A soft rock area coal exploration hole grading wall plugging method based on real-time judgment

Abstract

The present application relates to the technical field of soft rock area coal exploration hole construction, and discloses a soft rock area coal exploration hole grading wall protection and leakage stopping method based on real-time determination; the present application constructs a multi-source monitoring network to cooperate with an edge computing module to form a real-time data acquisition and processing system, designs a double determination model composed of a grading threshold dynamic correction model driven by a stratum parameter inversion algorithm and a composite risk identification model with a hybrid architecture of a rule engine cooperating with a lightweight decision tree algorithm, forms a complete technical system of data acquisition, dynamic determination, man-machine collaborative disposal and full-process data closed-loop evolution, realizes accurate determination of the composite risk of the soft rock area coal exploration hole, reduces the in-hole accident rate from more than 30% to less than 5%, improves the hole forming efficiency by 15%-20%, reduces the drilling fluid loss by 25%, realizes dynamic risk determination, customized disposal scheme and closed-loop technology system, and the industrial-grade hardware is adapted to the complex environment of high temperature, high vibration and strong interference underground, has strong environmental adaptability and continuous optimization capacity.

Description

Technical Field

[0001] This invention relates to the field of coal exploration borehole construction technology in soft rock areas, and in particular to a graded wall protection and leakage plugging method for coal exploration boreholes in soft rock areas based on real-time judgment. Background Technology

[0002] Soft rock formations are characterized by swelling upon contact with water, developing fissures, and poor mechanical properties. During the construction of coal exploration boreholes, they are prone to accidents such as leakage, collapse, and water inrush. Moreover, most accidents occur in the form of complex risks, which seriously affect construction safety and borehole efficiency.

[0003] The above-mentioned and existing related technologies have the following drawbacks: Traditional methods for sealing and plugging coal exploration boreholes in soft rock areas generally use fixed thresholds to determine risks, which cannot adapt to dynamic changes in the formation. They adopt a one-size-fits-all approach, which makes it difficult to take into account the multiple prevention and control needs of complex risks. In addition, traditional methods lack a data closed-loop mechanism, and construction experience cannot be transformed into the basis for optimizing technical parameters, resulting in repeated accidents in similar constructions, a high accident rate in the borehole, and long borehole construction cycles and high costs. Summary of the Invention

[0004] The technical problem to be solved by this invention is that the existing technology has the disadvantages of inaccurate risk assessment, poor adaptability of treatment plan, and lack of continuous optimization mechanism. To this end, we propose a graded wall protection and leakage plugging method for coal exploration holes in soft rock areas based on real-time assessment.

[0005] To achieve the above objectives, this application adopts the following technical solution: a method for graded wall protection and leakage plugging of coal exploration holes in soft rock areas based on real-time judgment, comprising the following steps: Data Acquisition: Collect geological data of the target mining area and construction cases of similar soft rock areas. Based on the geological data and construction cases, obtain basic materials. Based on the basic materials, pre-set characteristic parameters of different types of soft rock. Analyze the geological data and construction cases to obtain the dynamic change law of soft rock area geology and complex risk pain points. System construction: Deploy a multi-source monitoring network and select an appropriate edge computing module based on characteristic parameters. Obtain a dual-judgment model by analyzing the dynamic changes in the soft rock area and the complex risk pain points. Start the dual-judgment model and import expert experience rules and basic training data. Data preprocessing: Parameter data is collected synchronously through a multi-source monitoring network, and after noise reduction and standardization by the edge computing module, a downhole dynamic data pool is constructed. Risk assessment is then performed based on the dynamic data pool. Dynamic risk assessment: The dual-assessment model is used to analyze the dynamic data pool, correct the risk threshold, identify the working condition type, output the risk level of hole wall instability and leakage, and generate a support and treatment plan. Solution generation and human-machine collaboration: Based on the risk assessment results, a treatment plan containing material formulas and process parameters is generated, and an execution instruction is formed after manual review and confirmation; Execution and Coordination Control: Execute instructions to complete wall protection and leakage plugging operations, and simultaneously adjust drilling and borehole inclination control parameters according to risk level and working conditions; Effect feedback and model evolution: Monitor the effect of wall protection and leak sealing operations. If the results are not as expected, trigger a second optimization of the solution and feed the full process data back to the dual-judgment model to achieve incremental learning of the model.

[0006] Preferably, the geological data of the target mining area includes geological survey reports, columnar sections of coal seam roof boreholes, soft rock mechanical test data and groundwater data. Similar soft rock area construction cases cover three types of accidents: leakage and collapse, water inrush and fracture zone, and progressive hidden collapse. Accident triggering parameters, disposal plans and failure causes are recorded.

[0007] Preferably, the multi-source monitoring network consists of a drill pressure torque sensor, a borehole imager, and a logging-while-drilling (LWD) device. The sensor measurement accuracy is ±1%, the imager resolution is 1080P, and the LWD device measurement accuracy is ±2%. The devices transmit data synchronously via industrial Ethernet.

[0008] Preferably, the edge computing module is industrial grade, resistant to temperature -20~60℃, vibration 5-50Hz, and processing latency ≤50ms. After being adapted to the monitoring network interface and calibrated, the data conversion error is ≤0.5%.

[0009] Preferably, the dynamic correction model for the graded threshold in the dual-judgment model adopts a formation parameter inversion algorithm, the formula of which is: Critical leakage threshold = formation permeability × soft rock compressive strength × mining area correction coefficient, where the mining area correction coefficient is 0.75.

[0010] Preferably, the composite risk identification model adopts a hybrid architecture of rule engine and decision tree algorithm. The rule engine has 12 built-in rules, and the decision tree is trained based on 50 sets of composite risk samples with a depth of 6 layers. The input features include leakage and torque fluctuation.

[0011] Preferably, the monitoring network collects three types of data—engineering, geology, and geophysics—at a frequency of once per minute. Outliers exceeding ±20% are removed using Kalman filtering, and the data are standardized to the 0-1 range using min-max standardization to construct a downhole dynamic data pool.

[0012] Preferably, the supporting response plan is matched with known risk scenarios, with a response time ≤10ms. Unmatched items are analyzed using a decision tree, and the risk confidence score is calculated using the following formula: Confidence score = rule matching score × 0.6 + decision tree probability × 0.4.

[0013] Preferably, the treatment plan includes material formulation and process parameters. The on-site engineer, based on the real-time images from the in-hole imaging instrument, fine-tunes the material ratio and pumping pressure parameters before issuing the execution command.

[0014] Preferably, the effect monitoring includes short-term and long-term indicators, and the data of the whole process is fed back to the dual judgment model to realize incremental learning. If the target is not met, the secondary optimization of material particle size, pumping pressure and setting time is triggered.

[0015] The technical effects and advantages of this invention are as follows: This invention constructs a real-time data acquisition and processing system combining a multi-source monitoring network with an edge computing module. It designs a dual-judgment model consisting of a hierarchical threshold dynamic correction model driven by a formation parameter inversion algorithm and a hybrid risk identification model combining a rule engine with a lightweight decision tree algorithm. This forms a complete technical system encompassing data acquisition, dynamic judgment, human-machine collaborative handling, and a closed-loop evolution of the entire data process. This enables accurate judgment of composite risks in coal exploration boreholes in soft rock areas, reducing the borehole accident rate from over 30% to below 5%, increasing borehole formation efficiency by 15%-20%, and reducing drilling fluid consumption by 25%. It achieves dynamic risk judgment, customized handling solutions, and a closed-loop technical system. Furthermore, the industrial-grade hardware is adapted to the complex environment of high temperature, high vibration, and strong interference in downhole environments, possessing strong environmental adaptability and continuous optimization capabilities. Attached Figure Description

[0016] The disclosure of this invention is illustrated with reference to the accompanying drawings. It should be understood that the drawings are for illustrative purposes only and are not intended to limit the scope of protection of this invention. In the drawings, the same reference numerals are used to refer to the same parts: Figure 1 This is a flowchart of the graded wall protection and leakage plugging method for coal exploration holes in soft rock areas based on real-time judgment according to the present invention; Figure 2 This is a flowchart of the data acquisition process of the present invention; Figure 3 A flowchart illustrating the system construction of this invention; Figure 4 This is a flowchart of the dynamic risk assessment method of the present invention; Figure 5 This is a flowchart of the effect feedback and model evolution sub-process of the present invention. Detailed Implementation

[0017] It is readily understood that, based on the technical solution of this invention, those skilled in the art can propose various interchangeable structural methods and implementations without altering the essential spirit of the invention. Therefore, the following detailed embodiments and accompanying drawings are merely illustrative examples of the technical solution of this invention and should not be considered as the entirety of the invention or as limitations or restrictions on the technical solution of this invention.

[0018] According to the embodiments of the present invention, Figures 1 to 5 As shown.

[0019] Traditional methods for sealing and plugging leaks in coal exploration boreholes in soft rock areas generally rely on fixed thresholds to determine risk, which cannot adapt to dynamic changes in the formation. They employ a one-size-fits-all approach, failing to address the multiple prevention and control needs of complex risks. Furthermore, traditional methods lack a data closed-loop mechanism, preventing the conversion of construction experience into a basis for optimizing technical parameters. This leads to recurring accidents in similar operations, a persistently high rate of borehole accidents, and long drilling cycles with high costs. To address this issue, this invention incorporates the following design into a graded sealing and plugging method for coal exploration boreholes in soft rock areas based on real-time risk assessment: A method for graded wall protection and leakage plugging of coal exploration boreholes in soft rock areas based on real-time judgment includes the following steps: Data Acquisition: Collect geological data of the target mining area and construction cases of similar soft rock areas. Based on the geological data and construction cases, obtain basic materials. Based on the basic materials, pre-set characteristic parameters of different types of soft rock. Analyze the geological data and construction cases to obtain the dynamic change law of soft rock area geology and complex risk pain points. System construction: Deploy a multi-source monitoring network and select an appropriate edge computing module based on characteristic parameters. Obtain a dual-judgment model by analyzing the dynamic changes in the soft rock area and the complex risk pain points. Start the dual-judgment model and import expert experience rules and basic training data. Data preprocessing: Parameter data is collected synchronously through a multi-source monitoring network, and after noise reduction and standardization by the edge computing module, a downhole dynamic data pool is constructed. Risk assessment is then performed based on the dynamic data pool. Dynamic risk assessment: The dual-assessment model is used to analyze the dynamic data pool, correct the risk threshold, identify the working condition type, output the risk level of hole wall instability and leakage, and generate a support and treatment plan. Solution generation and human-machine collaboration: Based on the risk assessment results, a treatment plan containing material formulas and process parameters is generated, and an execution instruction is formed after manual review and confirmation; Execution and Coordination Control: Execute instructions to complete wall protection and leakage plugging operations, and simultaneously adjust drilling and borehole inclination control parameters according to risk level and working conditions; Effect feedback and model evolution: Monitor the effect of wall protection and leak sealing operations. If the results are not as expected, trigger a second optimization of the solution and feed the full process data back to the dual-judgment model to achieve incremental learning of the model.

[0020] In this embodiment, data acquisition specifically includes: Collect geological data for the target mining area: Collect geological exploration reports and borehole columnar sections of the coal seam roof; simultaneously acquire indoor mechanical test data for soft rock and relevant groundwater data, including mudstone saturated compressive strength of 8-12 MPa, shale swelling rate of 15-20%, and formation permeability of [missing data]. The aquifer pressure is 0.8-1.2 MPa. Based on the above data, the primary physical and mechanical properties of the soft rock and the formation occurrence state are clarified.

[0021] Collect similar construction cases in soft rock areas: comprehensively organize the complete construction records of three similar coal exploration boreholes in the adjacent area, focus on sorting out the details of 12 borehole accidents, including 5 cases of leakage or collapse, 4 cases of water inrush or fracture zone, and 3 cases of progressive hidden collapse. Record the triggering parameters, treatment plan and failure cause of each accident in detail, and establish a related database.

[0022] Generate basic materials and pre-set soft rock characteristic parameters: Based on geological data and case data, through statistical analysis and cross-validation, the core characteristic parameters of different soft rocks are extracted and pre-set into the system to ensure that the parameters are accurately adapted to the construction scenario. Among them, the permeability of mudstone is... The shale has a softening coefficient of 0.65, a critical fracture aperture of 0.4 mm, an expansion rate of 18%, a compressive strength of 10 MPa, and a permeability coefficient of [missing information]. .

[0023] Extracting the dynamic patterns of geological changes and identifying complex risks: Using a combination of data analysis and case study analysis, the core geological dynamic patterns were derived. When drilling is disturbed, the fracture aperture expands at a rate of 0.1-0.3 mm / h. At this time, the soft rock absorbs water and expands, and the borehole wall deformation increases. Through accident attribution analysis, two core complex risks were identified: when leakage... Furthermore, when the torque fluctuation range is greater than 15%, it is judged as a risk of leakage and collapse; when the resistivity drops sharply by 30%, the leakage increases and the formation pressure is greater than 1.0 MPa, it is judged as a risk of coexistence of high-pressure water inrush and fracture zone; the above two core composite risk pain points provide a direct basis for the design of the dual judgment model.

[0024] In this embodiment, the system construction specifically includes: Multi-source monitoring network deployment: Based on pre-set soft rock characteristic parameters, targeted equipment is selected and deployed to ensure the accuracy and real-time nature of data acquisition. The drill pressure torque sensor selected is model ZCY-200, with a measurement range of 0-10000KN / 0-5000N. With an accuracy of ±1%, it adapts to the dynamic range of drilling pressure and torque changes in soft rock areas, accurately capturing risk signals with torque fluctuations exceeding 15%. The borehole imaging instrument selected is model KJ-600, with a resolution of 1080P and a detection depth of 0-1000m. It can clearly identify fracture openings larger than 0.1mm, providing intuitive geological evidence for composite risk assessment. The logging-while-drilling equipment selected is model SJD-300, which simultaneously acquires resistivity and natural gamma data with a measurement accuracy of ±2%. It can accurately capture abnormal signals with a resistivity drop of more than 30%. All equipment is connected via industrial Ethernet with a transmission delay of ≤100ms, achieving synchronous transmission of engineering parameters, geological parameters, and geophysical parameters, and ensuring consistency of timestamps for multi-source data.

[0025] Edge computing module selection and calibration: Considering the complex environment of downhole equipment vibration and signal interference, a suitable edge computing module was selected to ensure the real-time performance and reliability of data preprocessing. The selected module is the ECM-800 industrial-grade edge computing module, which supports multi-protocol access, has a real-time data processing latency of ≤50ms, and is adaptable to environments with temperatures ranging from -20 to 60℃ and vibration frequencies from 5 to 50Hz, fully matching the harsh downhole operating conditions. The module's RS485 Ethernet interface was adapted to the multi-source monitoring network and calibrated using a standard signal source to ensure a data conversion error of ≤0.5%, providing hardware support for subsequent data processing.

[0026] Dual-judgment model construction: Based on the refined geological dynamic change patterns and complex risk pain points, a dual-judgment model is designed to ensure that the model has dynamic adaptation and complex recognition capabilities. The core algorithm of the graded threshold dynamic correction model adopts the formation parameter inversion algorithm. Based on the positive correlation between risk threshold and formation dynamic parameters, a customized formula is designed: Critical leakage threshold = Formation permeability × Soft rock compressive strength × Mining area correction coefficient; where the mining area correction coefficient = 0.75, which is obtained by fitting 30 sets of historical data of the mining area to ensure that the formula is adapted to local geological conditions. Dynamic parameters such as formation permeability and soft rock compressive strength are received in real time and substituted into the formula to calculate the specific critical leakage threshold of the current formation, replacing the traditional fixed threshold and realizing adaptive judgment that the threshold changes with the formation.

[0027] Composite Risk Identification Model: Employing a hybrid architecture of rule engine and lightweight decision tree algorithm, it balances rapid matching of known risks with adaptability to new risks. It transforms the two core composite risk pain points and industry expert experience into 12 executable rules: When leakage exceeds the critical threshold and torque fluctuation exceeds 15%, it is identified as leakage + collapse risk; when resistivity drops sharply by more than 30% and formation pressure exceeds 1.0 MPa and leakage increases, it is identified as high-pressure water inrush + fracture zone coexistence risk.

[0028] Decision tree training: 50 typical composite risk samples were selected, covering risk levels I-IV. With leakage, torque fluctuation, resistivity change, formation pressure, and fracture aperture as input features and risk type as output label, a decision tree model with a depth of 6 layers was trained and generated. The model training accuracy reached over 90%.

[0029] Collaborative logic: The rule engine prioritizes matching known complex risk scenarios, where the response time is ≤10ms. New parameter combinations that are not matched are analyzed by the decision tree algorithm to ensure that no complex risk is missed in the identification.

[0030] Model initialization and calibration: Start the dual-judgment model, import basic data to complete initialization, and ensure that the model has initial judgment capability. Transform 15 domain expert experience rules into conditional statements that the model can recognize and embed them into the rule engine. For example, when the fracture aperture is >0.5mm, the leakage risk threshold is reduced by 25%. When the natural gamma value is >80API and the leakage amount increases, it is judged as a water-rich fracture zone. Import 100 sets of typical soft rock area construction history data, including risk levels I-IV, single or compound conditions, covering different soft rock types such as mudstone and shale, for model parameter initialization. Use 5 sets of measured data that were not used in training for verification. Adjust the model weight parameters to ensure that the accuracy of risk level judgment is ≥88% and the accuracy of compound risk type identification is ≥85%. The model initialization is completed.

[0031] In this embodiment, data preprocessing specifically includes: Data Acquisition: After drilling commences, three types of core data are simultaneously collected once per minute via a multi-source monitoring network to ensure data coverage of the dimensions required for risk assessment. These engineering parameters include drilling pressure of 4800-6500 KN and torque of 1200-2800 N. m, leakage Geological parameters include core fragmentation of 0-100% and fracture aperture of 0-2mm; geophysical parameters include resistivity of 20-100Ω. m, Natural Gamma 30-120 API.

[0032] Data processing: Two-step processing is performed using the ECM-800 edge computing module. A Kalman filter algorithm is employed to filter parameters susceptible to vibration interference, such as drill pressure and torque, eliminating anomalous data with instantaneous fluctuations exceeding ±20%, retaining accurate formation response data. The min-max normalization method is used to unify data of different units and magnitudes into a 0-1 numerical range, eliminating dimensional interference. For example, core fragmentation of 0-100% is directly converted to 0-1, and resistivity of 20-100Ω. The standardized formula for m is: .

[0033] Construct a dynamic downhole data pool: Store the processed high-quality data in a structured format according to acquisition time, parameter type, value, and reliability. Reliability is determined by a combination of data integrity and equipment status. The reliability is 0.95 when the equipment is working normally and 0.6 when there is signal interference. The data pool adopts a dual storage mode of local caching and cloud backup. The local cache capacity is ≥10GB, and the cloud is synchronized in real time to ensure that the data is not lost and is traceable, providing accurate input for dynamic risk assessment.

[0034] In this embodiment, dynamic risk assessment specifically includes: Model calling and data input: The system extracts the latest structured data from the downhole dynamic data pool once per minute and automatically inputs it into the initialized dual-judgment model. The data input delay is ≤20ms to ensure the real-time nature of the judgment.

[0035] Dual-model collaborative analysis: When leakage is collected Torque fluctuation range 18%, resistivity 65Ω When real-time data of m, formation pressure 1.1 MPa, and fracture aperture 0.6 mm are available, based on the real-time acquired fracture aperture of 0.6 mm and soft rock compressive strength of 10 MPa, the current formation permeability is calculated using a formation parameter inversion algorithm. to increase the permeability of the formation Substituting the soft rock compressive strength of 10 MPa and the mining area correction factor of 0.75 into the formula, the current formation critical leakage threshold is calculated. The system automatically adjusts the minor omission threshold from the initial value. Upgraded to To ensure that the risk level assessment is consistent with the actual formation, real-time data is compared with 12 built-in rules. The rule matching is found to be a combination of high-pressure water inrush and fractured zone coexisting. The real-time data is then input into the trained decision tree model, which outputs a matching probability of 93% for this condition. This is cross-validated with the rule engine results. The confidence level of the judgment is calculated as (rule matching score × 0.6 + decision tree probability × 0.4) = 92%, ensuring the reliability of the judgment result.

[0036] Risk information output: Risk level III, operating condition attribute is composite risk, judgment confidence level is 92%, the core issues are high-pressure water inrush leading to leakage and fracture zone causing borehole wall instability, and the leakage amount of supporting data is... Supercritical threshold The resistivity decreased by 32%, and the formation pressure was 1.1 MPa.

[0037] In this embodiment, the solution generation and human-machine collaboration specifically include: Intelligent prescription automatic generation: Addressing the combined risks of high-pressure water inrush and fractured zones, the system simultaneously meets three major requirements: rapid water plugging, fracture reinforcement, and resistance to soft rock expansion. Based on pre-set soft rock characteristic parameters (mudstone softening coefficient 0.65, shale expansion rate 18%), an integrated solution combining material formulation and process parameters is automatically generated. The base slurry uses high-strength cement slurry with a compressive strength ≥42.5MPa, adding 9% quick-setting water plugging agent (model SND-1) with an initial setting time ≤30min, 3% flexible fiber, and 6% expanding particles to fill micro-fractures and counteract soft rock expansion deformation. Based on real-time data of a formation pressure of 1.1MPa and a fracture aperture of 0.6mm, a pumping pressure of 1.2-1.5MPa is calculated, with an initial step pumping displacement of [missing information]. Voltage stabilization Sealing holes The condensation time is 4 hours.

[0038] Human-machine collaborative decision-making: The system automatically generates a solution and pushes it to the on-site engineer's terminal in the form of a graphic report. The report includes solution details, generation basis, and links to real-time images inside the borehole. The engineer combines the real-time images transmitted by the borehole imager to analyze that the fracture density in the area is higher than the average level, and it is necessary to increase the grout solidification speed. The proportion of the quick-setting water plugging agent is slightly adjusted to 10%. At the same time, considering that the formation compressive strength at this depth is low, the upper limit of the pumping pressure is adjusted to 1.4MPa to avoid secondary fractures caused by excessive pressure. After the engineer approves the solution, the final execution instruction is issued through the terminal. The instruction includes complete material formula, process parameters, and execution time window.

[0039] In this embodiment, execution and coordinated control specifically include: Implementation of wall protection and leak sealing operations: Mix high-strength cement slurry + 10% quick-setting water-stopping agent + 3% flexible fiber + 6% expanding granules according to the adjusted formula, and stir for 15 minutes. Control the slurry density at [specific values ​​to be filled in]. The grout injection adopts a stepped pumping mode, starting from a depth of 350m, with an initial discharge rate of... After 30 minutes, adjust to Stabilize the voltage for 2 hours, and finally... The borehole was sealed for 30 minutes, with real-time monitoring via a pump pressure sensor to ensure the pressure remained stable at 1.2-1.4 MPa. Based on the Level III composite risk conditions, drilling parameters were dynamically adjusted, with drilling pressure reduced to 5000-5800 KN and the rotation speed adjusted to... The spacing between guide tubes was reduced from 8m / tube to 5m / tube. The borehole inclination was monitored in real time by the drilling rig attitude sensor to ensure that the borehole inclination deviation was ≤0.5%, thus achieving seamless connection between wall protection and drilling operations and avoiding mutual interference.

[0040] In this embodiment, the effect feedback and model evolution specifically include: Comprehensive Effect Monitoring: Four hours after condensation, short-term and long-term indicators are continuously monitored through a multi-source monitoring network to comprehensively verify the treatment effect. Short-term indicators are monitored using leakage sensors. Down to The borehole wall deformation was measured using a borehole imaging system, with a maximum deformation of 3 mm. The pumping pressure remained stable at 1.3 MPa without significant fluctuations, indicating that the grout filling was dense. Long-term performance indicators were monitored using core sampling tests, reaching 8 MPa. The groundwater permeability coefficient was measured using pumping tests, decreasing to [a value missing]. The continuous drilling verification showed no abnormalities in the subsequent 24 hours, with no leakage or collapse, and the treatment effect met the standards.

[0041] Incremental model learning: The entire construction process data was labeled as typical samples of coexistence of Level III high-pressure water inrush and fractured zones, including 560 sets of real-time parameters collected, risk level, working condition type, and confidence level of the judgment results. The material formula after parameter adjustment included 10% quick-setting water-blocking agent and the process parameter pumping pressure of 1.2-1.4 MPa. The effect feedback was provided by short-term and long-term monitoring data, which were stored in the model training sample library. New samples were input into the dual-judgment model to perform incremental learning. The graded threshold dynamic correction model was based on the formation permeability measured in this study. Critical leakage threshold The data was refitted to improve the correction coefficient for the mining area from 0.75 to 0.78, making the formula more consistent with the actual situation of the mining area. The composite risk identification model added new samples to the decision tree training set and readjusted the tree node weights to improve the sensitivity to parameter combinations such as sudden drop in resistivity and excessive formation pressure. After optimization, the confidence level of the model in judging this type of working condition increased to 96%. If the treatment effect does not meet the standard, the system automatically links real-time data and judgment logic to trigger a second optimization of the scheme, adjusts the particle size of the expansion particles to 1-2mm, increases the pumping pressure to 1.6MPa, and extends the waiting time to 6 hours.

[0042] Evolution effect verification: Through subsequent construction verification under three similar working conditions, the confidence level of the optimized model was increased to an average of 95%, the adaptability of the treatment plan was improved by 15%, the amount of grout used was reduced by 10%, and the accident rate in the borehole was reduced from 32% in history to 5%, thus verifying the effectiveness of the model evolution.

[0043] Example 1 In response to the core characteristics of shale, such as a swelling rate of 25-30%, rapid cementation upon contact with cement, and a fracture propagation rate of 0.3-0.5 mm / h, the focus is on addressing the combined risks of swelling and shrinkage combined with local leakage in highly expansive shale mining areas.

[0044] First, data collection was conducted by retrieving the 1:5000 geological survey report of the mining area and the shale indoor mechanical test report to obtain the shale compressive strength (6-8 MPa) and permeability. The basic parameters of aquifer pressure (0.6-0.9 MPa) were established. Simultaneously, construction records from four similar coal exploration boreholes in adjacent areas were compiled, and details of 16 borehole accidents were analyzed. The core pain point was identified as fracture aperture > 0.3 mm + expansion rate > 25% = diameter reduction + leakage risk. Subsequently, a system was constructed. Based on the high expansion characteristics of shale, the correction factor for the mining area was set to 0.8. ZCY-200 drilling pressure torque sensors, KJ-600 borehole imaging instruments, and a supporting multi-source monitoring network were deployed. An ECM-800 edge computing module was selected to complete interface adaptation and calibration. Expansion was added to the dual-judgment model rule engine. The threshold adjustment rules were correlated with the risk level. 120 sets of typical shale mining area construction data were imported to complete model initialization, ensuring a risk level assessment accuracy of ≥88%. After drilling commenced, data preprocessing was performed. Three core data types—leakage, torque fluctuation, and fracture aperture—were collected via a multi-source monitoring network at a frequency of once per minute. The Kalman filter algorithm of the edge computing module was used to remove outliers with instantaneous fluctuations exceeding ±20%. The min-max standardization method was used to unify the data format, and a downhole dynamic data pool was constructed. In the dynamic risk assessment phase, the system extracted leakage data from the data pool at a frequency of once per minute. Real-time data of torque fluctuation of 20% and crack opening of 0.4mm were substituted into the formula to calculate the critical leakage threshold. After comprehensive analysis by the dual-judgment model, a judgment result of Level II composite risk with a confidence level of 90% is output.

[0045] In the scheme generation and human-machine collaboration phase, the system automatically generates a material formula of high-strength cement slurry + 5% anti-expansion agent + 3% flexible fiber, along with process parameters of pumping pressure 0.8-1.0 MPa and setting time of 6 hours. The on-site engineer, combining the fracture distribution images transmitted from the borehole imaging instrument, confirms that the scheme requires no adjustment before issuing the execution command. In the execution and collaborative control phase, construction personnel mix the slurry according to the formula, controlling the mixing time to 18 minutes, and maintaining the slurry density at [specific parameters missing]. A stepped pumping method was used to inject the material into the borehole, and the drilling parameters were simultaneously adjusted to a drilling pressure of 4500-5000 KN and a rotation speed of [missing information]. The spacing between the guide tubes was maintained at 5m / tube. Finally, feedback and model evolution were conducted. After 6 hours of condensation, the leakage was monitored through a multi-source monitoring network and reduced to [a certain value]. The deformation of the borehole wall is ≤2mm, and the treatment effect meets the standard. The 620 sets of full-process data of this construction are marked as typical samples and fed back to the dual judgment model. After incremental learning, the confidence of judgment for similar working conditions is increased to 94%.

[0046] Example 2 In view of the characteristics of sandstone, such as low cementation strength, susceptibility to collapse and sand loss, formation permeability A weakly cemented sandstone mining area with a compressive strength of 15-20MPa and a historical accident rate of 35%.

[0047] First, data collection was conducted, gathering geological survey data from the mining area and construction records from three similar coal exploration boreholes in adjacent areas. The focus was on analyzing the triggering parameters and causes of failure in handling 18 collapse and sand-carrying loss accidents, and extracting the loss volume. The core pain point is that a core fragmentation degree greater than 80% indicates a risk of collapse and sand loss. The system was then constructed, and considering the uniform porosity of sandstone, a correction factor of 0.7 was set for the mining area. When deploying a multi-source monitoring network, the borehole imaging system was upgraded to a 4K resolution model to enhance the accuracy of core fragmentation identification. An ECM-800 edge computing module was selected and calibrated. A dual-judgment model was built, and 150 sets of sandstone mining area construction data were imported to complete model initialization. After drilling commenced, core data such as leakage, core fragmentation degree, and torque fluctuations were collected at a frequency of 1 time per minute. After noise reduction and standardization by the edge computing module, a dynamic downhole data pool was constructed. In the dynamic risk assessment stage, the system extracted leakage data. Real-time data of 85% core fragmentation and 25% torque fluctuation were used to calculate the critical leakage threshold. The dual-judgment model outputs a Level IV composite risk assessment result with a 93% confidence level. During the scheme generation and human-machine collaboration phase, the system automatically generates a material formula of high-strength cement grout + 8% flexible fiber + 6% expanded particles, along with pumping pressure of 1.5-1.8 MPa and process parameters for grouting before drilling. After engineer review and confirmation, in the execution and collaborative control phase, construction personnel employ segmented grouting operations, controlling the length of each segment to 10m, with a grout mixing time of 20 minutes and a grout density... During pumping, pressure was monitored in real time, and the spacing between guide pipes was shortened to 4m / pipe. Drilling parameters were adjusted to a drilling pressure of 5500-6000KN and a rotation speed of [missing information]. During the feedback and model evolution phase, monitoring data after 8 hours showed that the leakage rate had decreased. The risk of collapse has been completely eliminated. The 750 sets of full-process data from this construction project were fed back to the model. After incremental learning, the correction coefficient for the mining area was optimized to 0.72, and the confidence level for similar working conditions was increased to 97%.

[0048] Example 3 This study addresses the characteristics of mudstone with a softening coefficient of 0.5-0.6, aquifer pressure of 1.2-1.5 MPa, and the risk of combined high-pressure water inrush and fracture zone, resulting in a historical accident rate of 40% in mudstone-rich mining areas.

[0049] First, data collection was conducted, retrieving geological survey reports from the mining area and indoor mechanical test data of mudstone to obtain mudstone permeability. Based on the fundamental parameters of compressive strength of 9-11 MPa, and by compiling construction records from five similar coal exploration boreholes in neighboring areas, the core pain point was identified as a 40% drop in resistivity plus formation pressure > 1.2 MPa, indicating high-pressure water inrush and risks associated with fractured zones. Subsequently, a system was constructed, setting the correction factor for the mining area to 0.85, considering the high pressure characteristics of water-rich formations. Water pressure sensors were added to the multi-source monitoring network, and hardware such as drill pressure torque sensors and logging-while-drilling equipment were deployed. An ECM-800 edge computing module was selected and calibrated. A dual-judgment model was constructed, and 130 sets of construction data from water-rich cement rock mining areas were imported for initialization. After drilling commenced, core data such as leakage, resistivity, and formation pressure were collected at a frequency of once per minute. After processing by the edge computing module, a dynamic downhole data pool was constructed. In the dynamic risk assessment phase, the system extracted leakage data. Real-time data showing a 42% drop in resistivity and a formation pressure of 1.4 MPa were used to calculate the critical leakage threshold using the formula. The dual-judgment model outputs a Level III composite risk assessment result with a 92% confidence level. During the scheme generation and human-machine collaboration phase, the system automatically generates a material formula of high-strength cement grout + 10% quick-setting plugging agent + 5% flexible fiber, along with pumping pressure of 1.8-2.0 MPa and segmented grouting and sealing process parameters. Engineers confirm the scheme using images from the borehole imaging instrument. In the execution and collaborative control phase, construction personnel employ a segmented grouting and sealing process, with each segment grouting length of 15m, a grout mixing time of 15 minutes, and a grout density... During pumping, the pressure range is strictly controlled, and drilling parameters are adjusted synchronously to a drilling pressure of 5000-5500 kN and a rotation speed of [missing information]. During the feedback and model evolution phase, monitoring after 5 hours of waiting showed that the leakage rate had decreased. The water pressure stabilized below 0.5MPa, and the treatment effect met the standards. The 680 sets of full-process data from this construction were fed back to the model. After incremental learning, the correction coefficient for the mining area was optimized to 0.88, and the confidence level for similar working conditions was increased to 96%.

[0050] The following is a comparison table of the effects given in Examples 1, 2, and 3:

[0051] A comparison of the three sets of implementation examples shows that Example 2 resulted in the largest decrease in accident rate and the most significant improvement in borehole formation efficiency. It constructed a real-time data acquisition and processing system using a multi-source monitoring network and edge computing modules. A dual-judgment model was designed, consisting of a hierarchical threshold dynamic correction model driven by formation parameter inversion algorithms and a hybrid risk identification model combining a rule engine and a lightweight decision tree algorithm. This formed a complete technical system encompassing data acquisition, dynamic judgment, human-machine collaborative handling, and a closed-loop evolution of the entire data process. This enabled accurate judgment of composite risks in coal exploration boreholes in soft rock areas, reducing the borehole accident rate from over 30% to below 5%, increasing borehole formation efficiency by 15%-20%, and reducing drilling fluid consumption by 25%. It achieved dynamic risk judgment, customized handling solutions, and a closed-loop technical system. Furthermore, the industrial-grade hardware is adapted to the complex environment of high temperature, high vibration, and strong interference in the well, exhibiting strong environmental adaptability and continuous optimization capabilities.

[0052] The technical scope of this invention is not limited to the content described above. Those skilled in the art can make various modifications and variations to the above embodiments without departing from the technical concept of this invention, and all such modifications and variations should fall within the protection scope of this invention.

Claims

1. A method for graded wall protection and leakage plugging of coal exploration boreholes in soft rock areas based on real-time judgment, characterized in that, Includes the following steps: Data Acquisition: Collect geological data of the target mining area and construction cases of similar soft rock areas. Based on the geological data and construction cases, obtain basic materials. Based on the basic materials, pre-set characteristic parameters of different types of soft rock. Analyze the geological data and construction cases to obtain the dynamic change law of soft rock area geology and complex risk pain points. System construction: Deploy a multi-source monitoring network and select an appropriate edge computing module based on characteristic parameters. Obtain a dual-judgment model by analyzing the dynamic changes in the soft rock area and the complex risk pain points. Start the dual-judgment model and import expert experience rules and basic training data. Data preprocessing: Parameter data is collected synchronously through a multi-source monitoring network, and after noise reduction and standardization by the edge computing module, a downhole dynamic data pool is constructed. Risk assessment is then performed based on the dynamic data pool. Dynamic risk assessment: The dual-assessment model is used to analyze the dynamic data pool, correct the risk threshold, identify the working condition type, output the risk level of hole wall instability and leakage, and generate a support and treatment plan. Solution generation and human-machine collaboration: Based on the risk assessment results, a treatment plan containing material formulas and process parameters is generated, and an execution instruction is formed after manual review and confirmation; Execution and Coordination Control: Execute instructions to complete wall protection and leakage plugging operations, and simultaneously adjust drilling and borehole inclination control parameters according to risk level and working conditions; Effect feedback and model evolution: Monitor the effect of wall protection and leak sealing operations. If the results are not as expected, trigger a second optimization of the solution and feed the full process data back to the dual-judgment model to achieve incremental learning of the model.

2. The method for graded wall protection and leakage sealing of coal exploration boreholes in soft rock areas based on real-time judgment as described in claim 1, characterized in that: The geological data of the target mining area includes geological survey reports, columnar sections of coal seam roof boreholes, soft rock mechanical test data, and groundwater data. Similar soft rock area construction cases cover three types of accidents: leakage and collapse, water inrush and fracture zone, and progressive hidden collapse. The accident triggering parameters, disposal plans, and failure causes are recorded.

3. The method for graded wall protection and leakage sealing of coal exploration boreholes in soft rock areas based on real-time judgment as described in claim 1, characterized in that: The multi-source monitoring network consists of a drill pressure torque sensor, a borehole imaging device, and a logging-while-drilling (LWD) device. The sensor has a measurement accuracy of ±1%, the imaging device has a resolution of 1080P, and the LWD device has a measurement accuracy of ±2%. The devices transmit data synchronously via industrial Ethernet.

4. The method for graded wall protection and leakage sealing of coal exploration boreholes in soft rock areas based on real-time judgment as described in claim 1, characterized in that: The edge computing module is industrial grade, resistant to temperature -20~60℃, vibration 5-50Hz, and processing latency ≤50ms. After being adapted and calibrated with the monitoring network interface, the data conversion error is ≤0.5%.

5. The method for graded wall protection and leakage sealing of coal exploration boreholes in soft rock areas based on real-time judgment as described in claim 1, characterized in that: The dynamic correction model for the graded threshold in the dual-judgment model adopts a formation parameter inversion algorithm, and the formula is as follows: Critical leakage threshold = formation permeability × soft rock compressive strength × mining area correction coefficient, wherein the mining area correction coefficient is 0.

75.

6. The method for graded wall protection and leakage plugging of coal exploration boreholes in soft rock areas based on real-time judgment as described in claim 1, characterized in that: The composite risk identification model adopts a hybrid architecture of rule engine and decision tree algorithm. The rule engine has 12 built-in rules, and the decision tree is trained based on 50 sets of composite risk samples with a depth of 6 layers. The input features include leakage and torque fluctuation.

7. The method for graded wall protection and leakage sealing of coal exploration boreholes in soft rock areas based on real-time judgment as described in claim 1, characterized in that: The monitoring network collects three types of data—engineering, geology, and geophysics—at a frequency of once per minute. Outliers exceeding ±20% are removed using Kalman filtering, and the data are standardized to the 0-1 range using min-max to construct a downhole dynamic data pool.

8. The method for graded wall protection and leakage sealing of coal exploration boreholes in soft rock areas based on real-time judgment as described in claim 1, characterized in that: The supporting response plan prioritizes matching known risk scenarios with a response time ≤10ms. Unmatched items are analyzed using a decision tree, and the risk confidence level is calculated using the following formula: Confidence score = rule matching score × 0.6 + decision tree probability × 0.

4.

9. The method for graded wall protection and leakage plugging of coal exploration boreholes in soft rock areas based on real-time judgment as described in claim 1, characterized in that: The treatment plan includes material formulation and process parameters. The on-site engineer, based on the real-time images from the in-hole imaging instrument, fine-tunes the material ratio and pumping pressure parameters before issuing the execution command.

10. The method for graded wall protection and leakage sealing of coal exploration boreholes in soft rock areas based on real-time judgment as described in claim 1, characterized in that: The effect monitoring includes short-term and long-term indicators. Data from the entire process is fed back to the dual-judgment model to achieve incremental learning. If the target is not met, secondary optimization of material particle size, pumping pressure, and setting time is triggered.

Images