve the whole-process management from real-time monitoring to data analysis and early warning, providing technical support for the safe, efficient and green mining of water-rich roofs and steeply inclined working faces.

[0006] To achieve the above object, the present invention provides the following solutions:

[0007] An intelligent monitoring method for roof water disasters in steeply inclined working faces, comprising:

[0008] Collecting real-time monitoring data underground based on a preset sampling frequency; wherein, the real-time monitoring data includes: microseismic sensing data, water level data, water pressure data, humidity data;

[0009] Inputting the real-time monitoring data into an early warning model to obtain a prediction result; wherein, the prediction result includes: disaster type, occurrence area and time; the early warning model is obtained by training and adjusting a long short-term memory network model with historical monitoring data;

[0010] When the prediction result exceeds a preset early warning threshold, an early warning is triggered.

[0011] Optionally, the microseismic sensing data includes: earthquake source location, magnitude, frequency;

[0012] Collecting the earthquake source location includes: recording the arrival time of microseismic events, combining the spatial coordinates of sensors, and calculating the earthquake source location using the Geiger iterative method.

[0013] Optionally, training the long short-term memory network model with historical monitoring data includes:

[0014] Performing data analysis on the historical monitoring data to obtain the corresponding disaster type, occurrence area and time;

[0015] Based on the historical monitoring data and the corresponding analysis results, constructing a data set; wherein, the historical monitoring data is divided into a training set (70%), a validation set (15%) and a test set (15%) in chronological order, ensuring that the training set data time is earlier than the validation set and the test set to avoid future data leakage;

[0016] Training the long short-term memory network model with the data set, adjusting the model parameters, and validating the model using the cross-validation method.

[0017] Optionally, performing data analysis on the historical monitoring data includes:

[0018] Preprocessing the historical monitoring data;

[0019] Extract features from the preprocessed data to obtain time-domain features;

[0020] Based on the time-domain features, use the Fourier transform method to extract frequency-domain features;

[0021] Use the wavelet transform method for the frequency-domain features to extract time-frequency features;

[0022] Based on the time-frequency features, perform pattern recognition, use the Geiger iteration method to determine the source location of microseisms, and identify potential disaster sources;

[0023] Use the support vector machine algorithm to classify microseismic events and distinguish normal microseisms from abnormal microseisms.

[0024] Optionally, the preprocessing of the historical monitoring data includes:

[0025] Clean the historical monitoring data using the median filtering method;

[0026] Use the Min-Max scaling method to normalize the cleaned data.

[0027] An intelligent monitoring system for roof water disasters in steeply inclined working faces, the system includes: a sensor layout module, a data acquisition module, a data transmission module, a data storage module, and a data analysis and early warning model;

[0028] The sensor layout module is used to layout a sensor network based on a preset position; wherein, the sensor network includes: microseismic sensors and collectors, water level gauges, water pressure gauges, and hygrometers;

[0029] The data acquisition module is used to acquire multi-source data underground based on a preset sampling frequency through the sensor network;

[0030] The data transmission module is used to transmit the acquired multi-source data by means of optical fiber cabling and wireless transmission;

[0031] The data storage module is used to store the multi-source data;

[0032] The data analysis and early warning model is used to analyze and process the multi-source data and issue an early warning.

[0033] Optionally, in the sensor layout module,

[0034] The microseismic sensors are fixed on the protruding bolts in the coal rib, the microseismic sensors are connected by cables, the collectors are suspended on the side rib of the roadway, and each collector receives the monitoring data from 2 microseismic sensors; wherein, the collectors are densely laid out in high-risk areas;

[0035] The water level gauge is arranged near the lower part of the working face and the floor fissures;

[0036] The water pressure gauge is arranged in the surrounding rock and the area where fissures develop;

[0037] The humidity gauge is arranged in the area where the air circulation in the working face is poor.

[0038] Optionally, the data storage module adopts a distributed storage architecture, supporting horizontal expansion of data and multi-copy storage;

[0039] The data storage module establishes multi-level indexes for multi-source data, including: time index, geographical location index, and sensor type index;

[0040] The data storage module adopts a redundant storage design.

[0041] Optionally, the data analysis and early warning model includes: a model construction sub-module and an analysis and early warning sub-module;

[0042] The model construction sub-module is used to train and adjust the long short-term memory network model through historical monitoring data to construct an early warning model;

[0043] The analysis and early warning sub-module is used to input real-time monitoring data into the early warning model to obtain a prediction result; when the prediction result exceeds a preset early warning threshold, an early warning is triggered; wherein, the prediction result includes: disaster type, occurrence area, and time.

[0044] Optionally, the model construction sub-module includes: a data analysis unit, a data set construction unit, and a model training unit;

[0045] The data analysis unit is used to perform data analysis on the historical monitoring data to obtain the corresponding disaster type, occurrence area, and time;

[0046] The data set construction unit is used to construct a data set based on the historical monitoring data and the corresponding analysis results;

[0047] The model training unit is used to train the long short-term memory network model with the data set, adjust the model parameters, and verify the model using the cross-validation method.

[0048] The beneficial effects of the present invention are:

[0049] By introducing advanced monitoring and analysis technologies, the present invention realizes efficient, accurate, and automated monitoring and early warning of roof water disasters in steeply inclined working faces. The present invention does not rely on a single type of sensor, but integrates multiple sensors such as microseismic sensors, water level gauges, water pressure gauges, and hygrometers. This cross-field sensor integration design can real-time obtain key environmental parameters such as vibration, pressure, water level, and humidity within the working face, providing multi-dimensional data support for the monitoring of roof water disasters. Through advanced data fusion algorithms, the present invention can integrate data from different sensors and extract comprehensive information with higher information content and higher signal-to-noise ratio. This data fusion technology not only improves the accuracy and sensitivity of monitoring, but also enables the system to timely identify disaster signs that may be overlooked by a single sensor. This method not only improves the efficiency and accuracy of detection, but also reduces costs and maintenance difficulties, providing a solid guarantee for the safe production of mines. With the further development and improvement of technology, this method is expected to be popularized and applied in more mines, promoting the development of mining production towards a more intelligent and safer direction. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0051] Figure 1 It is the general flow chart of the intelligent monitoring method for roof water disasters in steeply inclined working faces of the embodiments of the present invention;

[0052] Figure 2 It is the schematic diagram of the layout of microseismic monitoring sensors of the embodiments of the present invention;

[0053] Figure 3 It is the flow chart of the seismic source location algorithm of the embodiments of the present invention;

[0054] Figure 4 It is the schematic diagram of the LSTM structure of the embodiments of the present invention;

[0055] Figure 5 It is the data analysis flow chart of the embodiments of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0056] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to 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.

[0057] In order to make the above objects, features, and advantages of the present invention more obvious and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0058] This embodiment provides an intelligent monitoring system for roof water disasters in steeply inclined working faces, including: a sensor layout module, a data acquisition module, a data transmission module, a data storage module, and a data analysis and early warning model;

[0059] The sensor layout module is used to layout a sensor network based on preset positions; wherein, the sensor network includes: microseismic sensors and acquisition instruments, water level gauges, water pressure gauges, and hygrometers;

[0060] The data acquisition module is used to acquire multi-source data underground through the sensor network based on a preset sampling frequency;

[0061] The data transmission module is used to transmit the acquired multi-source data by means of optical fiber wiring and wireless transmission; wherein, in areas where wiring is difficult or the cost is high, wireless transmission can be used;

[0062] The data storage module is used to store multi-source data;

[0063] The data analysis and early warning model is used to analyze and process multi-source data and issue early warnings.

[0064] Further, in the sensor layout module,

[0065] The microseismic sensors are fixed on the protruding anchor bolts in the coal wall, the microseismic sensors are connected by cables, and the acquisition instruments are suspended on the side wall of the roadway. Each acquisition instrument receives the monitoring data from 2 microseismic sensors; wherein, in high-risk areas, the acquisition instruments are densely arranged; the installation methods of the remaining sensors are selected according to different situations, such as wireless transmission and transmission through wifi, etc.;

[0066] The water level gauges are arranged at the lower part of the working face and near the floor fissures;

[0067] The water pressure gauges are arranged in the surrounding rock and areas with developed fissures;

[0068] The hygrometers are arranged in areas with poor air circulation in the working face.

[0069] Specifically, in this embodiment, the sensor layout is as follows:

[0070] For the microseismic sensors, highly sensitive microseismic sensors suitable for the mine environment are selected. These sensors need to have high resolution and high-frequency response capabilities to capture tiny vibration signals. According to the noise level and geological conditions of the steeply inclined working face, highly sensitive microseismic sensors are chosen to ensure that the sensors can accurately capture tiny vibration signals in a high-noise environment.

[0071] The microseismic monitoring system consists of two parts: underground and aboveground. Underground, acquisition instruments and sensors are arranged in the transportation roadway and the return airway. Each acquisition instrument receives monitoring data from 2 sensors. The sensors are fixed on the protruding bolts in the coal rib, and the acquisition instruments are suspended on the side wall of the roadway through hooks. The acquisition instruments and sensors are connected by cables to achieve real-time data acquisition and transmission. One microseismic acquisition instrument is arranged every 20 - 30 meters, and in high-risk areas, it can be encrypted to 10 - 15 meters. Aboveground, microseismic data collectors, recorders, and analyzers are arranged to collect, record, and analyze and save the microseismic data. The layout diagram of the microseismic monitoring system is as Figure 2 shown. Special installation brackets and fixing devices are used to ensure that the sensors are firmly installed and prevent displacement during mining operations. Check the environmental conditions at the installation location to ensure that there is no accumulated water and excessive dust around the sensors.

[0072] By recording the arrival time of microseismic events and combining with the spatial coordinates of the sensors, in this embodiment, the Geiger iterative method is used to calculate the location of the seismic source: The Geiger iterative method is a numerical method for solving non-linear equations and is commonly used in seismic source location problems. It gradually approaches the true location of the seismic source through iteration.

[0073] 1. Assume an initial seismic source location (x0, y0, z0) and the earthquake occurrence time t0;

[0074] 2. Calculate the theoretical arrival time:

[0075]

[0076] 3. Calculate the residual:

[0077] Δt i =t i -t i calc

[0078] 4. Through Taylor expansion, linearize the non-linear equations and solve for the correction amounts (Δx, Δy, Δz, Δt0) of the seismic source location and the earthquake occurrence time;

[0079] 5. Update the seismic source location:

[0080] x0 = x0 + Δx, y0 = y0 + Δy, z0 = z0 + Δz, t0 = t0 + Δt

[0081] Repeat steps 2 - 5 until the residual meets the accuracy requirement. The flowchart is as Figure 3 shown.

[0082] Select appropriate water level gauges, water pressure gauges, and humidity gauges according to the measurement accuracy and environmental adaptability. These sensors should be able to work stably in the mine environment with high humidity and a lot of dust.

[0083] The water level gauges should be installed in areas where water accumulation or seepage may occur, such as the lower part of the working face and near the floor fissures. The sensors should be installed at positions where they can directly contact the water body to ensure the accuracy of the measurement. The water pressure gauges should be installed in the surrounding rock and areas where water accumulation may occur, especially in the areas with developed fissures and possible water conduction. The sensors should be fixed firmly to avoid damage caused by rock deformation or pressure change. The humidity gauges should be installed in areas with poor air circulation in the working face and locations where leakage may occur. During installation, ensure that the sensors can accurately reflect the humidity change in the air and regularly clean the dust and water vapor around the sensors.

[0084] Specifically, in this embodiment, the data acquisition module collects multi-source data underground through the sensing network based on a preset sampling frequency, including:

[0085] According to the monitoring requirements and the working face environment, reasonably set the sampling frequency of the sensors. Set the sampling frequency of the microseismic sensors between 500 Hz and 2000 Hz to ensure capturing the subtle vibration signals in the working face. Dynamically adjust the sampling frequency according to the vibration intensity and the working face activities to ensure the data acquisition accuracy at critical moments.

[0086] Water level gauges, water pressure gauges, and humidity gauges: The sampling frequencies of these sensors should be set according to the actual on-site situation. Generally, the sampling frequency of the water level gauges and water pressure gauges can be set to once per minute. When significant changes are detected, the sampling frequency should be increased to once per 10 seconds. The humidity gauges can be set to sampling frequencies ranging from 10 seconds to 1 minute according to the change of air humidity.

[0087] The microseismic sensor records key parameters such as the source location, magnitude, and frequency of earthquakes. The data is converted into digital signals through the built-in ADC and stored in the local storage device of the sensor in real time to prevent data loss. The water level gauge records the water level changes in real time. Especially when the water level shows an obvious upward trend, the sensor will automatically increase the data recording frequency and store the data in the local storage device to ensure the integrity and continuity of the data. The water pressure gauge monitors the changes in the surrounding rock and water pressure in real time. After the data is converted into digital signals through the ADC, it is recorded in the sensor memory card in real time for subsequent analysis. The hygrometer records the changes in air humidity in real time. Especially when the humidity changes suddenly, the sensor will increase the data acquisition frequency and store the data in the local storage device to ensure data integrity.

[0088] Specifically, in this embodiment, the data transmission module uses fiber optic cabling to transmit the multi-source data collected; including:

[0089] In areas with high requirements for stability and safety in the mine, fiber optic cabling is used as the main data transmission method. Fiber optic has the advantages of anti-electromagnetic interference, long transmission distance, and large bandwidth, and is suitable for long-distance signal transmission. When cabling, the fiber optic line should be avoided passing through areas with frequent operation of mechanical equipment to avoid external force damage.

[0090] In areas with short transmission distances, shielded cables are used for data transmission. Shielded cables can effectively prevent electromagnetic interference and improve the integrity of data transmission. When laying cables, they should be protected through cable trays or pipes to avoid cable damage caused by mechanical wear or environmental changes (such as humidity and temperature changes).

[0091] Suitable for areas where cabling is difficult, especially for sensors that are temporarily installed or quickly installed in the mine. Wi-Fi transmission requires a repeater to be installed in places with good signal strength to ensure that the signal coverage reaches the requirements. Considering the complex terrain structure in the mine, the sensor should maintain a line of sight or a path with the least obstacles with the repeater to ensure the stability of signal transmission.

[0092] Used for long-distance and low-power data transmission, suitable for sensors with low requirements for transmission rate (such as hygrometers, water pressure gauges, etc.). These technologies can expand the signal coverage through multiple repeaters. Signal strength tests should be carried out for possible electromagnetic interference and obstacles in the mine, and the signal quality should be optimized by adjusting the position of the relay station and the antenna direction.

[0093] Relay stations and signal amplifiers are arranged at key positions in the mine. The relay station should be set in a place where it is not affected by electromagnetic interference and the environment, and should be maintained regularly to ensure the normal operation of the equipment. The signal amplifier should select an appropriate amplification factor according to the actual needs in the mine to avoid signal distortion or energy consumption caused by excessive amplification.

[0094] Furthermore, the data storage module adopts a distributed storage architecture and supports horizontal expansion of data and multi-copy storage.

[0095] The data storage module establishes multi-level indexes for multi-source data, including: time index, geographical location index, and sensor type index.

[0096] The data storage module adopts a redundant storage design.

[0097] Specifically, in this embodiment, data reception and storage include the following:

[0098] Receiving device configuration: The ground monitoring center or relay station should be configured with multi-channel and multi-protocol data receiving devices to support parallel reception of different sensor data. The receiving device should have a redundant design to ensure that other channels can continue to work when some channels fail.

[0099] Use a receiving device with high bandwidth and low latency characteristics to ensure that the data collected by the sensor can be transmitted to the monitoring center in real time. The receiving device should reduce the latency in data transmission and improve the effect of real-time monitoring through regular system diagnosis and network optimization. The receiving device should support multiple transmission protocols (such as TCP / IP, MQTT, etc.) to ensure data compatibility with different sensors. At the same time, the device should have an automatic protocol recognition function and can adapt to the data transmission of different sensors without manual switching.

[0100] Data storage: Select a large-capacity database system with high read and write performance, such as an SQL database, to meet the storage requirements of massive sensor data. The database system should have a distributed storage architecture and support horizontal expansion of data and multi-copy storage. To improve the query efficiency of data, multi-level indexes should be established for sensor data, including time index, geographical location index, sensor type index, etc. The database system should support efficient full-text retrieval and real-time data analysis to ensure quick access and processing of relevant data in case of emergencies. Set up an automated data backup system to support scheduled backup and off-site backup. The database system should adopt a redundant storage design to ensure that data can be restored from backups or other redundant nodes when a storage node fails. The backup data should be regularly verified to prevent recovery failures caused by data corruption or loss. Encrypt the stored data to prevent unauthorized access. Set up a fine-grained access control mechanism to ensure that only authorized personnel or systems can access specific data. The database system should have an auditing function to record all data access and operation logs for review and analysis.

[0101] Integrate the database system with the real-time data processing engine to achieve real-time analysis and processing of data. Through integration with the early warning system, the real-time data analysis results can directly trigger early warning events, shortening the response time. Fusion of historical data and real-time data: The data storage system should support the fusion analysis of historical data and real-time data, use historical data for model training and prediction, and compare and verify with real-time data to improve the accuracy of early warnings.

[0102] Furthermore, the data analysis and early warning model includes: a model construction sub-module and an analysis and early warning sub-module;

[0103] The model construction sub-module is used to train and adjust the long short-term memory network model through historical monitoring data to construct an early warning model;

[0104] The analysis and early warning sub-module is used to input real-time monitoring data into the early warning model to obtain prediction results; when the prediction results exceed the preset early warning threshold, an early warning is triggered; among them, the prediction results include: disaster type, occurrence area and time.

[0105] Furthermore, the model construction sub-module includes: a data analysis unit, a data set construction unit, and a model training unit;

[0106] The data analysis unit is used to perform data analysis on historical monitoring data to obtain the corresponding disaster type, occurrence area and time;

[0107] The data set construction unit is used to construct a data set based on historical monitoring data and the corresponding analysis results;

[0108] Divide the historical monitoring data into a training set (70%), a validation set (15%) and a test set (15%) in chronological order to ensure that the training set data time is earlier than the validation set and the test set, and avoid future data leakage.

[0109] The model training unit is used to train the long short-term memory network model using the data set, adjust the model parameters, and verify the model using the cross-validation method.

[0110] Furthermore, the data analysis unit's data analysis of historical monitoring data includes:

[0111] Preprocess the historical monitoring data;

[0112] Extract features from the preprocessed data to obtain time-domain features;

[0113] Based on the time-domain features, use the Fourier transform method to extract frequency-domain features;

[0114] Use the wavelet transform method on the frequency-domain features to extract time-frequency features; that is, extract characteristic data in the time and frequency dimensions;

[0115] Based on time-frequency features, perform pattern recognition, use the Geiger iterative method to determine the source location of microseisms, and identify potential disaster sources;

[0116] Adopt the support vector machine algorithm to classify microseismic events and distinguish normal microseisms from abnormal microseisms.

[0117] Specifically, in this embodiment, data analysis and early warning specifically include:

[0118] Clean the multi-source data collected, remove noise and outliers, and use the median filtering method for processing to improve data quality. The calculation formula is as follows

[0119] g(x,y) = med{f(x - m,y - n),(m,n∈w)}

[0120] Among them, f(x,y) represents the original data, and g(x,y) represents the processed data.

[0121] Perform normalization processing on the processed data, using the Min-Max scaling method. The calculation formula is as follows:

[0122]

[0123] Min-Max normalization converts the data into the range of [0,1].

[0124] Extract features from the data that has been processed again, extract time-domain features such as peak value, mean value, standard deviation, skewness, and kurtosis, etc., to reflect the distribution and change trend of the data. Use the Fourier transform method to extract frequency-domain features to reveal the periodicity and frequency components in the data. Use the wavelet transform method to extract time-frequency features to comprehensively reflect the time and frequency information of the data.

[0125]

[0126] The above formula is the Fourier formula. Among them, w represents frequency, and t represents time. It represents the function in the frequency domain as the integral of the time-domain function f(t).

[0127] Then perform the pattern recognition step. Use the source location method to determine the source location of microseisms and identify potential disaster sources. Adopt the support vector machine algorithm to classify microseismic events and distinguish normal microseisms from abnormal microseisms. The essence of the support vector machine (SVM) is to use the supervised learning method to solve the binary classification problem.

[0128] Finally, perform model training: collect historical microseismic data, including information such as source location, magnitude, frequency, time, etc. The data should include normal and abnormal events. Select the long short-term memory network (LSTM) model. The structural schematic diagram is as Figure 4As shown. The reason for choosing this model is the advantage of LSTM in dealing with time series data and long-term dependency relationships, making it an ideal choice for the monitoring and early warning of mine roof water disasters. By using the LSTM model, the accuracy and real-time performance of monitoring can be effectively improved, providing reliable disaster early warning and trend analysis, and providing a solid guarantee for the safe production of mines. The model is trained using training data, and the model parameters are adjusted to improve the prediction accuracy of the model. The cross-validation method is adopted to prevent overfitting.

[0129] The real-time monitoring data is input into the trained early warning model, and the model conducts real-time analysis and prediction on the input data. Then, the model outputs the prediction results, including information such as possible disaster types, occurrence areas, and times. According to historical data and expert experience, an early warning threshold is set. When the prediction result exceeds the set threshold, an early warning is triggered. When the model prediction result indicates the existence of potential disaster risks, the system automatically triggers an alarm to notify relevant personnel to take emergency measures. The alarm methods include audible and visual alarms, text message notifications, email notifications, etc. After receiving the alarm, relevant personnel take emergency measures in a timely manner to prevent disasters from occurring. The data analysis flow chart is as Figure 5 shown.

[0130] This embodiment proposes an intelligent monitoring method for roof water disasters that comprehensively utilizes microseismic monitoring, sensor networks, and artificial intelligence technology. This method realizes the scientific and precise prevention and control of roof water disasters by real-time monitoring of microseismic activities, combining sensor network data, and using artificial intelligence algorithms for data analysis and early warning.

[0131] In practical applications, this method significantly improves the accuracy and real-time performance of monitoring and greatly reduces the need for manual intervention. By combining sensors and microseismic monitoring equipment, data can be collected and transmitted in real-time, and artificial intelligence technology conducts in-depth analysis and prediction on this data, so as to timely discover potential risks and issue early warnings. This not only improves work efficiency but also greatly reduces production risks.

[0132] In addition, this comprehensive monitoring method also shows significant advantages in equipment maintenance. Traditional monitoring equipment usually requires frequent manual inspections and maintenance, while the new method maintains the stability and accuracy of the system through sensor and algorithm updates, thereby reducing the complexity and frequency of maintenance. This intelligent monitoring and maintenance method helps to achieve comprehensive monitoring and effective prevention and control of mine roof water disasters and ensure the safe production of steeply inclined working faces.

[0133] In this embodiment, the use of microseismic monitoring, sensor networks, and artificial intelligence technology to monitor and warn of roof water disasters in steeply inclined working faces is a technology with great potential. It is not only stable and reliable but also easy to operate. Although the current research is only preliminary, with further improvement and perfection, this invention is expected to play an important role in the field of mine safety monitoring. This method of roof water disaster monitoring based on multiple advanced technologies not only improves the efficiency and accuracy of monitoring but also reduces labor costs and maintenance difficulties, providing a practical new approach for the intelligent construction of mines. With the further development and perfection of technology, this method is expected to be popularized and applied in more mines, promoting the development of mining production towards a more intelligent and safe direction.

[0134] As Figure 1 shown, this embodiment also proposes an intelligent monitoring method for roof water disasters in steeply inclined working faces, including:

[0135] Collecting real-time monitoring data underground based on a preset sampling frequency; among them, the real-time monitoring data includes: microseismic sensing data, water level data, water pressure data, and humidity data;

[0136] Inputting the real-time monitoring data into the warning model to obtain the prediction result; among them, the prediction result includes: disaster type, occurrence area, and time; the warning model is obtained by training and adjusting the long short-term memory network model through historical monitoring data;

[0137] When the prediction result exceeds the preset warning threshold, a warning is triggered.

[0138] Furthermore, the microseismic sensing data includes: source location, magnitude, and frequency;

[0139] Collecting the source location includes: recording the arrival time of microseismic events, combining the spatial coordinates of the sensors, and using the Geiger iterative method to calculate the source location.

[0140] Furthermore, training the long short-term memory network model through historical monitoring data includes:

[0141] Performing data analysis on historical monitoring data to obtain the corresponding disaster type, occurrence area, and time;

[0142] Based on historical monitoring data and the corresponding analysis results, constructing a data set;

[0143] Using the data set to train the long short-term memory network model, adjusting the model parameters, and using the cross-validation method to verify the model.

[0144] Furthermore, performing data analysis on historical monitoring data includes:

[0145] Performing preprocessing on historical monitoring data;

[0146] Perform feature extraction on the preprocessed data to obtain time domain features;

[0147] Based on the time domain features, the Fourier transform method is used to extract the frequency domain features;

[0148] The frequency domain features are used to extract time-frequency features using wavelet transform method;

[0149] Based on time-frequency characteristics, pattern recognition is performed and source location methods are used to determine the source of microseismic events and identify potential disaster sources.

[0150] The support vector machine algorithm is used to classify microseismic events and distinguish normal microseismic events from abnormal microseismic events.

[0151] Furthermore, preprocessing of historical monitoring data includes:

[0152] The historical monitoring data is cleaned using the median filtering method;

[0153] The cleaned data were normalized using the Min-Max scaling method.

[0154] Specifically, in this embodiment, sensors are arranged including microseismic sensors and water level gauges, water pressure gauges, and hygrometers to form a sensor network. High-sensitivity microseismic sensors are selected, which are suitable for the high noise level of the steeply inclined working face environment. The purpose of selecting the water level gauge is to monitor the water level changes in the working face, especially in areas with higher water damage risks. The water pressure gauge is used to monitor the changes in water pressure in the working face, especially in the surrounding rock or areas where water accumulation may exist. The hygrometer is used to monitor the changes in air humidity in the working face to evaluate the water seepage of the surrounding rock. Sensors are arranged in key areas of the roof, floor and surrounding rock of the steeply inclined working face, especially in areas where there may be hidden dangers of water damage. Microseismic sensors are organically combined with water level gauges, water pressure gauges, hygrometers and other sensors to form a comprehensive sensor network. This integrated design allows multiple data sources to provide information simultaneously, providing a multi-dimensional monitoring method for the overall stability of the working face. The sensors should be firmly installed to avoid vibration or movement affecting the data accuracy.

[0155] Real-time data acquisition: Set the sampling frequency of the sensor, typically from hundreds of hertz to thousands of hertz, to ensure that even small vibrations are captured. Microseismic sensors record microseismic activity in real time, including parameters such as source location, magnitude, and frequency. Water level gauges are used to monitor water level changes within the working face in real time. Water pressure gauges measure water pressure changes in the surrounding rock and areas of potential water accumulation in real time. Hygrometers monitor air humidity in real time. The data from these different sensors is converted into digital signals using the sensor's built-in ADC (analog-to-digital converter). Real-time data is stored locally using the sensor's built-in storage device (such as a built-in memory card) to ensure data integrity.

[0156] Data transmission method: In scenarios with high requirements for stability and data integrity, wired transmission methods such as optical fibers and cables are preferred. The sensors are connected to the ground monitoring center or relay station by wire. In areas where wiring is difficult or costly, wireless transmission methods can be adopted.

[0157] Data reception and storage: The ground monitoring center or relay station is equipped with data reception equipment to receive data from sensors in real time. The reception equipment should have the characteristics of high bandwidth and low latency. The received data should be stored in the database in a timely manner, and the database should have efficient read and write performance and large-capacity storage capabilities.

[0158] Data analysis and early warning: Clean, normalize, and extract features from the multi-source data collected. Extract features through time domain, frequency domain, and time-frequency analysis, and use machine learning algorithms for event classification and trend analysis. Based on historical data and expert experience, train an early warning model to achieve real-time prediction and early warning. When the prediction result exceeds the set threshold, an alarm is automatically triggered to notify relevant personnel to take emergency measures.

[0159] Advantages and characteristics:

[0160] 1. High degree of automation

[0161] Reduce manual intervention: The entire monitoring process does not require a large amount of manpower for on-site measurement, reducing labor costs and labor intensity. The system can operate automatically to achieve all-weather real-time monitoring.

[0162] Efficient data processing: Utilize sensor networks and artificial intelligence technologies to automatically collect and process data, significantly improving work efficiency and response speed.

[0163] 2. High precision and high reliability

[0164] Accurate monitoring: Through highly sensitive microseismic sensors and advanced data analysis algorithms, it is able to accurately identify and locate microseismic events and changes in roof water disasters.

[0165] Multi-dimensional feature analysis: Combine time domain, frequency domain, and time-frequency analysis technologies to comprehensively capture data features, ensuring the reliability and accuracy of monitoring results.

[0166] 3. Strong real-time performance

[0167] Real-time data collection and transmission: The sensor network collects environmental data in real time and transmits it to the ground monitoring center in real time by wire or wireless means to ensure the real-time nature of the data.

[0168] Instant early warning: Use artificial intelligence technology to analyze and predict real-time data. The system can detect potential risks and issue early warnings in the first time, providing sufficient response time.

[0169] 4. Easy to maintain

[0170] The location selection of sensors is very important. Sensors should be installed in easily accessible areas so that when maintenance or replacement is required, staff can reach these positions conveniently and quickly. For example, in steeply inclined working faces, sensors should be arranged in areas that do not require additional support or climbing, avoiding complex operation steps. Reasonable layout should take into account the normal operations within the working face and should not interfere with the operation of mining equipment or the signal reception of other sensors. Sensors arranged on non-critical equipment paths are avoided from being interfered due to equipment movement or operations. Sensors are installed in relatively dry, dust-free or less dusty areas, reducing the frequent maintenance requirements caused by environmental impacts. For example, avoid directly installing sensors in low-lying areas with concentrated water flow or areas prone to water accumulation, but choose to install them in relatively high and dry places.

[0171] Sensors and camera devices adopt modular design, that is, each sensor unit can be independently disassembled and replaced. When a certain sensor fails, maintenance personnel do not need to replace the entire system, but only need to replace the faulty unit. This not only improves the maintenance efficiency but also reduces the replacement cost. All sensors should preferably adopt unified interfaces and connection methods to simplify the connection operations during replacement. Standardized interface design reduces the risk of incorrect connection during equipment replacement and enables good compatibility between different models or batches of sensors.

[0172] Quick installation: Reasonable installation brackets and fixing devices enable the installation and replacement processes of sensors and camera devices to be completed quickly. The installation positions should be designed to be operable with common tools and have sufficient space for connection and fixing. For example, sensors can be installed through snap-on brackets, so that maintenance personnel can complete installation or disassembly with simple pushing and pulling actions.

[0173] Clearly marked: Each sensor installation position should have clear markings, and the corresponding sensor numbers and position information should be recorded in the system. In this way, during maintenance, maintenance personnel can quickly locate the sensors that need to be replaced, avoiding long-term searching and misoperations.

[0174] Easy algorithm update: Through software updates, the algorithm performance and accuracy can be continuously improved, ensuring the long-term stable operation of the system and reducing the maintenance cost.

[0175] 5. Strong comprehensiveness

[0176] Multi-source data fusion: Comprehensive utilization of microseismic monitoring, sensor networks and artificial intelligence technologies to achieve the fusion and comprehensive analysis of multi-source data, improving the overall monitoring effect.

[0177] Comprehensive Risk Assessment: By combining historical data and real-time data, the system can conduct comprehensive risk assessments, providing detailed risk reports and decision-making support.

[0178] 6. High Adaptability

[0179] Adapt to Complex Environments: The system design can adapt to the complex geological conditions and working environments of steeply inclined working faces, ensuring the stability and accuracy of monitoring. The layout of the sensor network and monitoring equipment is flexible and can be adjusted and optimized according to the specific conditions of the mine, being applicable to different types of mine environments.

[0180] In summary, through the introduction of advanced monitoring and analysis technologies, the present invention realizes efficient, precise, and automated monitoring and early warning of roof water disasters in steeply inclined working faces. The present invention does not rely on a single type of sensor, but integrates multiple sensors such as microseismic sensors, water level gauges, water pressure gauges, and hygrometers. This cross-field sensor integration design can obtain key environmental parameters such as vibrations, pressures, water levels, and humidity within the working face in real time, providing multi-dimensional data support for the monitoring of roof water disasters. Through advanced data fusion algorithms, the present invention can integrate data from different sensors and extract comprehensive information with higher information content and higher signal-to-noise ratio. This data fusion technology not only improves the accuracy and sensitivity of monitoring, but also enables the system to promptly identify disaster signs that may be overlooked by a single sensor. This method not only improves the efficiency and accuracy of detection, but also reduces costs and maintenance difficulties, providing a solid guarantee for the safe production of mines. With the further development and improvement of technology, this method is expected to be popularized and applied in more mines, promoting the development of mining production towards a more intelligent and safer direction.

[0181] The embodiments described above are only descriptions of the preferred embodiments of the present invention and do not limit the scope of the present invention. Without departing from the design spirit of the present invention, various deformations and improvements made by those of ordinary skill in the art to the technical solutions of the present invention shall fall within the protection scope determined by the claims of the present invention.

Claims

1. An intelligent monitoring method for roof water disasters in steeply inclined working faces, characterized in that, Including: Collect real-time monitoring data underground based on a preset sampling frequency; wherein, the real-time monitoring data includes: microseismic sensing data, water level data, water pressure data, and humidity data; Input the real-time monitoring data into an early warning model to obtain a prediction result; wherein, the prediction result includes: disaster type, occurrence area, and time; the early warning model is obtained by training and adjusting a long short-term memory network model with historical monitoring data; When the prediction result exceeds a preset early warning threshold, trigger an early warning.

2. The intelligent monitoring method for roof water disasters in steeply inclined working faces according to claim 1, wherein, The microseismic sensing data includes: seismic source location, magnitude, and frequency; Collecting the seismic source location includes: recording the arrival time of microseismic events, combining with the spatial coordinates of sensors, and using the Geiger iterative method to calculate the seismic source location.

3. The intelligent monitoring method for roof water disasters in steeply inclined working faces according to claim 1, characterized in that Training the long short-term memory network model with historical monitoring data includes: Conduct data analysis on the historical monitoring data to obtain the corresponding disaster type, occurrence area, and time; Based on the historical monitoring data and the corresponding analysis results, construct a data set; Use the data set to train the long short-term memory network model, adjust the model parameters, and verify the model using the cross-validation method.

4. The intelligent monitoring method for roof water disasters in steeply inclined working faces according to claim 3, characterized in that, Conducting data analysis on the historical monitoring data includes: Preprocess the historical monitoring data; Extract features from the preprocessed data to obtain time-domain features; Based on the time-domain features, use the Fourier transform method to extract frequency-domain features; Use the wavelet transform method for the frequency-domain features to extract time-frequency features; Based on the time-frequency features, conduct pattern recognition, use the Geiger iterative method to determine the seismic source location of microseisms, and identify potential disaster sources; Adopt the support vector machine algorithm to classify microseismic events and distinguish normal microseisms from abnormal microseisms.

5. The intelligent monitoring method for roof water disasters in steeply inclined working faces according to claim 4, characterized in that, Preprocessing the historical monitoring data includes: Clean the historical monitoring data using the median filtering method; Use the Min-Max scaling method to normalize the cleaned data.

6. An intelligent monitoring system for roof water disasters in steeply inclined working faces, characterized in that, For implementing the intelligent monitoring method for water disasters on the roof of steeply inclined working faces as described in any one of claims 1-5; the system includes: a sensor layout module, a data collection module, a data transmission module, a data storage module, a data analysis and early warning model; The sensor layout module is used to layout a sensor network based on a preset position; wherein, the sensor network includes: microseismic sensors and acquisition instruments, water level gauges, water pressure gauges, and humidity gauges; The data collection module is used to collect multi-source data underground through the sensor network based on a preset sampling frequency; The data transmission module is used to transmit the collected multi-source data by means of fiber optic cabling and wireless transmission; The data storage module is used to store the multi-source data; The data analysis and early warning model is used to analyze and process the multi-source data and issue an early warning.

7. The intelligent monitoring system for roof water disasters in steeply inclined working faces according to claim 6, characterized in that In the sensor layout module, The microseismic sensors are fixed on the protruding bolts in the coal rib, the microseismic sensors are connected by cables, the acquisition instruments are suspended on the side rib of the roadway, and each acquisition instrument receives the monitoring data from 2 microseismic sensors; wherein, the acquisition instruments are densely laid out in high-risk areas; The water level gauge is arranged near the lower part of the working face and the floor fissures; The water pressure gauge is arranged in the surrounding rock and the area where fissures develop; The humidity gauge is arranged in the area where the air circulation in the working face is poor.

8. The intelligent monitoring system for roof water disasters in steeply inclined working faces according to claim 6, characterized in that The data storage module adopts a distributed storage architecture, supporting horizontal expansion of data and multi-copy storage; The data storage module establishes multi-level indexes for multi-source data, including: time index, geographical location index, and sensor type index; The data storage module adopts a redundant storage design.

9. The intelligent monitoring system for roof water disasters in steeply inclined working faces according to claim 6, characterized in that, The data analysis and early warning model includes: a model construction sub-module and an analysis and early warning sub-module; The model construction sub-module is used to train and adjust the long short-term memory network model through historical monitoring data to construct an early warning model; The analysis and early warning sub-module is used to input real-time monitoring data into the early warning model to obtain a prediction result; when the prediction result exceeds a preset early warning threshold, an early warning is triggered; wherein, the prediction result includes: disaster type, occurrence area, and time.

10. The intelligent monitoring system for roof water disasters in steeply inclined working faces according to claim 9, characterized in that, The model construction sub-module includes: a data analysis unit, a data set construction unit, and a model training unit; The data analysis unit is used to perform data analysis on the historical monitoring data to obtain the corresponding disaster type, occurrence area, and time; The data set construction unit is used to construct a data set based on the historical monitoring data and the corresponding analysis results; The model training unit is used to train the long short-term memory network model using the data set, adjust the model parameters, and verify the model using the cross-validation method.

Images