Monitoring system for underground surrounding rock deformation

Abstract

The invention discloses a monitoring system for underground surrounding rock deformation, and relates to the technical field of mine safety monitoring, the monitoring system comprises an acquisition module, an analysis module and a decision module, the acquisition module is used for acquiring three-dimensional space geometric data and real-time data, establishing codes and dynamically expanding a sample library; the analysis module establishes a grading early warning system and a visual model based on the sample library, and analyzes a deformation trend through machine learning; and the decision module predicts the deformation development trend and rate and triggers graded early warning and maintenance measures. Through multi-sensor data fusion, an intelligent algorithm and visual display, real-time and accurate monitoring and prospective early warning of underground surrounding rock deformation are realized, the automation and intelligent level of mine safety production is effectively improved, and accident risks and manual intervention requirements are reduced. The system has the advantages of fast response, high precision, strong self-adaption and the like.

Description

Technical Field

[0001] This invention relates to the field of coal mine safety monitoring technology, and in particular to a monitoring system for underground surrounding rock deformation. Background Technology

[0002] In the mining and construction of underground projects such as coal mines, the stability of the surrounding rock in underground roadways is directly related to the safety of workers and the smooth progress of production. The surrounding rock will deform under the influence of mining stress, geological structure and other factors. If it is not monitored and warned in time, it may lead to serious accidents such as roadway collapse and roof fall.

[0003] Traditional methods for monitoring surrounding rock largely rely on manual, periodic measurements using tools such as convergence meters and delamination meters. This approach is not only inefficient and produces discrete data, but also suffers from monitoring blind spots, making it difficult to capture the dynamic process of surrounding rock deformation in a timely manner. Furthermore, some existing automated monitoring systems often focus on monitoring single physical quantities, such as displacement or stress, lacking the fusion of multi-source information. They are unable to construct a comprehensive three-dimensional spatial picture of surrounding rock deformation, resulting in limited early warning methods and a lack of predictive capabilities based on historical data, leading to insufficient accuracy and foresight in early warning systems.

[0004] Currently, Chinese invention patent application number CN202510625829.9 discloses a method for monitoring the condition of surrounding rock and providing early warning of disasters in underground coal mine roadways. This method uses a self-organizing network of seismic sensors to construct a polygonal monitoring network to perceive surrounding rock deformation in real time. Combined with multi-point delamination displacement sensors on anchor bolts, it captures fracture development characteristics and triggers millimeter-wave radar to dynamically scan abnormal areas. A data fusion algorithm is used to analyze deformation gradients and strain tensors, achieving precise location of disasters and prediction of their evolution trends. Its technical advantage lies in overcoming the limitations of single monitoring technologies, simultaneously improving the accuracy of disaster location and the timeliness of early warning, and providing multi-dimensional decision support for assessing the stability of surrounding rock in deep roadways.

[0005] The aforementioned technologies are insufficient to assess the severity of warnings, and each warning may result in a waste of resources. The prediction results are not verified or adjusted, and corresponding warning methods are not provided for different situations. It is also difficult to distinguish the severity of the surrounding rock conditions. Furthermore, the signal in coal mines is poor, data transmission is slow, and data loss is common. Summary of the Invention

[0006] The technical problem solved by this invention is that existing technologies are difficult to judge the danger of early warnings, each early warning may cause waste of resources, the prediction results are not verified and adjusted, and corresponding early warning methods are not provided for the corresponding early warning situations. It is not easy to distinguish the severity of the surrounding rock conditions, and the signal in coal mines is poor, the data transmission is slow and easy to be lost.

[0007] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a monitoring system for downhole surrounding rock deformation, comprising a data acquisition module, an analysis module, and a decision-making module:

[0008] The acquisition module is used to acquire three-dimensional spatial geometric data, obtain a three-dimensional model, acquire real-time data through different sensor nodes, establish codes based on the real-time data, and add samples to the database based on the codes.

[0009] The analysis module is used to obtain historical sample data from the sample library, establish a tiered early warning system based on the historical sample data, and obtain a visualization model through a 3D model based on the tiered early warning system and real-time data.

[0010] The decision-making module is used to take the sample library and the hierarchical early warning system as the first instruction to obtain the deformation development trend and rate, take the real-time data, deformation development trend and rate as the second instruction to obtain the prediction result, retrieve the corresponding prediction result through the code, obtain the early warning level and location based on the prediction result, and carry out maintenance and alarm.

[0011] As a preferred embodiment of the monitoring system for downhole surrounding rock deformation according to the present invention, the three-dimensional model specifically includes:

[0012] The three-dimensional spatial geometric data includes tunnel cross-sectional contour data, surrounding rock surface morphology data, and borehole observation data;

[0013] The tunnel cross-sectional profile data includes the initial shape and size of the tunnel; the surrounding rock surface morphology data includes three-dimensional coordinate information of the undulations and fissure distribution of the surrounding rock surface after tunnel excavation; and the borehole inspection data includes three-dimensional spatial information of the internal structure of the surrounding rock.

[0014] The cross-sectional contour data of the tunnel was obtained using a total station and a 3D laser scanner;

[0015] The borehole observation data is obtained through borehole cameras and radar;

[0016] A three-dimensional model is obtained based on the three-dimensional spatial geometric data.

[0017] As a preferred embodiment of the monitoring system for downhole surrounding rock deformation according to the present invention, the real-time data specifically includes:

[0018] The real-time data includes displacement data, stress data, and microseismic data. The displacement data includes the relative convergence of the roof and floor of the underground coal mine roadway, the convergence displacement of the two sides, and the three-dimensional coordinate changes of specific points on the roadway surface.

[0019] The stress data includes the internal stress of the surrounding rock, the stress of the anchor bolts, the stress of the anchor cables, and the working resistance of the hydraulic supports.

[0020] The microseismic data includes the energy, location, and timing of microseismic events generated by rock mass fracturing.

[0021] The displacement data is acquired using a displacement sensor;

[0022] The stress data is acquired using a stress sensor;

[0023] The microseismic data is acquired using a microseismic sensor.

[0024] As a preferred embodiment of the monitoring system for downhole surrounding rock deformation according to the present invention, the step of establishing a code and adding a sample library based on the code specifically includes:

[0025] By setting up underground nodes and cloud nodes through a communication bus, the real-time data is transmitted to the underground nodes and matched with the historical data range;

[0026] If the real-time data is within the range of the historical data, then the encoding of the historical data is transmitted to the cloud node;

[0027] If the real-time data is not within the range of the historical data, a new code is established based on the real-time data, the real-time data is expanded according to a preset expansion range, the range of historical data is obtained, and it is saved to the underground node. Then, the real-time data and the new code are transmitted to the cloud node. The cloud node expands the sample library according to the real-time data and the preset expansion range, and the sample library includes sample historical data, codes and prediction results.

[0028] The historical data range is defined by taking real-time data as the center, using a preset expansion range as the interval radius, determining an expanded numerical interval, expanding the real-time data, establishing sample codes for all virtual data points within the expanded numerical interval, and treating the virtual data points as acquired real-time data.

[0029] The historical data of the sample consists of all virtual data points within the expanded numerical range, and these virtual data points are sorted according to time.

[0030] As a preferred embodiment of the monitoring system for downhole surrounding rock deformation described in this invention, the graded early warning system specifically includes:

[0031] Based on the sample library, historical sample data is obtained; based on the historical sample data, data ranges are divided; based on the data ranges, warning levels are set; the data ranges include a first range, a second range, and a third range.

[0032] The warning level is set based on the real-time data of the degree of deformation in the historical data of the sample. The data range is the maximum and minimum value of the real-time data of normal deformation, the maximum and minimum value of the real-time data of deformation that needs to be checked, and the maximum and minimum value of the real-time data of deformation that needs to be repaired.

[0033] The warning level corresponding to the first range is a primary warning, the warning level corresponding to the second range is a medium warning, and the warning level corresponding to the third range is a high warning.

[0034] As a preferred embodiment of the monitoring system for downhole surrounding rock deformation according to the present invention, the visualization model specifically includes:

[0035] The warning levels are displayed in different colors at corresponding positions on the 3D model to obtain a visualization model. The primary warning corresponds to the first color, the intermediate warning corresponds to the second color, and the advanced warning corresponds to the third color. The corresponding position is a color-coded position that corresponds to a specific area of ​​the collected real-time data.

[0036] As a preferred embodiment of the monitoring system for downhole surrounding rock deformation according to the present invention, the analysis module specifically includes:

[0037] The sample library and the hierarchical early warning system are used as the first instruction and input into the machine learning model to perform regression analysis and time series analysis to obtain the deformation development trend and rate. Real-time data, deformation development trend and rate are used as the second instruction and input into the machine learning model to obtain the prediction results. The prediction results are theoretical real-time data, risk level and location information.

[0038] Based on the prediction results, the corresponding warning level is determined by the color displayed by the visualization model, the corresponding location information is obtained based on the position of the color display, and the corresponding warning method is triggered based on the warning level.

[0039] As a preferred embodiment of the monitoring system for downhole surrounding rock deformation according to the present invention, the corresponding early warning method specifically includes:

[0040] The warning method corresponding to the primary warning is to display the real-time data through the monitoring center and send a message that everything is normal and the real-time data to designated personnel.

[0041] The intermediate-level early warning method is to trigger an audible and visual alarm, and link the downhole broadcasting system to release data anomaly messages and display location information.

[0042] The advanced early warning system uses a warning method that triggers an audible and visual alarm, links the underground broadcasting system to issue evacuation instructions, automatically locks the equipment power supply, and displays location information.

[0043] As a preferred embodiment of the monitoring system for downhole surrounding rock deformation according to the present invention, the decision module specifically includes:

[0044] The difference between the theoretical real-time data and the real-time data is obtained. If the difference is greater than or equal to the first threshold and less than or equal to the second threshold, the theoretical real-time data is used as the real-time data to obtain a new prediction result.

[0045] If the difference is less than the first threshold or greater than the second threshold, the theoretical real-time data is discarded, and a new prediction result is obtained based on the real-time data, deformation development trend and rate.

[0046] As a preferred embodiment of the monitoring system for downhole surrounding rock deformation according to the present invention, the decision module further includes:

[0047] The tiered early warning system is updated based on the newly added historical sample data, and the corresponding data range is adjusted based on the updated tiered early warning system.

[0048] Once the historical data range of the sample library covers all expected working conditions of downhole surrounding rock deformation, real-time data is acquired, and the code corresponding to the real-time data is sent to the cloud node. The cloud node directly retrieves the corresponding prediction result based on the code.

[0049] The beneficial effects of this invention are as follows: By employing multi-sensor fusion technology, three-dimensional spatial geometric data, including tunnel cross-sectional contour data, surrounding rock surface morphology data, and borehole inspection data, as well as real-time data such as displacement, stress, and microseismic activity, are collected. This ensures the comprehensiveness and accuracy of data acquisition, providing a reliable foundation for constructing a high-precision three-dimensional model, thus more realistically reflecting the deformation state of the surrounding rock. Through the establishment of a coding mechanism and a dynamic sample library, efficient data management and expansion are achieved. The sample library is intelligently expanded based on historical data ranges, covering various expected working conditions, improving the system's adaptability and robustness, and reducing the risk of misjudgment due to data gaps. The analysis module establishes a graded early warning system based on the sample library, dividing early warning levels into primary, intermediate, and advanced levels, which are intuitively displayed in the visualization model through color coding, enabling managers to quickly identify risk areas and levels. This system enhances the intuitiveness and operability of monitoring, facilitating timely responses. The decision-making module utilizes machine learning models for regression and time series analysis to predict deformation trends and rates, generating forecast results based on real-time data. This enables proactive early warning and automatically triggers corresponding warning mechanisms, such as audible and visual alarms, broadcast notifications, and equipment interlocking, ensuring timely and effective emergency responses. Finally, it possesses self-optimization capabilities, dynamically adjusting the prediction model and updating the tiered early warning system by comparing the difference between theoretical and actual data, thus guaranteeing the accuracy of long-term monitoring. This not only improves the precision and efficiency of underground rock deformation monitoring but also reduces labor costs, enhances the automation and intelligence level of mine safety production, and provides reliable technical support for accident prevention. Furthermore, through visualization and intelligent decision-making, it improves the overall level of safety management. Attached Figure Description

[0050] Figure 1 This is a basic flowchart of a monitoring system for downhole rock deformation, provided as an embodiment of the present invention. Detailed Implementation

[0051] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0052] This invention provides a monitoring system for underground rock deformation. The system can be applied to underground safety monitoring platforms in coal mines, metal mines, and other similar facilities, and is used for real-time monitoring and early warning of roadway rock deformation. It is suitable for safety monitoring scenarios in underground engineering such as mining and tunnel construction. Addressing the technical problems of incomplete underground rock deformation monitoring data, delayed early warning, and insufficient decision support in traditional technologies, the inventors propose this underground rock deformation monitoring system. The core idea of ​​this invention is to collect three-dimensional spatial geometric data and real-time data through multiple sensors, establish a coding system and sample library, construct a hierarchical early warning system and visualization model based on the sample library, and use machine learning to analyze and predict deformation trends to achieve accurate early warning and automated response, thereby improving the safety and reliability of underground operations.

[0053] Example 1, referring to Figure 1 , Figure 1 This is a schematic diagram of a module for monitoring downhole rock deformation, provided as an embodiment of the present invention. Figure 1 As shown, in this embodiment, the system includes, but is not limited to, a data acquisition module, an analysis module, and a decision-making module:

[0054] In one embodiment, the acquisition module is used to acquire three-dimensional spatial geometric data, obtain a three-dimensional model, acquire real-time data through different sensor nodes, establish codes based on the real-time data, and add samples to the database based on the codes;

[0055] The data acquisition module provides raw input for surrounding rock deformation monitoring through multi-source data acquisition. It acquires tunnel cross-sectional contour data using a total station and a 3D laser scanner.

[0056] The roadway cross-sectional contour data acquisition unit is based on the principle of geometric measurement. Using a typical coal mine roadway as a prototype, a three-dimensional geometric model of the roadway is established. The key parameters of the roadway are defined as an arched cross-section with a width of 4.5 meters and a height of 3.2 meters. The coordinates of key points in the roadway, such as control points on the roof, floor, and sides, are measured using a Leica TS60 total station with a measurement accuracy of ±1 mm and a scanning rate of 1000 points / second. Initial shape and dimension data are acquired. High-density scanning is then performed using a 3D laser scanner, such as a Faro Focus S350, to generate point cloud data. The point cloud density is 1000 points per square meter, with a point cloud accuracy of ±2 mm. The scanning range covers the entire 100-meter length of the roadway. The output roadway cross-sectional contour data includes the roadway contour line, cross-sectional area, and geometric dimensions. During implementation, point cloud processing software, such as CloudCompare, is used for data registration and 3D reconstruction to ensure that the model is consistent with the actual roadway.

[0057] Data on borehole observation is obtained through borehole cameras and radar;

[0058] The borehole observation data acquisition unit, based on optical and electromagnetic wave detection principles, drills a 50 mm diameter, 5 m deep borehole inside the surrounding rock of the tunnel. Using borehole imaging equipment, such as an endoscope camera, it captures images and videos of the surrounding rock interior at a resolution of 1920×1080 and a frame rate of 30 fps, identifying fissures and cavities. A ground-penetrating radar, such as the GSSI SIR-4000, emits electromagnetic waves at a frequency of 100 MHz to detect the internal structure of the surrounding rock, acquiring three-dimensional spatial information such as fissure location, depth, and orientation. The output borehole observation data includes images of the internal structure, radar waveforms, and three-dimensional coordinates. In practice, a data fusion algorithm integrates the optical and radar data to generate a model of the internal structure of the surrounding rock.

[0059] Real-time data is acquired through displacement sensors, stress sensors, and microseismic sensors;

[0060] The real-time data acquisition unit deploys various sensor nodes, including displacement sensors, such as laser displacement sensors installed on the roof, floor, and sidewalls of the roadway to measure the relative approach of the roof and floor and the convergence displacement of the sidewalls, with an accuracy of ±0.1 mm and a sampling frequency of 10 Hz; stress sensors, such as resistance strain gauges embedded in the surrounding rock or installed on anchor bolts and cables, to measure the internal stress of the surrounding rock, anchor bolt stress, and anchor cable stress, with a range of 0-50 MPa and an accuracy of ±0.5%; and micro-vibration sensors, such as accelerometers arranged around the roadway to capture micro-vibration events, with an energy detection range of 10⁻⁻⁶. 6 Up to 10⁻³J, with a positioning accuracy of ±1 meter; output real-time data includes displacement data, such as changes in three-dimensional coordinates, stress data, such as the working resistance of hydraulic supports, and micro-vibration data, such as event energy and occurrence time; in practice, data is transmitted to underground nodes via wireless sensor networks, such as the ZigBee protocol, to ensure real-time performance and stability.

[0061] The acquisition module is also used to establish codes and expand the sample library;

[0062] The coding and sample library management unit uses a communication bus, such as an industrial Ethernet, to set up underground and cloud nodes. The underground nodes are deployed underground and are responsible for temporary data storage, while the cloud nodes are located on a surface server. After real-time data is transmitted to the underground nodes, it is matched with historical data ranges. The historical data range is a pre-defined expansion range of previous real-time data, for example, a numerical range generated with a radius of ±5% centered on the real-time data value. If the real-time data is within the historical data range, the system transmits the coding of that historical data to the cloud nodes. If the real-time data exceeds the historical data range, the system generates new coding and expands it according to the pre-defined expansion range, using the real-time data as the center, to generate a new historical data range, which is saved to the underground nodes. At the same time, the real-time data and the new coding are transmitted to the cloud nodes. The cloud nodes expand the sample library according to the pre-defined expansion range based on the real-time data, adding to the sample library. The sample library includes historical sample data, coding, and prediction results. The historical sample data consists of all virtual data points within the expanded numerical range, which are sorted by time to simulate a historical sequence. In practice, the sample library is stored and updated through a database management system, such as MySQL, to ensure data traceability and expansion.

[0063] In this embodiment of the invention, based on the principle of multi-source data fusion, the deformation state of the surrounding rock is comprehensively captured through the collaborative acquisition of three-dimensional geometric data and real-time data. The coding and sample library management utilize data mining and dynamic expansion technologies to compensate for data gaps and changes in working conditions. The three-dimensional model serves as the foundation, real-time data as the input, and the sample library as the support, forming a unified data foundation. Through the integration of total station, scanner, and sensor, the acquisition module accurately describes the changes in roadway geometry and internal structure, especially the real-time evolution of displacement and stress. This enhances the reliability of monitoring because multi-sensor data can cross-verify deformation information, thereby providing accurate input for subsequent analysis and reducing the risk of false alarms.

[0064] In one embodiment, the analysis module is used to obtain historical sample data from the sample library, establish a hierarchical early warning system, and obtain a visualization model through a three-dimensional model based on the hierarchical early warning system and real-time data.

[0065] The analysis module extracts data from the sample library and builds early warning and visualization outputs.

[0066] Used to establish a tiered early warning system;

[0067] The hierarchical early warning system construction unit divides the data range based on historical sample data in the sample database. This historical data includes time-series information on displacement, stress, and microseismic parameters. The data range is divided into three levels based on the degree of deformation: the first range corresponds to real-time data with normal deformation, such as displacement changes of 0-10 mm and stress changes of 0-5 MPa; the second range corresponds to real-time data requiring inspection, such as displacement changes of 10-20 mm and stress changes of 5-10 MPa; and the third range corresponds to real-time data requiring repair, such as displacement changes exceeding 20 mm and stress changes exceeding 10 MPa. Correspondingly, early warning levels are set: primary warning corresponds to the first range, intermediate warning to the second range, and advanced warning to the third range. Thresholds are set based on historical statistics and engineering experience, for example, through machine learning clustering algorithms such as K-means to automatically optimize the thresholds. In implementation, data analysis software, such as Python pandas, is used to process the historical sample data to ensure that the hierarchical classification is scientifically sound and reasonable.

[0068] Used to obtain a visual model;

[0069] The visualization model generation unit is based on a 3D model. It acquires and classifies early warning systems from the acquisition module, displaying the warning levels on the 3D model using color coding: primary warnings correspond to the first color, such as green; intermediate warnings correspond to the second color, such as yellow; and advanced warnings correspond to the third color, such as red. The color coding location corresponds to the specific area where real-time data is collected, such as displacement sensor detection points or stress measurement points. In implementation, 3D rendering is performed using computer graphics software such as OpenGL or Unity. The model can be colored on line segments or curved surfaces and supports rotation, scaling, and sectioning views, allowing managers to intuitively view the risk distribution on the monitoring center's display screen.

[0070] The analysis module is also used to obtain deformation development trends and rates through machine learning models;

[0071] The machine learning analysis unit takes the sample library and the hierarchical early warning system as the first instructions, inputting them into a machine learning model, such as a Long Short-Term Memory (LSTM) network or a random forest, to perform regression analysis and time series analysis. Regression analysis is used to fit the trend of displacement or stress data and predict future deformation parameters. Time series analysis is used to identify data periodicity or abrupt changes and obtain the deformation development trend and rate, such as predicting the displacement change and average rate within the next hour. Then, real-time data, deformation development trend, and rate are used as the second instructions, inputting them into the same machine learning model to obtain prediction results, including theoretical real-time data, predicted values, risk levels, and location information. The risk level is determined based on the hierarchical early warning system, and the location information comes from sensor coordinates. In implementation, the model is trained using TensorFlow or Scikit-learn frameworks, with training data from the sample library. The verification accuracy requirement is ≥90% to ensure the reliability of the prediction.

[0072] Used to obtain prediction results and trigger early warnings;

[0073] The prediction result processing unit retrieves prediction results through encoding, as the sample library stores the mapping relationship between encodings and prediction results. For example, when real-time data matches a certain encoding, the system directly retrieves the prediction result corresponding to that encoding, including theoretical real-time data, risk level, and location information. Based on the prediction results, the decision module obtains the warning level and location, and triggers the corresponding warning method: a primary warning corresponds to displaying real-time data through the monitoring center and sending a message indicating everything is normal, along with real-time data, to designated personnel, such as inspectors, via SMS or a dedicated APP push notification; a secondary warning corresponds to triggering audible and visual alarms, such as buzzers and LED lights installed in the tunnel, and linking with the underground broadcasting system to release abnormal data messages and display location information, reminding personnel to check; a high-level warning corresponds to triggering audible and visual alarms, linking with the underground broadcasting system to release evacuation instructions, automatically locking equipment power, such as cutting off regional power supply and displaying location information, ensuring safe evacuation. In practice, the warning action is executed through control logic circuits and communication protocols, such as Modbus, with a response time of less than 5 seconds.

[0074] This invention, based on data analysis and machine learning principles, achieves intelligent monitoring of deformation states through tiered early warning and visualization. The tiered system utilizes statistical learning principles to simplify complex data into actionable early warning levels. The visualization model, based on graphics rendering technology, enhances information density and readability. Historical sample data is directly input into the machine learning model, which outputs prediction results in real time. Through color coding and trend analysis, the analysis module identifies abnormal increases in roof displacement and stress concentration areas, enhancing the sensitivity of fault detection. Through control theory and adaptive learning principles, the system achieves intelligent decision-making through early warning triggering and model optimization. The early warning method utilizes multimodal alarm technology to ensure comprehensive emergency response. Model optimization, based on error feedback principles, improves prediction accuracy. Because multi-parameter collaboration can detect risks early, it provides a basis for the decision-making module, improving the system's practicality and response speed.

[0075] In one embodiment, the decision module is used to use the sample library and the hierarchical early warning system as the first instruction to obtain the deformation development trend and rate, and to use the real-time data, deformation development trend and rate as the second instruction to obtain the prediction result. The corresponding prediction result is retrieved through encoding, and the early warning level and location are obtained based on the prediction result for maintenance and alarm.

[0076] The decision-making module triggers early warnings and maintenance actions based on the prediction results.

[0077] Used to optimize forecasting models and update the tiered early warning system;

[0078] The model optimization unit performs self-correction by comparing the difference between theoretical real-time data, predicted values, and actual real-time data: It calculates the absolute value of the difference; if the difference is greater than or equal to a first threshold (e.g., 5% error) and less than or equal to a second threshold (e.g., 10% error), the theoretical real-time data is considered valid and used to update the prediction results; if the difference is less than the first threshold or greater than the second threshold, the theoretical real-time data is discarded, and new prediction results are generated based on the real-time data, deformation development trend, and rate. Simultaneously, the decision module updates the tiered early warning system based on newly added historical data, such as reclassifying data range thresholds to adapt to geological changes. Once the historical data range of the sample library covers all expected working conditions, the system optimization process proceeds as follows: after acquiring real-time data, its encoding is directly sent to the cloud node, which retrieves the pre-stored prediction results based on the encoding, eliminating the need for recalculation. In implementation, iterative algorithms and database update mechanisms ensure continuous system learning.

[0079] In this embodiment of the invention, the early warning action is directly driven by the prediction results, and the efficiency is improved by encoding and retrieval. By comparing the difference and the threshold, the decision module dynamically adjusts the output to avoid model drift. The early warning classification and location display make the measures accurate and effective. This enhances the reliability and security of the system because the automated response reduces human delay, thereby improving the overall monitoring performance.

[0080] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program code. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0081] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A monitoring system for deformation of surrounding rock in downhole wells, characterized in that, It includes a data acquisition module, an analysis module, and a decision-making module: The acquisition module is used to acquire three-dimensional spatial geometric data, obtain a three-dimensional model, acquire real-time data through different sensor nodes, establish codes based on the real-time data, and add samples to the database based on the codes. The analysis module is used to obtain historical sample data from the sample library, establish a tiered early warning system based on the historical sample data, and obtain a visualization model through a 3D model based on the tiered early warning system and real-time data. The decision-making module is used to take the sample library and the hierarchical early warning system as the first instruction to obtain the deformation development trend and rate, take the real-time data, deformation development trend and rate as the second instruction to obtain the prediction result, retrieve the corresponding prediction result through the code, obtain the early warning level and location based on the prediction result, and carry out maintenance and alarm.

2. The monitoring system for downhole surrounding rock deformation as described in claim 1, characterized in that, The three-dimensional model specifically includes: The three-dimensional spatial geometric data includes tunnel cross-sectional contour data, surrounding rock surface morphology data, and borehole observation data; The tunnel cross-sectional profile data includes the initial shape and size of the tunnel; the surrounding rock surface morphology data includes three-dimensional coordinate information of the undulations and fissure distribution of the surrounding rock surface after tunnel excavation; and the borehole inspection data includes three-dimensional spatial information of the internal structure of the surrounding rock. The cross-sectional contour data of the tunnel was obtained using a total station and a 3D laser scanner; The borehole observation data is obtained through borehole cameras and radar; A three-dimensional model is obtained based on the three-dimensional spatial geometric data.

3. A monitoring system for downhole surrounding rock deformation as described in claim 2, characterized in that, The real-time data specifically includes: The real-time data includes displacement data, stress data, and microseismic data. The displacement data includes the relative convergence of the roof and floor of the underground coal mine roadway, the convergence displacement of the two sides, and the three-dimensional coordinate changes of specific points on the roadway surface. The stress data includes the internal stress of the surrounding rock, the stress of the anchor bolts, the stress of the anchor cables, and the working resistance of the hydraulic supports. The microseismic data includes the energy, location, and timing of microseismic events generated by rock mass fracturing. The displacement data is acquired using a displacement sensor; The stress data is acquired using a stress sensor; The microseismic data is acquired using a microseismic sensor.

4. A monitoring system for downhole surrounding rock deformation as described in claim 3, characterized in that, The establishment of the encoding, and the addition of the sample library based on the encoding, specifically includes: By setting up underground nodes and cloud nodes through a communication bus, the real-time data is transmitted to the underground nodes and matched with the historical data range; If the real-time data is within the range of the historical data, then the encoding of the historical data is transmitted to the cloud node; If the real-time data is not within the range of the historical data, a new code is established based on the real-time data, the real-time data is expanded according to a preset expansion range, the range of historical data is obtained, and it is saved to the underground node. Then, the real-time data and the new code are transmitted to the cloud node. The cloud node expands the sample library according to the real-time data and the preset expansion range, and the sample library includes sample historical data, codes and prediction results. The historical data range is defined by taking real-time data as the center, using a preset expansion range as the interval radius, determining an expanded numerical interval, expanding the real-time data, establishing sample codes for all virtual data points within the expanded numerical interval, and treating the virtual data points as acquired real-time data. The historical data of the sample consists of all virtual data points within the expanded numerical range, and these virtual data points are sorted according to time.

5. A monitoring system for downhole surrounding rock deformation as described in claim 4, characterized in that, The tiered early warning system specifically includes: Based on the sample library, historical sample data is obtained; based on the historical sample data, data ranges are divided; based on the data ranges, warning levels are set; the data ranges include a first range, a second range, and a third range. The warning level is set based on the real-time data of the degree of deformation in the historical data of the sample. The data range is the maximum and minimum value of the real-time data of normal deformation, the maximum and minimum value of the real-time data of deformation that needs to be checked, and the maximum and minimum value of the real-time data of deformation that needs to be repaired. The warning level corresponding to the first range is a primary warning, the warning level corresponding to the second range is a medium warning, and the warning level corresponding to the third range is a high warning.

6. A monitoring system for downhole surrounding rock deformation as described in claim 5, characterized in that, The visualization model specifically includes: The warning levels are displayed in different colors at corresponding positions on the 3D model to obtain a visualization model. The primary warning corresponds to the first color, the intermediate warning corresponds to the second color, and the advanced warning corresponds to the third color. The corresponding position is a color-coded position that corresponds to a specific area of ​​the collected real-time data.

7. A monitoring system for downhole surrounding rock deformation as described in claim 6, characterized in that, The analysis module specifically includes: The sample library and the hierarchical early warning system are used as the first instruction and input into the machine learning model to perform regression analysis and time series analysis to obtain the deformation development trend and rate. Real-time data, deformation development trend and rate are used as the second instruction and input into the machine learning model to obtain the prediction results. The prediction results are theoretical real-time data, risk level and location information. Based on the prediction results, the corresponding warning level is determined by the color displayed by the visualization model, the corresponding location information is obtained based on the position of the color display, and the corresponding warning method is triggered based on the warning level.

8. A monitoring system for downhole surrounding rock deformation as described in claim 7, characterized in that, The corresponding early warning methods specifically include: The warning method corresponding to the primary warning is to display the real-time data through the monitoring center and send a message that everything is normal and the real-time data to designated personnel. The intermediate-level early warning method is to trigger an audible and visual alarm, and link the downhole broadcasting system to release data anomaly messages and display location information. The advanced early warning system uses a warning method that triggers an audible and visual alarm, links the underground broadcasting system to issue evacuation instructions, automatically locks the equipment power supply, and displays location information.

9. A monitoring system for downhole surrounding rock deformation as described in claim 8, characterized in that, The decision-making module specifically includes: The difference between the theoretical real-time data and the real-time data is obtained. If the difference is greater than or equal to the first threshold and less than or equal to the second threshold, the theoretical real-time data is used as the real-time data to obtain a new prediction result. If the difference is less than the first threshold or greater than the second threshold, the theoretical real-time data is discarded, and a new prediction result is obtained based on the real-time data, deformation development trend and rate.

10. A monitoring system for downhole surrounding rock deformation as described in claim 9, characterized in that, The decision-making module also includes: The tiered early warning system is updated based on the newly added historical sample data, and the corresponding data range is adjusted based on the updated tiered early warning system. Once the historical data range of the sample library covers all expected working conditions of downhole surrounding rock deformation, real-time data is acquired, and the code corresponding to the real-time data is sent to the cloud node. The cloud node directly retrieves the corresponding prediction result based on the code.

Images