Fine-grained control system for the entire life cycle of drilling

Abstract

The present application provides a refined management and control system for the entire life cycle of a borehole, and relates to the technical field of coal mine information systems. The system includes: a sealing parameter perception unit, which is used to collect sealing sensor data based on a sealing sensor group; a sealing abnormality repair unit, which is used to trigger and start a secondary sealing program when abnormal sealing conditions are detected, and the corresponding sealing optimization strategy is determined by a sealing repair model; an extraction parameter perception unit, which is used to collect gas extraction sensor data based on an extraction sensor group; an extraction effect prediction unit, which is used to predict the gas extraction volume prediction results and gas concentration prediction results; an extraction abnormality warning unit, which is used to generate an extraction warning notification when a preset extraction abnormality condition is detected. Thus, a refined management and control system for the entire life cycle of a borehole is established by performing real-time working status perception of the sealing stage and the extraction stage of the borehole based on multimodal sensing technology.

Description

Technical Field

[0001] The present application relates to the technical field of coal mine information systems, and in particular to a refined management and control system for the entire life cycle of drilling. Background Art

[0002] Gas extraction is an extremely important part of the coal mine production process, especially in high-gas mines. Effective gas extraction is directly related to production safety and economic benefits. In order to prevent accidents caused by gas accumulation, some relevant regulations clearly require that sufficient gas pre-extraction must be carried out before mining operations to ensure that the prescribed safety standards are met. However, the existing gas extraction technology has some obvious deficiencies in various key links and cannot meet the increasingly complex needs of coal mine operations.

[0003] On the one hand, hole sealing is a basic step to ensure the effectiveness of gas extraction, and its quality directly determines the concentration and efficiency of gas extraction during the extraction process. However, traditional hole sealing processes mostly use manual inspections and empirical judgments, lacking systematic detection methods, resulting in unstable hole sealing effects, prone to gas leakage or hole sealing failure, and thus affecting the extraction effect.

[0004] On the other hand, during the gas extraction process, the borehole is in a complex geological environment for a long time and is easily affected by factors such as surrounding stress changes, stratum collapse, and groundwater accumulation, which can cause borehole deformation, blockage, or pipeline leakage. During the operation of the borehole, common faults include water accumulation in the hole, deformation, and pipeline leakage, which will seriously affect the extraction efficiency and even cause the borehole to fail. However, the traditional technology is manual regular inspection, which lacks real-time fault monitoring means and cannot be discovered and handled in time at the early stage of the fault, resulting in many faulty boreholes being scrapped prematurely, shortening their service life.

[0005] In response to the above problems, the industry has not yet proposed a better technical solution. Summary of the invention

[0006] The present application provides a refined management and control system for the entire life cycle of drilling, which is used to at least solve the problems caused by relying on traditional manual inspections and experience-based controls, such as long monitoring intervals, high risks of missed detections, and inability to meet the personalized needs of gas extraction at different stages.

[0007] The embodiment of the present application provides a refined management and control system for the entire life cycle of drilling, including: a sealing parameter perception unit, which is used to collect sealing sensor data based on a sealing sensor group, and the sealing sensor data includes the following sensor parameters: sealing material stress distribution, sealing gas concentration and sealing area negative pressure information; a sealing abnormality repair unit, which is used to trigger and start a secondary sealing program when it is detected that any sensor parameter in the sealing sensor data meets a preset sealing abnormality condition, and input the sealing sensor data into a sealing repair model to determine a corresponding sealing optimization strategy; the sealing optimization strategy includes adjusting the sealing material ratio, the sealing material injection speed and the sealing pressure; the sealing repair model adopts a reinforcement learning model, the state of the reinforcement learning model is defined by the sealing material stress distribution, the sealing gas concentration and the sealing area negative pressure, and the action of the reinforcement learning model is determined by the sealing material ratio, the sealing gas concentration and the sealing area negative pressure. The reward function of the reinforcement learning model is defined by the optimization degree of the sealing effect and the cost of the sealing repair operation; an extraction parameter sensing unit is used to collect gas extraction sensor data based on the extraction sensor group, and the gas extraction sensor data includes pipeline flow parameters, pipeline gas concentration, pipeline ultrasonic signals and pipeline negative pressure information; an extraction effect prediction unit is used to send the gas extraction sensor time series data corresponding to the first time period of history to the extraction effect prediction model to predict the gas extraction volume prediction result and the gas concentration prediction result corresponding to the second time period of the future; the backbone of the extraction effect prediction model adopts LSTM; an extraction abnormality warning unit is used to generate an extraction warning notification when it is detected that the gas extraction volume prediction result or the gas concentration prediction result meets the preset extraction abnormality condition.

[0008] The drilling life cycle refined management and control system provided by this application can produce at least the following technical effects:

[0009] (1) Through the joint use of the sealing parameter sensing unit and the extraction parameter sensing unit, the monitoring sealing and extraction data are integrated to enable the system to have the ability to detect faults throughout the entire life cycle. Common problems in the sealing and extraction process, such as sealing leakage, borehole deformation, and pipeline blockage, can be discovered at the early stage of the fault through the linkage analysis of sensor data. In addition, when the system detects an abnormal situation, it immediately generates an early warning notification to avoid safety hazards caused by potential faults and ensure safe operation throughout the life cycle of gas extraction.

[0010] (2) During the sealing stage, the intelligent sealing and repair model automatically adjusts the sealing material ratio, injection speed and sealing pressure based on the reinforcement learning model to ensure continuous optimization of the sealing quality. When any parameter in the sealing sensor data is abnormal, the system can trigger the secondary sealing program in real time through data feedback and intelligently select the best sealing and repair strategy. Therefore, the stability and durability of the sealing effect are guaranteed through the intelligent dynamic sealing adjustment mechanism.

[0011] (3) During the extraction stage, the extraction effect prediction unit uses the LSTM (Long Short-Term Memory) model to comprehensively analyze multi-dimensional data such as pipeline flow, gas concentration, ultrasonic signals and negative pressure information to predict future gas extraction volume and concentration. It can continuously monitor the pipeline status and accurately predict future extraction volume and gas concentration, which can effectively reduce abnormal extraction volume or safety hazards caused by unpredictable factors.

[0012] Through this technical solution, based on multimodal sensing technology, real-time working status perception is achieved in the sealing and extraction stages of the borehole, and a refined management and control system for the entire life cycle of the borehole is established. This not only improves the stability of the sealing process, but also optimizes the long-term effect of gas extraction, effectively improving the overall safety, efficiency and economic benefits of gas extraction. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0014] Figure 1 A structural block diagram of an example of a drilling life cycle refined management and control system according to an embodiment of the present application is shown;

[0015] Figure 2 A schematic diagram of the system architecture effect of an example of a drilling full life cycle refined management and control system according to an embodiment of the present application is shown;

[0016] Figure 3 An on-site installation diagram showing an example of an installation module for a high-position drilling site according to an embodiment of the present application;

[0017] Figure 4 A schematic diagram showing an example of a state transition action in a reinforcement learning model;

[0018] Figure 5A schematic diagram of structural connection of an example of a drilling full life cycle refined management and control system according to an embodiment of the present application is shown;

[0019] Figure 6 A schematic diagram of the structural connection of an example of a extraction effect prediction model according to an embodiment of the present application is shown;

[0020] Figure 7 An operation flow chart of an example of a drilling life cycle refined management and control method according to an embodiment of the present application is shown;

[0021] Figure 8 It is a schematic structural diagram of an embodiment of an electronic device of the present application. DETAILED DESCRIPTION

[0022] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0023] In the technical solution of this application, the collection, storage, use, processing, transmission, provision and disclosure of user personal information involved shall comply with the provisions of relevant laws and regulations and shall not violate public order and good morals.

[0024] Figure 1 A structural block diagram of an example of a drilling life cycle refined management and control system according to an embodiment of the present application is shown.

[0025] Regarding the carrier of the system of the embodiment of the present application, it can be any controller or processor with computing or processing capabilities. Through the real-time collection of multi-dimensional sensor data, the intelligent optimization of the reinforcement learning model, and the introduction of prediction and early warning mechanisms, the sealing effect is improved and the life of the borehole is extended. Through intelligent means, human intervention is reduced, the safety and economy of the system are improved, and the refined management and control of the entire life cycle of the gas extraction system is achieved.

[0026] In some examples, it can be integrated and configured in an electronic device or terminal through software, hardware, or a combination of software and hardware, and the type of the terminal or electronic device can be diverse, such as a mobile phone, a tablet computer, or a desktop computer, etc.

[0027] like Figure 1 As shown, the drilling life cycle refined management and control system 100 includes a sealing parameter sensing unit 110, a sealing abnormality repair unit 120, a drainage parameter sensing unit 130 and a drainage effect prediction unit 140.

[0028] The sealing parameter sensing unit 110 is used to collect sealing sensor data based on the sealing sensor group.

[0029] Here, the sealing sensor data includes the following sensing parameters: stress distribution of sealing materials, sealing gas concentration and negative pressure information of the sealing area. In some embodiments, the sealing sensor group is installed at a key position in the sealing area to monitor the physical and chemical changes in the sealing process in real time. The sealing sensor group includes a stress sensor, a gas concentration sensor and a negative pressure sensor. The stress sensor is used to detect the stress distribution of the sealing material under pressure to ensure that the sealing material is evenly applied to the borehole wall; the gas concentration sensor is used to detect the gas concentration in the sealing area to prevent gas leakage; the negative pressure sensor is used to monitor the air pressure in the sealing area to ensure the stability of the negative pressure in the sealing area and prevent external gas from entering the sealing area. Furthermore, the data collected by the sensor is transmitted to the central control system by wireless or wired means, so that the control system uses a data analysis algorithm to evaluate the sealing quality and determine whether the sealing is successful or abnormal.

[0030] Through the deployment of the sealing sensor group and the real-time transmission of data, the sealing effect can be monitored and evaluated in real time, avoiding the lag of manual inspection. Real-time monitoring of stress, gas concentration and negative pressure information ensures comprehensive control of the sealing process, effectively preventing the occurrence of sealing failure, gas leakage and other problems, and greatly improving the reliability and long-term stability of the sealing.

[0031] Figure 2 The following is a schematic diagram of the system architecture effect of an example of a refined management and control system for the entire life cycle of a borehole according to an embodiment of the present application. The information hardware system is a prerequisite for the operation of the system software, the transmission of relevant information, and the publication of evaluation results, and mainly includes servers, clients, office networks, switches, etc. Here, drilling pipeline monitoring points are deployed in two high-position drilling sites, and the working status and real-time parameters of the measuring points are collected through sensor equipment. Reasonable deployment of measuring points plays a decisive role in the evaluation of the effect of coal mine gas extraction.

[0032] In some examples of the embodiments of the present application, the sealing sensor group includes a sealing material stress sensor installed in the sealing area, a first gas concentration sensor having a first probe installed on the inside of the sealing area and a second probe installed on the outside of the sealing area, and a first negative pressure sensor installed in the extraction pipe near the sealing.

[0033] Here, by installing a material stress sensor in the sealing area, the stress distribution of the sealing material at different positions can be monitored in real time to ensure uniform material stress, which helps to improve the stability and durability of the seal. Through the first gas concentration sensor, the sealing effect of the seal can be more accurately judged by comparing the gas concentrations on the inside and outside. The inner probe measures the concentration of the gas inside the seal, and the outer probe measures the concentration difference outside the seal. By comparing the two, it can be determined in time whether there is a leak in the seal, thereby improving the reliability of sealing monitoring. Through the first negative pressure sensor close to the seal, the pressure state of the sealing area can be fed back in time to ensure that the sealing material can effectively seal the gas source. Real-time monitoring of negative pressure ensures the smooth discharge of gas and reduces the risk of gas reflux.

[0034] Figure 3 An on-site installation diagram of an example of an installation module for a high-position drilling site according to an embodiment of the present application is shown.

[0035] like Figure 3 As shown, there are 12 boreholes in the two high-position drilling sites of the No. 2 Coal Mine, and each drilling site is equipped with a manifold. Therefore, each drilling site is equipped with 12 single-hole gas extraction pipeline comprehensive parameter measuring instruments, 12 single-hole gas pipeline electric valves, 1 manifold gas extraction pipeline comprehensive parameter measuring instrument, and 1 manifold electric valve; in order to achieve better data transmission and remote control, each drilling site is equipped with 6 mine-use intrinsically safe wireless gateways, of which 2 are used for wireless transmission of monitoring data and 4 are used to control electric valves.

[0036] The sealing abnormality repair unit 120 is used to trigger the secondary sealing procedure when it is detected that any sensor parameter in the sealing sensor data meets the preset sealing abnormality condition, and input the sealing sensor data into the sealing repair model to determine the corresponding sealing optimization strategy. Here, the sealing optimization strategy includes adjusting the sealing material ratio, the sealing material injection speed and the sealing pressure. The sealing repair model adopts a reinforcement learning model.

[0037] Figure 4 A schematic diagram showing an example of state transition actions in a reinforcement learning model.

[0038] like Figure 4 As shown, it involves multiple states S The action corresponding to the state transition in the state space of 1~Sn, for example a 1 means from S 1 to S 2. The action of state transition, a 2 means from S 2 to S 1 action, a 3 means from S 1 toS 3 state transition, and so on. Here, the corresponding state transition can occur based on the strategy, and each state transition strategy can be used to cause different transitions. S 1 transfer strategy, action can occur a 2 or a 3.

[0039] It should be noted that the range of states that a state can transfer to (also called transferable states) may be restricted or conditional, for example S 1~ S None of the 3 S 4~ S n occurs between states, and for states S 1 can be transferred to the state S 2 and S 3, etc.

[0040] In some embodiments, each action has a corresponding action reward, and each action reward can be determined based on a preset reward function. Generally, if the transfer reward is larger, the transfer action is considered more valuable, and the system will give priority to executing this action. a The reward corresponding to 1 is greater than a The reward corresponding to 3 indicates the transfer action. a 1 is more valuable.

[0041] In some examples of the embodiments of the present application, the state of the reinforcement learning model is defined by the stress distribution of the sealing material, the sealing gas concentration and the negative pressure of the sealing area, the action of the reinforcement learning model is defined by the adjustment range of the sealing material ratio, the sealing material injection speed and the sealing pressure, and the reward function of the reinforcement learning model is defined by the optimization degree of the sealing effect and the cost of the sealing repair operation. Therefore, the reinforcement learning model is motivated to make a sealing optimization strategy that can simultaneously support the maximization of the optimization degree of the sealing effect and the minimization of the sealing repair operation cost, so as to realize the intelligent agent to find the best sealing optimization strategy.

[0042] Through the embodiments of the present application, combined with the optimization decision of the reinforcement learning model, the system can continuously optimize the sealing and repair strategy, making it adaptive in a dynamic network environment. By adjusting the material ratio in real time, the material performance is maximized under different geological conditions and the sealing effect is enhanced. According to the feedback of sensor data on gas concentration, pressure and other parameters, the injection speed of the sealing material is dynamically adjusted to ensure that the sealing material can be evenly distributed to prevent material waste or insufficient filling. In addition, according to the stress sensor data, the sealing pressure is optimized to ensure that the material is completely filled into the sealing area to avoid material leakage or incomplete sealing.

[0043] The extraction parameter sensing unit 130 is used to collect gas extraction sensor data based on the extraction sensor group.

[0044] It should be noted that the evaluation of gas extraction effect is an important basis for ensuring the smooth progress of extraction projects, but the existing extraction effect evaluation mainly relies on the actual measurement of gas concentration after extraction, and cannot predict and adjust the extraction effect of the borehole before or during construction. This leads to unreasonable layout of some boreholes or improper parameter settings, affecting the extraction efficiency. In addition, during the extraction process, gas concentration, pressure and other parameters will fluctuate with the change of time and environmental conditions. The existing evaluation system cannot track and adjust these parameters in real time, resulting in fluctuations in extraction effect. It is easy to carry out mining operations before the extraction effect meets the standards, which poses a major safety hazard.

[0045] More specifically, the gas extraction sensor data includes pipeline flow parameters, pipeline gas concentration, pipeline ultrasonic signals and pipeline negative pressure information.

[0046] In some examples of the embodiments of the present application, the extraction sensor group includes a flow sensor installed in the middle and end of the extraction pipeline, a second gas concentration sensor having a third probe installed in the middle of the extraction pipeline and a fourth probe at the borehole mouth, an ultrasonic sensor installed at the pipe connection or bend, and a second negative pressure sensor installed in the middle and end of the extraction pipeline.

[0047] By installing flow sensors in the middle and end of the extraction pipeline, the changes in gas flow in the extraction pipeline can be accurately monitored, and the dynamic changes in flow during the extraction process can be more comprehensively understood, especially when the gas flow at the end changes dramatically, early warning can be given. By installing the third and fourth probes in the middle and borehole respectively through the second gas concentration sensor, it is helpful to accurately monitor the gas concentration from the borehole to the pipeline, monitor the spatial distribution of gas concentration, better judge the extraction effect, and optimize the extraction strategy. By installing ultrasonic sensors at the connection or bend of the pipeline, the changes in the flow state of the airflow can be accurately detected, especially the gas behavior under complex pipeline geometry. The airflow anomalies in the pipeline (such as airflow blockage or turbulence) can be identified in advance, and the high-frequency sound waves generated by the leak can be used to quickly locate the leak point by analyzing the frequency and intensity of the ultrasonic wave. The second negative pressure sensor monitors the changes in negative pressure in the middle and end of the extraction pipeline to ensure that the negative pressure is evenly distributed throughout the pipeline.

[0048] The extraction effect prediction unit 140 is used to send the gas extraction sensor time series data corresponding to the first historical time period to the extraction effect prediction model to predict the gas extraction volume prediction result and the gas concentration prediction result corresponding to the second future time period.

[0049] Here, the backbone of the extraction effect prediction model uses LSTM. The LSTM model has the advantage of processing time series data. It can learn the changing trends of gas extraction volume and concentration from a large amount of historical data, and predict future changes in extraction volume and gas concentration, so that the system can judge the future extraction situation in advance and detect potential extraction anomalies in time.

[0050] Combination Figure 2 and 3 For example, the dynamic source evaluation of gas extraction effect requires a sufficient number and reasonable deployment of measuring points, which can effectively ensure the monitoring and measurement of coal mine gas extraction areas. Specifically, borehole pipeline monitoring points are deployed in two high-level drilling sites to analyze and evaluate the gas extraction effect of the boreholes, and intelligent valves are deployed on the extraction pipelines to intelligently control the borehole extraction pipelines. Furthermore, unit measuring points and intelligent valves can be deployed at the high-level drilling site pipeline aggregation point to intelligently control the evaluation unit extraction pipelines.

[0051] Figure 4 The following is a schematic diagram showing an example of an interface for intelligently judging the effect of gas extraction according to an embodiment of the present application. Figure 4 As shown in the figure, the system has established an intelligent evaluation system for the entire process of gas extraction, realizing intelligent evaluation analysis and management of the entire life cycle, including definition of effective control range of drilling holes, evaluation of uniformity of pre-extraction drilling holes, division of pre-extraction compliance evaluation units, dynamic calculation of pre-extraction effect indicators, and management of measured indicators.

[0052] The extraction abnormality warning unit 150 is used to generate an extraction warning notification when it is detected that the gas extraction volume prediction result or the gas concentration prediction result meets the preset extraction abnormality condition.

[0053] In some examples of the embodiments of the present application, when the future extraction volume predicted by the extraction effect prediction unit 140 is too low or the gas concentration is too high, the system can make adjustments and interventions before the problem worsens, which helps to optimize the extraction plan, avoid the occurrence of safety hazards, and improve the efficiency of gas resource extraction. Here, the extraction abnormality warning unit sends an early warning notification to relevant personnel through an intelligent terminal or a remote monitoring system to prompt possible extraction failures. The operator can manually intervene in the system based on the early warning information, or the system can automatically adjust the extraction strategy to avoid serious failures.

[0054] In some embodiments, the extraction abnormality warning unit 150 is used to obtain the extraction pipeline identification corresponding to the extraction abnormality when the gas extraction volume prediction result or the gas concentration prediction result meets the preset extraction abnormality condition, and use the obtained extraction pipeline identification to query the matching target management terminal information from the preset identification terminal association table, and send the extraction warning notification to the target management terminal device corresponding to the target management terminal information. Here, the identification terminal association table pre-stores multiple management terminal information and corresponding calibrated extraction pipeline identifications. Exemplarily, each extraction pipeline has a unique identification code in the system, which is not only used to identify the pipeline itself, but also used to associate the management terminal. After detecting an abnormal situation, the system will automatically obtain the identification code of the extraction pipeline with the problem to ensure that subsequent operations can accurately locate the corresponding pipeline and responsible person. Therefore, through the automated early warning notification system, relevant managers can receive accurate fault information in a timely manner and take countermeasures quickly, effectively improving the fault response speed and reducing safety accidents or production losses caused by human delays.

[0055] The core of the detailed description of the sealing and repair model is the selection and design of the reward function of the reinforcement learning model. As described above, the reward function is defined by the optimization degree of the sealing effect and the cost of the sealing and repair operation. Exemplarily, the sealing effect is measured by the reduction of gas concentration after sealing, the uniformity of the stress distribution of the sealing material, and the stability of the negative pressure. The reward function can positively reward the uniform filling of the sealing material, the significant decrease in gas concentration, and the negative pressure kept within a safe range; negatively reward the abnormal increase in gas concentration, uneven material stress distribution, or excessive fluctuations in negative pressure. In addition, the reward function also introduces an evaluation of the cost of the sealing and repair operation, which encourages the system to complete the repair with as few sealing operations as possible and at a low economic cost, avoiding excessive consumption of sealing materials and energy.

[0056] Preferably, a system instability penalty term is introduced into the reward function of the reinforcement learning model, which is used to suppress the drastic fluctuations of material stress and negative pressure during the sealing process and ensure the stability of the operation. In this way, while optimizing the sealing effect, the stability of the material stress and negative pressure changes during the sealing process is ensured to avoid unnecessary fluctuations.

[0057] More specifically, the reward function of the reinforcement learning model is expressed as follows:

[0058] , Formula (1)

[0059] , Formula (2)

[0060] , Formula (3)

[0061] , Formula (4)

[0062] In the formula, Represents the reward value, represents the weight coefficient for optimizing the sealing effect, represents the weight coefficient of the operation cost, The weight coefficient representing the penalty for system instability; represents the optimization increment of the sealing effect, is the operating cost; is the system instability penalty term, which represents the volatility of material stress and negative pressure during the sealing process; They represent the weight coefficients used to measure the importance of gas concentration, material stress and negative pressure in the sealing effect; It is the change in gas concentration, indicating the degree of reduction in gas leakage; It is the change in the stress distribution of the sealing material, indicating that the uniformity of the force on the sealing material has improved; is the optimization degree of sealing negative pressure, which indicates the matching degree between the negative pressure in the sealing area and the ideal negative pressure value in the sealing area; is the material cost, and is the energy cost; is the penalty coefficient for material stress fluctuations, is the penalty coefficient for negative pressure fluctuation; is the variance of the stress distribution of the sealing material, which indicates the stress fluctuation of the sealing material at different positions; is the negative pressure fluctuation variance, which indicates the fluctuation of negative pressure in the sealing area.

[0063] Regarding the explanation of the above formula (1), the reward function Comprehensively measure the improvement of sealing effect ( )、Operation cost( ) and system stability ( ). Through the weight coefficient , and , the model is able to adjust the importance of each factor in the total reward.

[0064] According to the above formula (2), the sealing effect increment It is one of the most important indicators in the sealing operation, which is determined by three key factors: gas concentration, material stress distribution and negative pressure change. ): After sealing, the reduction of gas concentration is a direct reflection of the sealing effect. The lower the concentration, the better the sealing. Stress distribution uniformity ( ): The more uniform the material stress is, the more stable the sealing effect will be. Uneven stress on the sealing material will lead to a decrease in the sealing effect. Negative pressure adjustment ( ): Reasonable negative pressure helps to improve the efficiency of gas emission, and the adjustment of negative pressure reflects the control ability of the extraction system.

[0065] By increasing the sealing effect in the reward function The model can accurately measure the actual effect of the sealing process, and maximize the sealing effect by continuously adjusting the sealing material ratio, injection speed and pressure, ensuring that the gas concentration is reduced and the sealing effect of the sealing area is more stable. In this way, the optimization of the sealing process can effectively reduce gas leakage, ensure that the sealing area reaches the best sealing state, and improve the efficiency and safety of gas extraction. According to the formula, the model can continuously adjust the negative pressure state of the sealing area to make it close to the ideal negative pressure, ensuring that the gas can be effectively discharged.

[0066] Regarding the explanation of the above formula (3), the operating cost Reflects the resource consumption when completing the sealing operation, with the aim of enabling the system to optimize the sealing effect while minimizing the waste of materials and energy. ): The sealing material consumed during the sealing process. The higher the cost, the lower the reward. Energy consumption ( ): The energy consumption of the sealing equipment, such as electricity or energy for mechanical operation, the greater the consumption, the heavier the penalty.

[0067] By introducing the operation cost into the reward function , including the consumption of sealing materials and energy. By optimizing consumption, the system can minimize the waste of materials and energy without sacrificing the sealing effect.

[0068] According to the above explanation of formula (4), the system instability penalty term This is to make the system operation more stable and avoid drastic fluctuations that may lead to sealing failure or reduced efficiency. ): The force fluctuation of the sealing material at different positions should be as small as possible. The greater the fluctuation, the more likely the sealing material will experience stress concentration, leading to sealing failure. Negative pressure fluctuation ( ): Drastic changes in negative pressure will also affect the stability of the sealing process and the extraction effect. Smooth negative pressure control can ensure the effective discharge of gas.

[0069] By introducing a penalty term for system instability The model can suppress the drastic fluctuations of material stress and negative pressure during the sealing process. The smaller the stress and negative pressure fluctuations, the smoother the sealing process, thereby improving the sealing success rate and reducing the risk of sealing failure.

[0070] Through the reward function design of the reinforcement learning model in the embodiment of the present application, the system can adaptively adjust the operating parameters in a complex sealing environment, gradually optimize the sealing process, automatically select the optimal sealing strategy, significantly reduce manual intervention, realize intelligent sealing and repair solutions, and maintain efficient operation under different geological conditions based on the adaptive adjustment capability.

[0071] Figure 5 A schematic diagram of the structural connections of an example of a drilling life cycle refined management and control system according to an embodiment of the present application is shown.

[0072] like Figure 5 As shown, the drilling life cycle refined management and control system 500 includes a sealing parameter perception unit 510, a sealing abnormality repair unit 520, an extraction parameter perception unit 530, an extraction effect prediction unit 540, an extraction abnormality warning unit 550 and an extraction abnormality repair unit 560.

[0073] For details and effects of the sealing parameter sensing unit 510, the sealing abnormality repairing unit 520, the extraction parameter sensing unit 530, the extraction effect prediction unit 540 and the extraction abnormality warning unit 550, please refer to the description in combination with other examples above, which will not be repeated here.

[0074] In some examples of the embodiments of the present application, the extraction abnormality repair unit 560 is used to input the gas extraction volume prediction result and the gas concentration prediction result into the extraction parameter optimization model to determine the corresponding extraction parameter optimization strategy when it is detected that the gas extraction volume prediction result or the gas concentration prediction result meets the preset extraction abnormality condition.

[0075] In some embodiments, when the system detects an abnormal trend in gas extraction volume or gas concentration, the system will use these two parameters (gas extraction volume prediction results and gas concentration prediction results) as input and send them to the extraction parameter optimization model, so that the model will evaluate the current extraction system status based on these input data and provide a basis for the system to adjust the extraction parameters.

[0076] Specifically, the extraction parameter optimization model adopts an adaptive Bayesian optimization model, the extraction parameter optimization strategy includes the target negative pressure parameter and target gas flow to be adjusted, and the adaptive Bayesian optimization model adopts an adaptive search strategy to dynamically adjust the search range for the negative pressure parameter and gas flow according to the input data.

[0077] Here, the extraction parameter optimization model uses an adaptive Bayesian optimization model, which infers the optimal extraction parameter combination (such as negative pressure and gas flow) through the relationship between historical data and current input data. It should be noted that the Bayesian optimization model is particularly suitable for high-dimensional, nonlinear and uncertain parameter optimization scenarios, and can find the optimal solution with a small amount of data and a limited number of searches.

[0078] Through repeated iterations of the adaptive Bayesian optimization model, the system will gradually converge to the ideal combination of negative pressure and gas concentration parameters. The specific manifestation of the adaptive process is as follows: in the initial stage, the adjustment range of gas concentration and negative pressure is large. As the optimization process proceeds, the fluctuation range will gradually narrow and tend to stabilize. In addition, during the gas extraction process, geological conditions and other external factors may change, but the system can automatically adjust the parameter combination according to the changes in sensor data to ensure that the negative pressure and concentration always remain within a reasonable range.

[0079] More specifically, adaptive Bayesian optimization updates the best estimate of the extraction parameters at each iteration by using a Gaussian process-based model. The model automatically adjusts the search range for negative pressure parameters and gas flow. As the amount of collected data increases, the model adaptively adjusts the range and accuracy of the parameter search, thereby gradually approaching the optimal extraction parameter configuration. In addition, the objective function of the Bayesian optimization model can also be set according to demand, such as maximizing the gas extraction volume and minimizing the gas concentration, or maximizing the match between the gas extraction volume and gas concentration and the corresponding ideal value.

[0080] Therefore, by using limited historical data to search for optimal parameters, the model can quickly adapt to the complex and changing environment on site. Compared with traditional fixed parameter settings, Bayesian optimization can be dynamically adjusted according to environmental changes, greatly improving the flexibility and adaptability of the system, and ensuring that the optimization strategy can be applied to the extraction system in real time and quickly.

[0081] As a further optimization of the embodiment of the present application, the adaptive Bayesian optimization model uses the negative pressure parameter and the gas flow rate to define the optimization parameter combination of the Bayesian optimization model. θ .

[0082] The adaptive search strategy relies on the gradient of the objective function , which is used to calculate the deviation of the current negative pressure and gas concentration from the ideal value, and dynamically adjust the parameters accordingly. The goal of optimization is not to maximize or minimize a single indicator, but to maintain these parameters within a normal range.

[0083] Specifically, each time an adjustment is made, the system adjusts the parameter combination according to the calculated gradient. The formula for adaptive parameter iterative search is:

[0084] , Formula (5)

[0085] In the formula, is the new parameter combination used in the next iteration, is the parameter combination used in the current round; It is the gradient of the objective function, which measures the gap between the current parameter combination and the ideal value; is the step size factor, which is used to control the update rate of the parameter combination.

[0086] In order to adjust the current negative pressure and gas concentration to make them close to the target value, the system needs to calculate the deviation between the current parameters and the ideal value through the objective function.

[0087] Specifically, the objective function of Bayesian optimization is It is expressed as:

[0088] , Formula (6)

[0089] In the formula, and They represent the ideal extraction volume and the ideal gas concentration, respectively. and They represent the current round of extraction volume and the current round of gas concentration, respectively, and the initial round of extraction volume and gas concentration are set by the gas extraction volume prediction result and gas concentration prediction result, respectively; and They respectively represent the fluctuation tolerance of extraction volume and gas concentration.

[0090] For example, if Deviation from target value If the magnitude is large, the system will adjust by reducing or increasing the negative pressure. And, if Exceed or fall below If the amplitude is large, the system will control the concentration by controlling the gas flow or increasing / decreasing the extraction intensity.

[0091] Step Length It is the key factor in controlling the speed of parameter adjustment. In practical applications, the step size needs to be dynamically adjusted according to the current state of the system.

[0092] , Formula (7)

[0093] In the formula, Represents a preset constant, Indicates the initial step value; is the number of optimization iterations, and as the number of iterations increases, the step size coefficient Gradually decrease accordingly.

[0094] It should be noted that in the initial stage of the optimization process, the system sets the initial parameter values ​​according to the gas extraction data collected by the sensor (such as negative pressure, gas concentration, gas flow, etc.). Since the deviation degree between the current parameters and the target range may not be known at the beginning, the adjustment step size in the initial stage is It is usually set larger so that it can converge quickly to a reasonable interval. In addition, as the system gradually approaches the target interval, the step size It should be reduced gradually to avoid excessive adjustment of the system, which may lead to repeated fluctuations or instability.

[0095] The optimal optimization parameter combination corresponding to the determined extraction parameter optimization strategy It is expressed by the following formula:

[0096] , formula (8)

[0097] In the formula, It means to find the parameter combination that minimizes the function through multiple iterative searches. θ , and use it as the optimal optimization parameter combination .

[0098] Here, the ultimate optimization goal is to find a parameter combination (optimal target negative pressure parameters and target gas flow rate) so that the system's extraction volume and gas concentration can match the ideal value to the maximum extent, that is, through multiple iterative optimizations, the extraction volume and gas concentration are stabilized within a safe range.

[0099] Through the adaptive Bayesian optimization model of the embodiment of the present application, the system can dynamically adjust the negative pressure parameters and gas flow, and adjust the fluctuation range of the gas extraction volume and gas concentration according to real-time data. The model ensures that the gas extraction volume and gas concentration can accurately match the preset ideal target value through the adaptive adjustment of the objective function and step size, avoiding the negative pressure and gas concentration from exceeding the safe range, thereby reducing the occurrence of excessive gas extraction, too high or too low concentration, and improving the extraction efficiency. In addition, the model can adaptively control the negative pressure parameters and gas flow during the extraction process, can find the optimal extraction parameters under different geological conditions, and can effectively improve the operating stability of the system.

[0100] Figure 6 A schematic diagram of the structural connection of an example of a extraction effect prediction model according to an embodiment of the present application is shown.

[0101] like Figure 6 As shown, the extraction effect prediction model 600 adopts the LSTM of the hierarchical model architecture, and the extraction effect prediction model 600 includes an input layer 610, a local hierarchical layer 620, a local LSTM layer 630, a global fusion layer 640, a global LSTM layer 650 and an output layer 660.

[0102] The input layer 610 is used to receive gas extraction sensor time series data.

[0103] The local hierarchical layer 620 is used to process the gas extraction sensor time series data through convolution operations to extract the local features of each mode:

[0104] , Formula (9)

[0105] In the formula, Indicates the gas extraction sensor time series data corresponding to the time step The sensor data collected is a collection of Indicates from The extracted The local convolution features of Represents the convolution kernel, which is used to extract local features. Represents the convolution operation symbol.

[0106] Here, a convolutional layer (CNN) is introduced in the initial stage to extract local features from the time series data of gas extraction, such as local gas flow and pressure changes. These local features reflect the gas extraction status in the short term, which helps to improve the model's perception of small fluctuations and local trends and enhance the robustness of the model in complex dynamic environments.

[0107] The local LSTM layer 630 is used to capture the temporal dependencies of each modality data in a short period of time by learning the time series data:

[0108] , Formula (10)

[0109] In the formula, is at the time step The local features of short-term temporal dependence represent the local features of short-term temporal dependence at time step Local LSTM output features; Represents a local LSTM unit, which is used to learn the temporal dependencies of local features; Indicates that at time step short-term local characteristics.

[0110] Here, each type of sensor data (such as gas concentration, flow rate, etc.) is processed by an independent LSTM unit to capture its short-term temporal dependency, so that each local LSTM unit outputs the short-term modal feature sequence of the corresponding sensor. In addition, the data of each sensor passes through its own local LSTM network to extract the local feature sequence. The processing of each sensor data is parallel and they will not affect each other.

[0111] The global fusion layer 640 is used to fuse the local time series information for each time step output by the local LSTM layer to generate corresponding global fusion features.

[0112] Here, the global features are integrated using the attention mechanism through the global fusion layer 640. By calculating the attention weights, the model can automatically adjust the attention to each time step according to the task, focusing on the importance of local features at different times to the current prediction.

[0113] Specifically, by introducing the attention mechanism, the feature similarity between the current time step and the historical time step is calculated, and these features are fused in a weighted manner. The attention mechanism can automatically identify which time steps are more important in the gas extraction process and give them greater weights. As a result, the model's global feature integration capability is enhanced, and it can dynamically focus on important time series data, improving the accuracy of gas extraction volume and concentration prediction.

[0114] The global fusion layer is used to perform the following operations:

[0115] For each time step , calculate the historical time step Similarity score with the current time step :

[0116] , Formula (11)

[0117] In the formula, Represents the time step With historical time step The feature similarity score between them is used to calculate the attention weight; represents the scoring function, which is used to calculate the similarity between the features of two time steps through the dot product algorithm; Represents the time step short-term local characteristics.

[0118] Score feature similarity The attention weight is obtained by normalizing the softmax function :

[0119] , Formula (12)

[0120] In the formula, Indicates The attention weights of time steps are normalized to sum to 1; express The exponential form of N Represents the total number of time steps corresponding to the gas extraction sensor time series data, represents the exponential summation of feature similarity scores over all time steps.

[0121] Use the attention weights to perform weighted integration on the local features of the historical time steps to obtain the global fusion features:

[0122] , Formula (13)

[0123] In the formula, is at the time step The global fusion features, Represents the time step The attention weight.

[0124] Through the embodiments of the present application, a global fusion layer is used to weight and integrate local features of different time periods into global features, so that the model can better identify complex patterns across time periods in the gas extraction process, improve the model's global understanding of the entire gas extraction process, and improve the model's ability to handle dynamic characteristics of multiple time periods under complex working conditions, thereby ensuring the prediction stability of the system in long-term operation.

[0125] The global LSTM layer 650 is used to capture the long-term trend of the entire gas extraction system through a long time window to determine the corresponding global LSTM output features:

[0126] , Formula (14)

[0127] In the formula, Indicates that at time step The global LSTM output features contain long-term dependent system states; Represents the global LSTM unit, which is used to process the temporal dependencies of global features; Indicates that at time step The global LSTM output features of .

[0128] Through the global LSTM unit, the fused and merged global features are learned in time series. Through layered processing, not only the short-term local features of the sensor are captured, but also the long-term dynamics of the entire system are learned through global feature fusion, ensuring the accuracy and comprehensiveness of the final output results.

[0129] Through the embodiment of the present application, a hierarchical architecture of local LSTM and global LSTM layers is introduced to capture short-term and long-term temporal dependencies respectively. The local LSTM layer focuses on capturing short-term unimodal features in the gas extraction system, such as short-term fluctuations in gas flow and concentration, while the global LSTM layer captures the long-term trend of the system by integrating global features. As a result, the model's ability to capture temporal dependencies is improved, and it can handle short-term fluctuations and long-term trend changes at the same time, enabling the system to respond more effectively to complex trends in the gas extraction process.

[0130] The output layer 660 is used to predict the gas extraction volume prediction result and the gas concentration prediction result corresponding to the second time period in the future according to the output result of the global LSTM layer:

[0131] , Formula (15)

[0132] In the formula, Indicates the predicted value for the second time period in the future, including the predicted results of gas extraction volume and gas concentration; and Represent the weight matrix and bias term of the output layer respectively.

[0133] Through the embodiments of the present application, the multi-level structure optimization design is used to quickly adapt to the dynamic changes of real-time sensor data in the gas extraction system, and adaptively adjust according to the data changes, thereby improving the real-time response capability of the system, and being able to quickly respond to abnormal situations in the gas extraction system, such as a sudden increase in gas concentration or too low negative pressure, to ensure the safety and stability of the system. In this way, more intelligent adaptive prediction is achieved, the need for human intervention is reduced, and the degree of automation of the system is improved.

[0134] Figure 7 An operational flowchart of an example of a method for refined management and control of the entire life cycle of drilling according to an embodiment of the present application is shown.

[0135] like Figure 7 As shown, in step S710, sealing hole sensing data is collected based on the sealing hole sensor group.

[0136] Here, the sealing sensing data includes the following sensing parameters: sealing material stress distribution, sealing gas concentration and sealing area negative pressure information.

[0137] In step S720, when it is detected that any sensor parameter in the sealing sensor data meets the preset sealing abnormality condition, the secondary sealing program is triggered and the sealing sensor data is input into the sealing repair model to determine the corresponding sealing optimization strategy.

[0138] Here, the sealing optimization strategy includes adjusting the sealing material ratio, sealing material injection speed and sealing pressure; the sealing repair model adopts a reinforcement learning model, and the state of the reinforcement learning model is defined by the stress distribution of the sealing material, the sealing gas concentration and the negative pressure in the sealing area. The action of the reinforcement learning model is defined by the adjustment range of the sealing material ratio, the sealing material injection speed and the sealing pressure. The reward function of the reinforcement learning model is defined by the optimization degree of the sealing effect and the sealing repair operation cost.

[0139] In step S730, gas extraction sensor data is collected based on the extraction sensor group.

[0140] Here, the gas extraction sensor data includes pipeline flow parameters, pipeline gas concentration, pipeline ultrasonic signals and pipeline negative pressure information.

[0141] In step S740, the gas extraction sensor time series data corresponding to the first historical time period is sent to the extraction effect prediction model to predict the gas extraction volume prediction result and the gas concentration prediction result corresponding to the second future time period.

[0142] Here, the backbone of the extraction effect prediction model uses LSTM;

[0143] In step S750, when it is detected that the gas extraction volume prediction result or the gas concentration prediction result meets the preset extraction abnormality condition, an extraction warning notification is generated.

[0144] For more details and technical effects of the method for refined control of the entire life cycle of drilling provided in the embodiment of the present application, reference can be made to and combined with the description of the processing details of the refined control system for the entire life cycle of drilling in other embodiments above, and the corresponding technical effects can be achieved.

[0145] It should be noted that, for the aforementioned method embodiments, for the sake of simplicity of description, they are all expressed as a series of actions combined, but those skilled in the art should be aware that the present application is not limited by the described order of actions, because according to the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present application. In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, please refer to the relevant description of other embodiments.

[0146] In some embodiments, an embodiment of the present application provides a non-volatile computer-readable storage medium, in which one or more programs including execution instructions are stored, and the execution instructions can be read and executed by an electronic device (including but not limited to a computer, a server, or a network device, etc.) to execute the steps of any of the above-mentioned methods for refined management and control of the entire life cycle of drilling in the present application.

[0147] In some embodiments, the embodiments of the present application also provide a computer program product, which includes a computer program stored on a non-volatile computer-readable storage medium, and the computer program includes program instructions. When the program instructions are executed by a computer, the computer executes the steps of any one of the above-mentioned methods for refined management and control of the entire life cycle of drilling.

[0148] In some embodiments, the embodiments of the present application also provide an electronic device, comprising: at least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the steps of the method for refined management and control of the entire life cycle of drilling.

[0149] Figure 8 is a schematic diagram of the hardware structure of an electronic device for executing a method for fine-grained control of the entire life cycle of drilling provided by another embodiment of the present application, such as Figure 8 As shown, the device includes:

[0150] One or more processors 810 and memory 820, Figure 8 A processor 810 is taken as an example.

[0151] The device for executing the method for refined management and control of the entire life cycle of drilling may also include: an input device 830 and an output device 840 .

[0152] The processor 810, the memory 820, the input device 830 and the output device 840 may be connected via a bus or other means. Figure 8 The example of connecting through bus is taken in the following.

[0153] The memory 820, as a non-volatile computer-readable storage medium, can be used to store non-volatile software programs, non-volatile computer executable programs and modules, such as program instructions / modules corresponding to the drilling life cycle fine control method in the embodiment of the present application. The processor 810 executes various functional applications and data processing of the server by running the non-volatile software programs, instructions and modules stored in the memory 820, that is, the drilling life cycle fine control method in the above method embodiment is implemented.

[0154] The memory 820 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created according to the use of the electronic device, etc. In addition, the memory 820 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device. In some embodiments, the memory 820 may optionally include a memory remotely arranged relative to the processor 810, and these remote memories may be connected to the electronic device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0155] The input device 830 may receive input digital or character information and generate signals related to user settings and function control of the electronic device. The output device 840 may include a display device such as a display screen.

[0156] The one or more modules are stored in the memory 820, and when executed by the one or more processors 810, the drilling life cycle refined management and control method in any of the above method embodiments is executed.

[0157] The above-mentioned product can execute the method provided in the embodiment of the present application, and has the functional modules and beneficial effects corresponding to the execution method. For technical details not fully described in this embodiment, please refer to the method provided in the embodiment of the present application.

[0158] The electronic devices of the embodiments of the present application exist in various forms, including but not limited to:

[0159] (1) Mobile communication equipment: This type of equipment is characterized by having mobile communication functions and its main purpose is to provide voice and data communications. This type of terminal includes: smart phones, multimedia phones, functional phones, and low-end phones.

[0160] (2) Ultra-mobile personal computer devices: These devices belong to the category of personal computers, have computing and processing functions, and generally also have mobile Internet access features. These terminals include: PDA, MID and UMPC devices, etc.

[0161] (3) Portable entertainment devices: These devices can display and play multimedia content. They include audio and video players, handheld game consoles, e-books, smart toys, and portable car navigation devices.

[0162] (4) Other onboard electronic devices with data interaction functions, such as on-board devices installed in vehicles.

[0163] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0164] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a general hardware platform, and of course, by hardware. Based on this understanding, the above technical solution, in essence, or the part that contributes to the relevant technology, can be embodied in the form of a software product, which can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiment.

[0165] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A drilling life cycle refined management and control system, including: A sealing parameter sensing unit, used for collecting sealing sensing data based on the sealing sensor group, wherein the sealing sensing data includes the following sensing parameters: sealing material stress distribution, sealing gas concentration and sealing area negative pressure information; A sealing abnormality repair unit, configured to trigger and start a secondary sealing program when it is detected that any sensor parameter in the sealing sensor data meets a preset sealing abnormality condition, and input the sealing sensor data into a sealing repair model to determine a corresponding sealing optimization strategy; The sealing optimization strategy includes adjusting the sealing material ratio, sealing material injection speed and sealing pressure; the sealing repair model adopts a reinforcement learning model, the state of the reinforcement learning model is defined by the sealing material stress distribution, sealing gas concentration and negative pressure in the sealing area, the action of the reinforcement learning model is defined by the adjustment range of the sealing material ratio, sealing material injection speed and sealing pressure, and the reward function of the reinforcement learning model is defined by the optimization degree of the sealing effect and the sealing repair operation cost; A drainage parameter sensing unit, used for collecting gas drainage sensor data based on the drainage sensor group, wherein the gas drainage sensor data includes pipeline flow parameters, pipeline gas concentration, pipeline ultrasonic signal and pipeline negative pressure information; The extraction effect prediction unit is used to send the gas extraction sensor time series data corresponding to the first time period of history to the extraction effect prediction model to predict the gas extraction volume prediction result and the gas concentration prediction result corresponding to the second time period of the future; the backbone of the extraction effect prediction model adopts LSTM; An extraction abnormality warning unit is used to generate an extraction warning notification when it is detected that the gas extraction amount prediction result or the gas concentration prediction result meets a preset extraction abnormality condition; A drainage anomaly repair unit, for inputting the gas drainage volume prediction result and the gas concentration prediction result into a drainage parameter optimization model to determine a corresponding drainage parameter optimization strategy when it is detected that the gas drainage volume prediction result or the gas concentration prediction result meets a preset drainage anomaly condition; The extraction parameter optimization model adopts an adaptive Bayesian optimization model; the extraction parameter optimization strategy includes a target negative pressure parameter and a target gas flow rate to be adjusted; The adaptive Bayesian optimization model adopts an adaptive search strategy to dynamically adjust the search range for negative pressure parameters and gas flow according to input data; The adaptive Bayesian optimization model uses negative pressure parameters and gas flow to define the optimization parameter combination of the Bayesian optimization model. θ ; The formula for adaptive parameter iterative search is: ， In the formula, is the new parameter combination used in the next iteration, is the parameter combination used in the current round; It is the gradient of the objective function, which measures the gap between the current parameter combination and the ideal value; is the step size coefficient, which is used to control the update rate of the parameter combination; Objective function of Bayesian optimization It is expressed as: ， In the formula, and They represent the ideal extraction volume and the ideal gas concentration, respectively. and They represent the current round of extraction volume and the current round of gas concentration, respectively, and the initial round of extraction volume and gas concentration are set by the gas extraction volume prediction result and gas concentration prediction result, respectively; and They represent the fluctuation tolerance of extraction volume and gas concentration respectively; ， In the formula, Represents a preset constant, Indicates the initial step value; is the number of optimization iterations, and as the number of iterations increases, the step size coefficient Gradually decrease accordingly; The optimal optimization parameter combination corresponding to the determined extraction parameter optimization strategy It is expressed by the following formula: ， In the formula, It means to find the parameter combination that minimizes the function through multiple iterative searches. θ , and use it as the optimal optimization parameter combination .

2. The system according to claim 1, wherein: The sealing sensor group includes a sealing material stress sensor installed in the sealing area, a first gas concentration sensor having a first probe installed inside the sealing area and a second probe installed outside the sealing area, and a first negative pressure sensor installed in the extraction pipeline close to the sealing hole; The extraction sensor group includes flow sensors installed in the middle and at the end of the extraction pipeline, a second gas concentration sensor having a third probe installed in the middle of the extraction pipeline and a fourth probe at the borehole, an ultrasonic sensor installed at the connection part or the bend part of the pipeline, and a second negative pressure sensor installed in the middle and at the end of the extraction pipeline.

3. The system according to claim 1, wherein: The extraction abnormality warning unit is used to obtain the extraction pipeline identification corresponding to the extraction abnormality when it is detected that the gas extraction volume prediction result or the gas concentration prediction result meets the preset extraction abnormality condition, and use the obtained extraction pipeline identification to query the matching target management terminal information from the preset identification terminal association table, and send the extraction warning notification to the target management terminal device corresponding to the target management terminal information; The identification terminal association table pre-stores information of multiple management terminals and corresponding calibrated extraction pipeline identifications.

4. The system according to claim 1, wherein: The reward function of the reinforcement learning model is expressed as follows: ， ， ， ， In the formula, Represents the reward value, represents the weight coefficient for optimizing the sealing effect, represents the weight coefficient of the operation cost, The weight coefficient representing the penalty for system instability; represents the optimization increment of the sealing effect, is the operating cost; is the system instability penalty term, which represents the volatility of material stress and negative pressure during the sealing process; They represent the weight coefficients used to measure the importance of gas concentration, material stress and negative pressure in the sealing effect; It is the change in gas concentration, indicating the degree of reduction in gas leakage; It is the change in the stress distribution of the sealing material, indicating that the uniformity of the force on the sealing material has improved; is the optimization degree of sealing negative pressure, which indicates the matching degree between the negative pressure in the sealing area and the ideal negative pressure value in the sealing area; is the material cost, and is the energy cost; is the penalty coefficient for material stress fluctuations, is the penalty coefficient for negative pressure fluctuation; is the variance of the stress distribution of the sealing material, which indicates the stress fluctuation of the sealing material at different positions; is the negative pressure fluctuation variance, which indicates the fluctuation of negative pressure in the sealing area.

5. The system according to claim 1, wherein: The extraction effect prediction model adopts LSTM of hierarchical model architecture, and the extraction effect prediction model includes an input layer, a local hierarchical layer, a local LSTM layer, a global fusion layer, a global LSTM layer and an output layer; The input layer is used to receive the gas extraction sensor time series data; The local hierarchical layer is used to process the gas extraction sensor time series data through convolution operations to extract local features of each mode: ， In the formula, Indicates the gas extraction sensor time series data corresponding to the time step The sensor data collected is a collection of Indicates from The extracted The local convolution features of Represents the convolution kernel, which is used to extract local features. Represents the convolution operation symbol; The local LSTM layer is used to capture the temporal dependencies of each modality data in a short period of time by learning the time series data: ， In the formula, is at the time step The local features of short-term temporal dependence represent the local features of short-term temporal dependence at time step Local LSTM output features; Represents a local LSTM unit, which is used to learn the temporal dependencies of local features; Indicates that at time step Short-term local characteristics of The global fusion layer is used to fuse the local time series information for each time step output by the local LSTM layer to generate corresponding global fusion features, specifically including: For each time step , calculate the historical time step Similarity score with the current time step : ， In the formula, Represents the time step With historical time step The feature similarity score between them is used to calculate the attention weight; represents the scoring function, which is used to calculate the similarity between the features of two time steps through the dot product algorithm; Represents the time step Short-term local characteristics of Score feature similarity The attention weight is obtained by normalizing the softmax function : ， In the formula, Indicates The attention weights of time steps are normalized to sum to 1; express The exponential form of N Represents the total number of time steps corresponding to the gas extraction sensor time series data, represents the exponential summation of feature similarity scores at all time steps; Use the attention weights to perform weighted integration on the local features of the historical time steps to obtain the global fusion features: ， In the formula, is at the time step The global fusion features, Represents the time step The attention weight of The global LSTM layer is used to capture the long-term trend of the entire gas drainage system through a long time window to determine the corresponding global LSTM output features: ， In the formula, Indicates that at time step The global LSTM output features contain long-term dependent system states; Represents the global LSTM unit, which is used to process the temporal dependencies of global features; Indicates that at time step The global LSTM output features of; The output layer is used to predict the gas extraction volume prediction result and the gas concentration prediction result corresponding to the second time period in the future according to the output result of the global LSTM layer: ， In the formula, Indicates the predicted value for the second time period in the future, including the predicted results of gas extraction volume and gas concentration; and Represent the weight matrix and bias term of the output layer respectively.

Images