Intelligent judgment, regulation and control and emission reduction integrated system and method for coal seam gas extraction

Abstract

The invention discloses a coal seam gas extraction intelligent judgment, regulation and control and emission reduction integrated system and method. The system comprises an intelligent sensing layer, a data processing layer, a regulation and control execution layer and an emission reduction and efficiency improvement layer which are connected in sequence. The intelligent sensing layer collects multi-source data such as gas concentration, coal seam stress and equipment operation in real time through multiple types of sensors; the data processing layer performs intelligent judgment based on an improved deep belief network, and generates a self-adaptive regulation and control strategy for collaborative optimization of extraction efficiency, energy consumption and carbon emission by using a reinforcement learning algorithm; the regulation and control execution layer dynamically regulates the extraction power, the valve opening degree and the pipeline topology according to the strategy; and the emission reduction and efficiency improvement layer is used for carrying out graded concentration and resource utilization on the gas and recycling reaction heat energy. Thus, gas extraction is converted from experience driving to data intelligent driving, extraction efficiency is improved on the premise that safety is guaranteed, and system energy consumption and carbon emission are reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of coal mines, and in particular to an intelligent evaluation, regulation, and emission reduction integrated system and method for coal seam gas extraction. BACKGROUND

[0002] Coal seam gas is not only a major safety hazard in the process of coal mining, but also a clean energy with high calorific value. Efficient extraction of coal seam gas has multiple meanings for ensuring coal mine safety production, realizing gas resource utilization, and reducing greenhouse gas emissions. At present, coal mines have generally established ground or underground gas extraction systems, but the level of intelligence and collaboration still needs to be improved, and there are severe challenges in extraction efficiency, operation energy consumption, and carbon emission control.

[0003] Traditional gas extraction systems usually adopt a basic mode of monitoring + manual regulation. The typical technical architecture is: a single gas concentration sensor is arranged at a key point of the extraction pipeline or roadway to monitor the gas concentration in real time; the extraction pump group usually operates at constant power in the form of power frequency; the regulating valve is mostly a manual or simple electric valve, and the opening adjustment is discontinuous and inaccurate; the subsystems such as monitoring, extraction, and utilization are relatively independent, and the data interaction is limited.

[0004] The existing technology mainly has the following defects: 1) Evaluation lag and regulation roughness: the existing system relies on fixed thresholds set by manual to regulate extraction, which cannot adapt to the dynamic characteristics of coal seam gas emission. For example, when the stress of coal seam suddenly changes, causing the gas concentration to instantaneously increase, the system response lags, which easily leads to gas overrun accidents; and in low-concentration gas areas, excessive extraction leads to invalid operation. According to statistics, the average extraction efficiency of systems using such traditional modes is generally low, and the timeliness and accuracy of gas control are insufficient.

[0005] High energy consumption and carbon emission: the power frequency extraction pump is in a constant power operation state for a long time, and cannot automatically reduce load or stop even when the gas amount is small, resulting in a high proportion of idle loss. At the same time, a large amount of low-concentration gas is directly discharged into the atmosphere due to poor utilization economy, which has a huge greenhouse effect potential, causing significant energy waste and environmental pressure.

[0006] Insufficient system collaboration: the extraction evaluation, equipment regulation, and safety warning modules are independent of each other, lack data fusion and global optimization capabilities, and cannot simultaneously meet the safety extraction, energy efficiency improvement, and emission reduction targets. The gas sensor is only used for overrun alarm and is not linked with pump group power regulation; energy efficiency optimization only focuses on pump group power consumption and does not consider the indirect energy consumption caused by waste of gas resources. When the gas concentration meets the standard, high-power extraction is still maintained, resulting in high monthly power consumption. SUMMARY

[0007] The embodiment of the application provides a coal seam gas extraction intelligent evaluation, regulation and control and emission reduction integrated system and method, and solves the problem of poor effect of the prior art in intelligent evaluation precision, dynamic regulation and control capability, system collaboration and emission reduction and efficiency improvement level.

[0008] To achieve the above object, the technical scheme of the embodiment of the application is as follows: In a first aspect, the embodiment of the application provides a coal seam gas extraction intelligent evaluation, regulation and control and emission reduction integrated system, which comprises an intelligent sensing layer, a data processing layer, a regulation and control execution layer and an emission reduction and efficiency improvement layer connected in sequence, wherein: the intelligent sensing layer comprises multiple types of sensors arranged in coal seam drill holes, roadways and extraction pipelines, and is used to obtain multi-source data in the coal seam gas extraction process; wherein the multi-source data at least includes gas concentration, coal seam stress, extraction flow, equipment vibration frequency and pipeline pressure parameters; the data processing layer comprises an intelligent evaluation module and a regulation and control decision module, the intelligent evaluation module fuses and extracts features of the multi-source data based on a deep belief network improved based on an attention mechanism, and outputs extraction efficiency grades, equipment health states and carbon emission prediction values; the regulation and control decision module generates an optimal regulation and control strategy including extraction pump power, valve opening degree and pipeline topology structure based on a reinforcement learning algorithm according to the evaluation result of the intelligent evaluation module; the regulation and control execution layer comprises a variable frequency extraction pump, an intelligent throttle valve and a branch pipeline switching device, and is used to receive and execute the regulation and control instructions generated by the regulation and control decision module, and perform dynamic adjustment of extraction parameters and real-time reconstruction of pipeline topology; the emission reduction and efficiency improvement layer comprises a gas staged utilization module and a heat recovery module, the gas staged utilization module is used to shunt gas to a membrane separation and concentration unit or a direct utilization unit according to the gas concentration, and the heat recovery module is used to recover and utilize heat generated by oxidation of the concentrated gas.

[0009] In some possible implementation manners, the multiple types of sensors in the intelligent sensing layer include: a gas concentration sensor adopting a tunable semiconductor laser absorption spectrum technology, a coal seam stress sensor based on a fiber Bragg grating principle and an extraction flow sensor arranged in main lines and branch pipelines of the extraction pipeline; wherein each sensor performs data acquisition through a preset sampling frequency, is arranged in a distributed networking manner, forms a three-dimensional monitoring network covering the coal seam drill holes, the roadways and the extraction pipeline, and uploads the collected data to the data processing layer through a wireless network.

[0010] In some possible implementation manners, the deep belief network has a multi-layer structure including an input layer, at least one hidden layer, an attention layer and an output layer, and the input is a multivariate feature vector composed of the multi-source data; the reinforcement learning algorithm is a Q-Learning algorithm, a state space includes gas concentration distribution, coal seam stress state and equipment operation parameters, and an action space includes extraction pump power adjustment, valve opening degree adjustment and pipeline switching instructions.

[0011] In some possible implementation manners, the variable frequency extraction pump adopts a permanent magnet synchronous motor and a vector control technology, and a vector controller is configured to realize continuous adjustment of power; the intelligent throttle valve adopts an electro-hydraulic servo control technology, and is used for adjusting the valve opening according to an optimal regulation and control strategy to realize accurate control of the extraction flow of each branch pipeline; and the branch pipeline switching device includes a plurality of electric valves, and the switching of different extraction pipeline topological structures is realized through the combined switching of the valves.

[0012] In some possible implementation manners, in the emission reduction and efficiency improvement layer: the gas staged utilization module further includes a ventilation air oxidation device, which is used for oxidizing and utilizing the gas processed by the membrane separation and concentration unit; and the heat recovery module is connected to the ventilation air oxidation device, and is used for recovering heat energy generated by the oxidation reaction.

[0013] In some possible implementation manners, the reward function of the reinforcement learning algorithm is a weighted sum of the extraction efficiency, system energy consumption and carbon emission, which is used for realizing multi-objective collaborative optimization when generating the optimal regulation and control strategy.

[0014] In some possible implementation manners, the intelligent perception layer further includes an edge computing node, which is used for performing space-time alignment and noise reduction preprocessing on the raw data collected by the multiple types of sensors, and then uploading the data to the data processing layer.

[0015] In some possible implementation manners, the system further includes a central control server, which is in communication connection with the data processing layer and the regulation and control execution layer respectively, and is used for coordinating data flow and issuing instructions, and realizing human-computer interaction and system state monitoring.

[0016] In the second aspect, the embodiment of the present application provides a coal seam gas extraction intelligent evaluation, regulation and control and emission reduction integrated method, which comprises the following steps: through multiple types of sensors arranged in coal seam drill holes, roadways and extraction pipelines, real-time collection of gas concentration, coal seam stress, extraction flow, equipment vibration frequency and pipeline pressure parameters, obtaining multi-source data in the coal seam gas extraction process; using a deep belief network improved based on an attention mechanism to fuse and extract features of the multi-source data, and outputting extraction efficiency grades, equipment health states and carbon emission prediction values; based on a reinforcement learning algorithm, generating an optimal regulation and control strategy including extraction pump power, valve opening and pipeline topological structure according to the extraction efficiency grades, the equipment health states and the carbon emission prediction values; executing the optimal regulation and control strategy through a variable frequency extraction pump, an intelligent throttle valve and a branch pipeline switching device to realize dynamic adjustment of extraction parameters and real-time reconstruction of pipeline topologies; and according to the gas concentration, the gas is shunted to a membrane separation and concentration unit or directly utilized, and heat generated by oxidation of the concentrated gas is recycled and utilized.

[0017] The one or more technical solutions provided in the embodiment of the present application have at least the following technical effects or advantages: In the embodiment of the present application, a plurality of types of sensors deployed in drill holes, roadways and pipelines are used to fuse multi-source data such as gas concentration, coal seam stress, equipment state and the like in real time; intelligent judgment is performed based on a deep belief network with an introduced attention mechanism, and a reinforcement learning algorithm with safety, energy efficiency and emission reduction as a reward function is combined to generate an adaptive optimal regulation and control strategy; then, through the collaborative execution of a variable frequency pump set, an intelligent valve and a reconfigurable pipeline topology, millisecond-level accurate regulation and control of the extraction parameters is realized. The present application overcomes the defects of the prior art, such as dependence on fixed thresholds, response lag and extensive regulation and control, and can significantly improve the average extraction efficiency, while greatly reducing system energy consumption through frequency regulation and intelligent start-stop strategies. In addition, the system integrates gas grading utilization and heat recovery modules, resources the low-concentration gas traditionally directly discharged after being enriched, recovers reaction heat energy, and builds an integrated green cycle of extraction, utilization and emission reduction, which greatly reduces direct emission of greenhouse gases while creating additional economic benefits. Finally, the three goals of safe extraction, energy saving and carbon emission reduction are synergistically optimized. BRIEF DESCRIPTION OF DRAWINGS

[0018] In order to more clearly illustrate the embodiments of the present application, the drawings required to be used in the embodiments of the present application will be briefly introduced as follows. Obviously, the drawings described in the following description are some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor on the basis of these drawings.

[0019] Figure 1 A structural schematic diagram of a coal seam gas extraction intelligent evaluation, regulation and control and emission reduction integrated system provided by the embodiment of the present application; Figure 2 An embodiment flowchart of a coal seam gas extraction intelligent evaluation, regulation and control and emission reduction integrated method in the embodiment of the present application. DETAILED DESCRIPTION

[0020] The technical solutions in the embodiments of the present application will be described clearly and completely in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0021] In the description of the embodiments, the terms "comprising", "containing", "having", etc. are open-ended terms that are generally preferred over the terms "consisting of", "consisting essentially of", and "consisting only of", which are closed terms. The term "at least one" is generally preferred over the term "one or more", which is a closed term. The term "at least one of" or similar expressions is intended to refer to any combination of the items that the term refers to, including single items or combinations of multiple items. For example, "at least one of a, b, or c" or "at least one of a, b, and c" can mean a, b, c, a-b (i.e., a and b), a-c, b-c, or a-b-c, where a, b, and c can each be a single item or multiple items. The symbol "A / B" is used to describe the selection relationship of the associated objects, which generally represents the "or" relationship between the front and back.

[0022] In the following description of the embodiments, the terms used in the embodiments of the present application are merely for the purpose of describing the specific embodiments and are not intended to limit the present application. The singular forms "a" and "the" used in the embodiments of the present application and the appended claims are also intended to include the plural forms, unless the context clearly indicates otherwise.

[0023] Those skilled in the art should understand that in the following description of the embodiments of the present application, the order of the serial numbers does not mean the order of execution, and some or all steps can be executed in parallel or in sequence, and the execution order of each process should be determined according to its function and inherent logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0024] Those skilled in the art should understand that in the following description of the embodiments of the present application, the numerical range should be understood as also specifically disclosing each intermediate value between the upper limit and the lower limit of the range. Each smaller range between any stated value or stated range of values and any other stated value or stated range of values is also included within the present application. The upper and lower limits of these smaller ranges can be included or excluded independently from the range.

[0025] Unless otherwise specified, the technical / scientific terms used herein have the same meaning as generally understood by those skilled in the art to which the present application belongs. Although only preferred methods and materials are described herein, any methods and materials similar or equivalent to those described herein can also be used in the implementation or testing of the present application. All documents mentioned in the specification are incorporated by reference to disclose and describe the methods and / or materials related to the documents. In the event of any conflict between the content of the specification and any incorporated document, the content of the specification shall prevail.

[0026] In order to illustrate the technical solutions of the present application, the following specific embodiments are described.

[0027] Figure 1A structural schematic diagram of a coal seam gas extraction intelligent evaluation, regulation, and emission reduction integrated system is provided for an embodiment of the present application, referring to Figure 1 The coal seam gas extraction intelligent evaluation, regulation, and emission reduction integrated system can include an intelligent perception layer, a data processing layer, a regulation and execution layer, and an emission reduction and efficiency improvement layer connected in sequence. The intelligent perception layer includes multiple types of sensors arranged in coal seam drill holes, roadways, and extraction pipelines to obtain multi-source data during the coal seam gas extraction process. It can be understood that the intelligent perception layer, as the data acquisition core of the entire integrated system, is responsible for the function of accurately capturing multi-dimensional information throughout the coal seam gas extraction process. It deploys multiple types of sensors at key positions in coal seam drill holes, roadways, and extraction pipelines to build a comprehensive and dead-angle-free monitoring unit for real-time acquisition of multi-source heterogeneous data during the coal seam gas extraction process, thereby providing data support for subsequent intelligent evaluation and dynamic regulation. The multi-source data can include at least gas concentration, coal seam stress, extraction flow rate, equipment vibration frequency, and pipeline pressure parameters. Additionally, auxiliary parameters such as pump group operating current and roadway environmental temperature can be collected based on actual monitoring needs to achieve comprehensive perception of extraction conditions.

[0028] In some embodiments, the multiple types of sensors in the intelligent perception layer can include a gas concentration sensor using tunable diode laser absorption spectroscopy (TDLAS) technology, a coal seam stress sensor based on the principle of fiber Bragg grating (FBG), and an extraction flow rate sensor arranged in the main line and branch pipelines of the extraction pipeline. Each sensor collects data at a preset sampling frequency and is arranged in a distributed networking manner to form a three-dimensional monitoring network covering coal seam drill holes, roadways, and extraction pipelines. The collected data is uploaded to the data processing layer through a wireless network.

[0029] Specifically, the TDLAS gas concentration sensor has high precision and strong anti-interference characteristics, supports distributed networking mode, and can be densely arranged at key monitoring points such as the inner wall of the coal seam drill hole, the roof and sidewall of the roadway, etc., to accurately capture the spatial distribution of gas concentration. The FBG coal seam stress sensor is deployed in a fixed-point manner, and the deployment density can be determined based on actual application needs. For example, one measurement point can be arranged every 10 meters of drill hole length to monitor the coal seam stress change trend in real time and provide early warning for gas outburst risks caused by stress mutations. The extraction flow rate sensor is designed for pipeline gas transportation characteristics and can support simultaneous collection and real-time comparative analysis of multi-pipeline flow rates to accurately feedback the extraction efficiency differences of each branch pipeline.

[0030] In some embodiments, each sensor can continuously collect data at a preset sampling frequency. For example, the sampling frequency can be greater than or equal to 10 Hz. All sensors are arranged in a distributed networking manner to form a three-level stereoscopic monitoring network covering the coal seam drilling, underground roadway, and extraction pipeline. The total number of sensors deployed in a single working face is 200-300, ensuring the spatial continuity and temporal integrity of the monitoring data.

[0031] In some embodiments, the intelligent perception layer further includes an edge computing node for performing spatio-temporal alignment and noise reduction preprocessing on the raw data collected by multiple types of sensors, and then uploading the data to the data processing layer.

[0032] The collected raw data can first be converted into a unified communication protocol by a mine intrinsically safe signal converter to avoid data transmission abnormalities caused by device interface differences. Then, the data is preprocessed by a spatio-temporal alignment algorithm (multi-sensor data fusion technology based on Kalman filtering) carried by the edge computing node. The wavelet transform filtering algorithm is used to effectively remove noise data caused by environmental interference and equipment vibration, and the data timestamps of each sensor are synchronized to ensure the spatio-temporal consistency of multi-source data. The preprocessed standardized data can be uploaded to the data processing layer through a 5G wireless network with a transmission rate of ≥100 Mbps. This transmission method not only meets the real-time requirements of a sampling frequency of ≥10 Hz, but also has the advantages of strong anti-interference ability and low transmission delay, which can adapt to the complex electromagnetic environment and narrow roadway transmission scenarios in coal mines.

[0033] In some alternative embodiments, the above-mentioned multi-type sensors can also use a Brillouin optical time domain reflectometry (BOTDR) technology alternative. This type of sensor can continuously monitor the gas concentration and temperature distribution along the extraction pipeline without the need for dense arrangement of measurement points to achieve long-distance coverage, with a monitoring distance of up to 10 km. The gas concentration monitoring accuracy can reach ±1.5%, which is suitable for complex roadway layout and long extraction pipeline scenarios in large coal mines. This can reduce installation and maintenance costs while effectively monitoring the macroscopic distribution trend of gas concentration.

[0034] The data processing layer includes an intelligent evaluation module and a regulation and decision-making module. The intelligent evaluation module uses a deep belief network improved based on attention mechanism to fuse and extract features from multi-source data, and outputs extraction efficiency level, device health status, and carbon emission prediction value. The regulation and decision-making module generates an optimal regulation strategy including extraction pump power, valve opening degree, and pipeline topology structure based on the evaluation results of the intelligent evaluation module. In some embodiments, the deep belief network (DBN) has a multi-layer structure including an input layer, at least one hidden layer, an attention layer, and an output layer. The input is a multi-element feature vector composed of multi-source data. Specifically, the intelligent evaluation module takes a deep belief network improved based on an attention mechanism as a core algorithm model, is designed for the fusion and feature extraction of multi-source heterogeneous data of coal seam gas extraction, and solves the problems of lag and one-sidedness of traditional single parameter evaluation.

[0035] For example, taking a 5-layer network architecture as an example, the model includes an input layer, 2 hidden layers, an attention layer, and an output layer: the input layer receives 12-dimensional feature parameters from the intelligent perception layer, in addition to the core gas concentration, coal seam stress, extraction flow, equipment vibration frequency, and pipeline pressure, the input layer also receives auxiliary parameters such as pump group operating current, roadway environment temperature, and valve operating state, to ensure the comprehensiveness of the feature dimension; the attention layer focuses on key features that have a significant impact on extraction effect and equipment state by dynamically allocating weights, such as high-concentration gas area concentration changes and coal seam stress mutation data, which can effectively improve the model's ability to capture core information and avoid irrelevant data interference. The 2 hidden layers realize feature depth mining and nonlinear mapping through multilayer perceptron, and the output layer outputs three evaluation results, including extraction efficiency level, equipment health status, and carbon emission prediction value, which can provide direct basis for subsequent regulation and control decisions.

[0036] In some embodiments, to ensure evaluation accuracy and engineering adaptability, the improved DBN model can use long-term measured multi-condition data in coal mine sites as training samples, which can cover typical scenarios such as normal extraction, gas outburst early warning, equipment failure, and low-concentration gas enrichment, and iteratively adjust network weight parameters through Adam optimization algorithm, with the combination of mean square error (MSE) and cross-entropy loss function as the optimization objective, to ensure the regression accuracy of continuous parameters such as carbon emission prediction value, and improve the classification accuracy of discrete parameters such as efficiency level and health status. The model prediction accuracy is 92%, and the identification accuracy of abnormal conditions is improved to 95%.

[0037] In some embodiments, the reinforcement learning algorithm is a Q-Learning algorithm, the state space includes gas concentration distribution, coal seam stress state, and equipment operating parameters, and the action space includes extraction pump power adjustment, valve opening adjustment, and pipeline switching instructions.

[0038] The control and decision-making module is built based on the Q-Learning reinforcement learning algorithm. Its core objective is to generate an optimal control strategy that balances extraction efficiency, energy consumption control, and carbon emission reduction based on the intelligent evaluation results. The module's state space consists of 20-dimensional state variables, including not only the extraction efficiency level, equipment health status, and carbon emission predictions output by the intelligent evaluation module, but also real-time dynamic state data such as gas concentration spatial distribution, coal seam stress field distribution, flow differences in various branch pipelines, and real-time pump power, ensuring a comprehensive understanding of the current extraction conditions. The module's action space covers the core controllable dimensions of the extraction system, specifically including: extraction pump power adjustment, intelligent throttle valve opening adjustment, and pipeline topology switching.

[0039] For example, the pump power adjustment range is 20%-100% with an adjustment step of 0.5%, supporting fine power matching; the intelligent throttle valve opening adjustment range is 0-100% with an adjustment step of 1%, adapting to different flow requirements; the pipeline topology switching supports multiple preset topology combinations, which can adapt to the differentiated management and control requirements of priority extraction in high-concentration areas and centralized concentration enhancement in low-concentration areas.

[0040] In some embodiments, to achieve multi-objective collaborative optimization, the reward function of the reinforcement learning algorithm is a weighted sum of extraction efficiency, system energy consumption, and carbon emissions, used to achieve multi-objective collaborative optimization when generating the optimal control strategy. The weight configuration of the weighted sum can be tailored to the core needs of coal mine production. For example, extraction efficiency weight 0.4 prioritizes ensuring gas extraction effectiveness and reducing safety risks; energy consumption weight 0.3 controls the power consumption of pump sets and auxiliary equipment; carbon emissions weight 0.3 reduces carbon emissions caused by gas venting and ineffective operations; the reward function guides the algorithm to iterate towards the optimal direction of high efficiency, low consumption, and emission reduction.

[0041] During the offline phase, the control and decision-making module can undergo extensive simulation training to iteratively optimize strategies for typical scenarios with different geological conditions, gas outburst intensity, and equipment operating status, forming an optimal control strategy library containing a massive number of operating conditions. During online operation, it can quickly retrieve suitable strategies from the strategy library by combining the real-time results of the intelligent evaluation module with the current system status, with a decision cycle of ≤10s, which can meet the real-time requirements of dynamic control and ensure that the extraction system can respond quickly to changes in operating conditions.

[0042] In other alternative embodiments, for the processing needs of unstructured data such as coal seam stress waveforms and pump vibration spectra, the intelligent evaluation module can use the Transformer architecture to extract features from unstructured data and build a prediction model by combining it with a temporal convolutional network. This solution can further improve the response speed to dynamic scenarios such as sudden gas outbursts and early equipment failures. According to actual test data, the model prediction latency can be reduced from 10s in DBN to 5s, and the response speed to sudden gas outbursts is improved by 20% on the original basis. It can be applied to complex mining scenarios with high outburst mines and strong dynamics of gas outbursts, and can be realized with the support of hardware with higher computing power.

[0043] The control and execution layer, including variable frequency extraction pumps, intelligent throttling valves and branch pipeline switching devices, is used to receive and execute control commands generated by the control decision module, and to dynamically adjust extraction parameters and reconstruct pipeline topology in real time. Among them, the variable frequency extraction pump adopts a permanent magnet synchronous motor and vector control technology, and is equipped with a vector controller to achieve continuous power adjustment; Understandably, as the core power source for gas extraction, the variable frequency extraction pump is designed with a permanent magnet synchronous motor and vector control technology. Compared with traditional power frequency pump sets, it has the advantages of a wide power adjustment range, fast response speed, and low energy loss. In this embodiment of the invention, the power adjustment of the variable frequency extraction pump can cover the full range of 0%-100%, supports fine adjustment in 0.5% steps, and can accurately match the gas emission intensity according to the control command; the pump set response time is less than 500ms, which can quickly follow up on the fluctuations in working conditions such as sudden changes in gas concentration and changes in coal seam stress, ensuring timely extraction of gas in high-concentration areas; the matching intelligent vector controller can effectively suppress the current surge during start-up and shutdown, reduce damage to the power grid and pump set mechanical structure, extend the service life of the equipment, and its power adjustment error is ≤±2%, which can meet the precise control requirements of extraction power.

[0044] The intelligent throttle valve employs electro-hydraulic servo control technology to adjust the valve opening according to the optimal control strategy, achieving precise control of the extraction flow rate of each branch pipeline. Its valve opening adjustment range is 0-100%, with an adjustment step of 1%, allowing for differentiated control based on the flow requirements of each branch pipeline. The control error is ≤3%, ensuring that the extraction flow rate of each pipeline matches the gas concentration distribution. A high-precision pressure sensor can be installed inside the valve to collect real-time pressure data within the pipeline and feed it back to the control decision module, forming a pressure closed-loop control. This prevents safety risks caused by pipeline overpressure operation and provides pipeline pressure parameter support for evaluating extraction efficiency.

[0045] The branch pipeline switching device includes multiple electric valves, which are used to switch between different extraction pipeline topologies through a combination of valve switching.

[0046] Specifically, the branch pipeline switching device consists of an electric ball valve assembly with eight branch interfaces and an intelligent controller. It is the core component for realizing pipeline topology reconfiguration, used to switch between differentiated pipeline topologies for priority extraction in high-concentration areas and concentrated enrichment in low-concentration areas. Through the electric ball valve assembly with eight branch interfaces, the device can support rapid switching between eight preset pipeline topologies, with a switching time of less than 30 seconds. It can dynamically adjust the connection relationship between the borehole group, the main extraction pipeline, and the enrichment branches based on gas concentration distribution and extraction efficiency assessment results.

[0047] For example, when the gas concentration in a certain area reaches the high concentration standard (≥30%), the system quickly switches to the main extraction pipeline for priority extraction; when the gas concentration in the area is low concentration (<30%), the system switches to the enrichment branch for centralized enrichment, ensuring efficient utilization of gas resources. To avoid pressure surges during pipeline switching affecting extraction stability, the device adopts a first-open-then-close control logic: the intelligent controller first drives the electric ball valve of the target branch to open, and after the pipeline pressure stabilizes, it closes the ball valve of the original branch, thereby ensuring the continuity and safety of the extraction process.

[0048] In this embodiment of the invention, the three core execution components—the variable frequency extraction pump, the intelligent throttling valve, and the branch pipeline switching device—are coordinated and linked through a unified control bus. Specifically, the power adjustment of the variable frequency extraction pump and the opening adjustment of the intelligent throttling valve respond synchronously to ensure the matching of pipeline negative pressure and flow rate; after the topology reconfiguration action of the branch pipeline switching device is triggered, the control decision module synchronously adjusts the pump power and valve opening of the corresponding pipeline to avoid flow imbalance or pressure abnormalities after switching. The entire operation response of the control execution layer is fully automated, requiring no manual intervention, and all execution parameters, such as real-time pump power, valve opening, and pipeline switching status, can be fed back to the data processing layer in real time, forming a closed-loop control of decision-making, execution, and feedback.

[0049] In other alternative embodiments, for emergency drainage needs in high-risk gas outburst mines, the variable frequency drainage pump can also be a high-power drainage pump driven by a high-efficiency permanent magnet motor, with pump group efficiency increased to ≥90% and power coverage ranging from 100-500kW. It is equipped with a high-precision electro-hydraulic servo valve with a response time of less than 100ms and control accuracy of ±0.1%, which can shorten the overall control cycle to 15s. It can quickly increase drainage capacity after the appearance of gas outburst signs, minimize the outburst risk, and is suitable for special mining scenarios with complex geological conditions and high gas outburst risk.

[0050] The emission reduction and efficiency enhancement layer includes a gas classification and utilization module and a heat recovery module. The gas classification and utilization module is used to divert gas to the membrane separation concentration unit or the direct utilization unit according to the gas concentration. The heat recovery module is used to recover and utilize the heat generated by the oxidation of gas after concentration.

[0051] In some embodiments, the gas classification and utilization module in the emission reduction and efficiency enhancement layer further includes a waste gas oxidation device for oxidizing and utilizing the gas after it has been treated by the membrane separation and concentration unit. The heat recovery module is connected to the exhaust gas oxidation device to recover the heat energy generated by the oxidation reaction.

[0052] Specifically, the gas classification and utilization module uses dynamic monitoring data of gas concentration as a basis. Through preset concentration thresholds and automatic switching control of intelligent valve groups, it achieves differentiated and precise utilization of gas with different concentrations. The preset concentration threshold can be taken from 30%, the key dividing point between low-concentration and high-concentration gas utilization in the industry. A concentration value above 30% is considered high concentration, and a concentration value of 30% is considered low concentration.

[0053] The membrane separation and concentration unit is specifically designed for the purification of low-concentration methane gas. It can utilize hollow fiber membrane modules that are resistant to methane corrosion and offer high separation precision. The membrane pore size is controlled within 0.1-1 μm, and the operating pressure is set at 0.3-0.5 MPa, enabling efficient separation of methane gas and air under ambient temperature conditions. Through multi-stage membrane separation optimization (e.g., a 3-stage series design), this unit can concentrate low-concentration methane gas to over 50%, with a stable concentration efficiency exceeding 90%. This solves the problem of traditionally low-concentration methane gas being forced to be vented due to its low energy density and inability to be directly utilized.

[0054] The exhaust gas oxidation unit can adopt a regenerative thermal oxidizer (RTO) structure for energy recovery from enriched methane. Its thermal efficiency is ≥85%, and it can withstand methane intake concentration fluctuations within ±5%, exhibiting strong anti-interference capabilities. The unit can be equipped with a ceramic regenerator, achieving efficient heat recovery through heat absorption and release cycles. The oxidation temperature is controlled at 850-950℃, completely oxidizing and decomposing methane in the methane into CO2 and water, avoiding the increased carbon emissions caused by direct methane emissions. Simultaneously, the exhaust gas oxidation unit can be equipped with a flame monitoring system and an explosion-proof pressure relief valve. When the methane concentration abnormally exceeds the standard or the oxidation temperature is too high, automatic pressure relief and gas shut-off protection are triggered to ensure safe underground operation.

[0055] In some embodiments, to meet energy conversion needs, the concentrated methane can be used for power generation through oxidation. For example, a gas generator set integrated with a waste heat recovery auxiliary system can be adopted, with the generator set power dynamically configured according to the amount of methane extracted from the coal mine. After dehydration and desulfurization pretreatment, the high-concentration methane is fed into the combustion chamber of the generator set for combustion, driving the generator to generate electricity. The electricity can be directly integrated into the underground power supply system of the coal mine, replacing purchased electricity and reducing the cost of electricity for production. The power generation unit can control the concentration of pollutants after combustion below national emission standards through a matching exhaust gas purification device, achieving clean power generation.

[0056] In some embodiments, the gas classification and utilization module can be equipped with an intelligent diversion and control system, which can consist of an electric three-way valve, redundant concentration monitoring sensors, and a PLC controller, and is linked with the gas concentration sensor in the intelligent sensing layer and the pipeline switching device in the control execution layer. The redundant concentration monitoring sensors collect real-time gas concentration data entering the emission reduction and efficiency enhancement layer, with the data sampling frequency consistent with the front end to ensure accurate concentration judgment. The PLC controller issues diversion commands based on the comparison between the concentration data and preset thresholds. When the concentration is ≥30%, the electric three-way valve switches to the power generation unit branch; when the concentration is <30%, it switches to the membrane separation and concentration unit branch; if the concentration fluctuates (e.g., between 28% and 32%), a buffer tank is activated to temporarily store the gas until the concentration stabilizes before diversion, avoiding equipment wear caused by frequent switching.

[0057] The specific graded utilization process may include: after the gas is transported to the emission reduction and efficiency enhancement layer via the branch pipeline switching device of the control and execution layer, it first enters the pretreatment buffer tank to balance the gas intake pressure and flow rate; then, after the concentration detection and judgment of the intelligent diversion control system, the corresponding utilization path is activated. For high-concentration pathways: gas is pretreated by dehydration and desulfurization → connected to gas generator set → combustion to generate electricity → electricity is connected to the underground power grid → combustion exhaust gas is purified and discharged in compliance with standards. For low-concentration pathways: Gas enters the membrane separation enrichment unit → is purified to over 50% through 3-stage membrane separation → is connected to the regenerative exhaust gas oxidation unit → releases heat energy through high-temperature oxidation → the oxidized tail gas recovers heat through a heat exchange system before being discharged in compliance with standards.

[0058] The entire process is automated and requires no manual intervention. The operating status of each unit is fed back to the data processing layer in real time, forming a closed-loop control system for concentration monitoring, diversion control, operation feedback, and parameter optimization.

[0059] In some embodiments, the heat recovery module serves as an energy supplement and energy-saving extension for gas grading and utilization. It is deeply integrated with the exhaust gas oxidation device of the gas grading and utilization module. Its goal is to recover a large amount of heat energy released during the oxidation of low-concentration gas, replace the traditional raw coal combustion heating mode, and further reduce energy consumption and carbon emissions in the coal mine production process. The core components may include a high-efficiency shell-and-tube heat exchange system, heat energy transmission pipeline network, and end-use equipment.

[0060] Specifically, the high-temperature exhaust gas generated by the oxidation of methane in the exhaust gas oxidation unit enters the shell side of the heat exchange system and undergoes countercurrent heat exchange with the heat transfer medium circulating in the tube side. After absorbing heat energy, the heat transfer medium's temperature rises to 120-150℃ and is transported to various end-use equipment via the heat energy transmission network. After releasing heat energy, the temperature drops to 60-80℃ and returns to the heat exchange system through the circulation pump to absorb heat again, forming a closed loop. The low-temperature exhaust gas (≤120℃) after heat exchange is purified by the exhaust gas purification device to remove particulate matter and is discharged underground through a dedicated exhaust pipe to avoid polluting the underground air.

[0061] The operating parameters of the heat recovery module are fed back to the data processing layer in real time, and are linked with the operating data of the gas classification and utilization module: when the data processing layer predicts an increase in gas oxidation, it adjusts the power of the circulating pump of the heat exchange system in advance and increases the flow rate of the heat transfer medium to ensure timely heat recovery; when the end-point heat demand decreases, it adjusts the opening of the bypass valve of the heat exchange system to reduce the amount of heat recovery, avoid system overpressure, and achieve dynamic matching between heat recovery and demand.

[0062] In this embodiment of the invention, the gas grading and utilization module and the heat recovery module do not operate independently, but form a closed loop with the front-end intelligent sensing layer and the data processing layer: the intelligent sensing layer monitors parameters such as gas concentration and flow rate in real time, the data processing layer predicts carbon emission trends through models, and then issues diversion control instructions to the gas grading and utilization module to ensure that the diversion decision is dynamically matched with the gas concentration; the heat recovery efficiency data of the heat recovery module is fed back to the data processing layer in real time, serving as an important basis for carbon emission calculation and energy efficiency optimization, and continuously adjusting the grading and utilization strategy.

[0063] In some embodiments, the above-mentioned integrated intelligent assessment, regulation, and emission reduction system for coal seam gas extraction may also include a central control server, which is connected to the above-mentioned data processing layer and the above-mentioned regulation execution layer via a 5G network for coordinating data flow and instruction issuance, and realizing human-machine interaction and system status monitoring.

[0064] Specifically, the central control server mainly performs the following functions: Data coordination and command distribution: Receive intelligent evaluation results and generated optimal control strategies uploaded from the data processing layer in real time, verify and convert the strategies, and then accurately and reliably distribute them to the corresponding variable frequency extraction pumps, intelligent throttle valves, and branch pipeline switching devices in the control execution layer.

[0065] Full-process real-time monitoring and human-machine interaction: The server runs an integrated monitoring software platform that provides a graphical human-machine interface (HMI). This interface can centrally display the real-time data and status of the intelligent sensing layer's three-dimensional monitoring network, the evaluation results and decision logic of the data processing layer, the action status and parameters of each device in the control and execution layer, and the gas utilization efficiency and heat recovery status of the emission reduction and efficiency enhancement layer.

[0066] System coordination and scheduling management: Responsible for coordinating the timing of model calculations and strategy generation in the data processing layer, and scheduling and controlling the sequence of coordinated actions of multiple devices in the execution layer. For example, when switching pipeline topologies, it controls the order in which valves open and close to avoid pressure surges, ensuring the efficient, safe, and stable closed-loop operation of the entire integrated system.

[0067] Status monitoring and alarms: Real-time monitoring of abnormal states in all aspects of the system, such as sensor failure, concentration exceeding limits, equipment malfunctions, etc., triggering audible and visual alarms, and initiating emergency interlock control procedures according to preset safety rules, such as urgently increasing extraction capacity or initiating safe discharge to ensure the safe operation of the system.

[0068] By introducing a central control server, this embodiment of the invention achieves an upgrade from decentralized control to centralized intelligent management, enhancing the system's integrity, synergy, and reliability, and providing operators with a unified, convenient, and efficient management entry point.

[0069] In this embodiment of the invention, multiple types of sensors deployed in boreholes, roadways, and pipelines are used to fuse multi-source data such as gas concentration, coal seam stress, and equipment status in real time. Intelligent evaluation is performed based on a deep belief network incorporating an attention mechanism, combined with a reinforcement learning algorithm that uses safety, energy efficiency, and emission reduction as reward functions to generate an adaptive optimal control strategy. Then, through the coordinated execution of variable frequency pump sets, intelligent valves, and reconfigurable pipeline topology, millisecond-level precise control of extraction parameters is achieved. This overcomes the shortcomings of existing technologies, such as reliance on fixed thresholds, response lag, and coarse control, significantly improving average extraction efficiency. Simultaneously, variable frequency adjustment and intelligent start-stop strategies greatly reduce system energy consumption. Furthermore, the system integrates gas classification and utilization and heat recovery modules, concentrating and utilizing low-concentration gas that was traditionally directly emitted, and recovering reaction heat energy, constructing an integrated green cycle of extraction, utilization, and emission reduction. This greatly reduces direct greenhouse gas emissions while creating additional economic benefits. Ultimately, the system achieves synergistic optimization of the three major objectives of safe extraction, energy conservation and emission reduction.

[0070] Based on the same inventive concept, this application also provides an integrated method for intelligent evaluation, control, and emission reduction of coal seam gas extraction. Figure 2 This is a schematic flowchart of an embodiment of an integrated method for intelligent assessment, control, and emission reduction of coal seam gas extraction according to an invention. (See attached diagram.)Figure 2 As shown, the method may include: S201 uses multiple types of sensors deployed in coal seam boreholes, roadways and extraction pipelines to collect real-time data on gas concentration, coal seam stress, extraction flow rate, equipment vibration frequency and pipeline pressure parameters, thereby obtaining multi-source data during the coal seam gas extraction process. S202 uses a deep belief network based on an improved attention mechanism to fuse and extract features from multi-source data, and outputs the extraction efficiency level, equipment health status and carbon emission prediction values. S203, based on reinforcement learning algorithm, generates optimal control strategy including pump power, valve opening and pipeline topology according to pumping efficiency level, equipment health status and carbon emission prediction. S204 implements the optimal control strategy through variable frequency extraction pumps, intelligent throttling valves and branch pipeline switching devices to achieve dynamic adjustment of extraction parameters and real-time reconstruction of pipeline topology. S205, depending on the gas concentration, diverts it to the membrane separation concentration unit or the direct utilization unit, and recovers and utilizes the heat generated by the oxidation of the concentrated gas.

[0071] The various embodiments in this specification are described in a progressive manner. For the same or similar parts between the various embodiments, please refer to each other. Each embodiment focuses on describing the differences from other embodiments.

[0072] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit this application. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of this application.

Claims

1. An integrated intelligent assessment, control, and emission reduction system for coal seam gas extraction, characterized in that, It includes, in sequence, an intelligent sensing layer, a data processing layer, a control and execution layer, and an emission reduction and efficiency enhancement layer, wherein: The intelligent sensing layer includes multiple types of sensors arranged in coal seam boreholes, roadways, and extraction pipelines to obtain multi-source data during the coal seam gas extraction process; wherein, the multi-source data includes at least gas concentration, coal seam stress, extraction flow rate, equipment vibration frequency, and pipeline pressure parameters. The data processing layer includes an intelligent evaluation module and a control decision module. The intelligent evaluation module fuses and extracts features from the multi-source data based on a deep belief network with an improved attention mechanism, and outputs the extraction efficiency level, equipment health status, and carbon emission prediction values. The control decision module generates an optimal control strategy based on the evaluation results of the intelligent evaluation module, including the extraction pump power, valve opening, and pipeline topology. The control execution layer includes a variable frequency extraction pump, an intelligent throttling valve, and a branch pipeline switching device, which are used to receive and execute the control instructions generated by the control decision module, and to dynamically adjust the extraction parameters and reconstruct the pipeline topology in real time. The emission reduction and efficiency enhancement layer includes a gas classification and utilization module and a heat recovery module. The gas classification and utilization module is used to divert gas to a membrane separation concentration unit or a direct utilization unit according to the gas concentration. The heat recovery module is used to recover and utilize the heat generated by the oxidation of gas after concentration.

2. The system according to claim 1, characterized in that, The intelligent sensing layer includes multiple types of sensors, such as a gas concentration sensor using tunable semiconductor laser absorption spectroscopy, a coal seam stress sensor based on the fiber Bragg grating principle, and a drainage flow sensor arranged in the main pipeline and branch pipelines of the extraction pipeline. Each sensor collects data at a preset sampling frequency and is arranged in a distributed network to form a three-dimensional monitoring network covering coal seam boreholes, roadways, and extraction pipelines. The collected data is uploaded to the data processing layer via a wireless network.

3. The system according to claim 2, characterized in that, The deep belief network has a multi-layer structure including an input layer, at least one hidden layer, an attention layer, and an output layer, and its input is a multi-dimensional feature vector composed of the multi-source data. The reinforcement learning algorithm is the Q-Learning algorithm. The state space includes gas concentration distribution, coal seam stress state and equipment operating parameters, and the action space includes extraction pump power adjustment, valve opening adjustment and pipeline switching commands.

4. The system according to claim 1, characterized in that, The variable frequency pumping pump adopts a permanent magnet synchronous motor and vector control technology, and is equipped with a vector controller to achieve continuous power adjustment; The intelligent throttling valve adopts electro-hydraulic servo control technology to adjust the valve opening according to the optimal control strategy, thereby achieving precise control of the extraction flow rate of each branch pipeline. The branch pipeline switching device includes multiple electric valves, which are used to switch between different extraction pipeline topologies through a combination of valve switching.

5. The system according to claim 4, characterized in that, In the emission reduction and efficiency enhancement layer: The gas classification and utilization module also includes a waste gas oxidation device, which is used to oxidize and utilize the gas after it has been treated by the membrane separation and concentration unit. The heat recovery module is connected to the exhaust gas oxidation device and is used to recover the heat energy generated by the oxidation reaction.

6. The system according to claim 3, characterized in that, The reward function of the reinforcement learning algorithm is a weighted sum of extraction efficiency, system energy consumption, and carbon emissions, which is used to achieve multi-objective collaborative optimization when generating the optimal control strategy.

7. The system according to claim 2, characterized in that, The intelligent sensing layer also includes edge computing nodes, which are used to perform spatiotemporal alignment and noise reduction preprocessing on the raw data collected by the multi-type sensors before uploading it to the data processing layer.

8. The system according to claim 1, characterized in that, The system also includes a central control server, which is communicatively connected to the data processing layer and the control execution layer, respectively, for coordinating data flow and instruction issuance, and realizing human-computer interaction and system status monitoring.

9. A method for intelligent assessment, control, and emission reduction of coal seam gas extraction, characterized in that, include: By deploying various types of sensors in coal seam boreholes, roadways and extraction pipelines, gas concentration, coal seam stress, extraction flow rate, equipment vibration frequency and pipeline pressure parameters are collected in real time to obtain multi-source data during the coal seam gas extraction process. A deep belief network based on an improved attention mechanism is used to fuse and extract features from the multi-source data, and output the extraction efficiency level, equipment health status and carbon emission prediction values. Based on reinforcement learning algorithms, an optimal control strategy is generated, including pump power, valve opening and pipeline topology, according to the extraction efficiency level, equipment health status and carbon emission prediction. The optimal control strategy is executed by a variable frequency extraction pump, an intelligent throttling valve, and a branch pipeline switching device to achieve dynamic adjustment of extraction parameters and real-time reconstruction of pipeline topology. Depending on the gas concentration, the gas is diverted to a membrane separation concentration unit or a direct utilization unit, and the heat generated by the oxidation of the gas after concentration is recovered and utilized.

Images