An intelligent control system and method for a coal mine gas extraction pipe network

Abstract

The application discloses a kind of coal mine gas extraction pipe network intelligent regulation and control system and method, it is related to intelligent regulation and control technical field, including data acquisition module, pipe network analysis module, intelligent regulation and control module, effect evaluation module and early warning response module, data acquisition module gathers gas concentration, gas flow, extraction negative pressure, pipeline temperature and valve opening degree real-time operation parameter, pipe network analysis module establishes pipe network topological model, calculates resistance coefficient and pipe network total resistance coefficient, intelligent regulation and control module with gas concentration deviation, gas flow deviation and extraction pump power three weighted and minimization as target to formulate regulation and control instruction, three weighted coefficients are adjusted with pipe network total resistance coefficient automatically;Valve opening degree and branch negative pressure, pump station frequency and liquid supply flow meet linkage constraint between during regulation and control instruction execution process, realizes the on-line self-adapting correction of regulation and control model and whole pipe network negative pressure dynamic equalization distribution, improves gas extraction pipe network regulation and control precision and system operation stability.

Description

Technical Field

[0001] This invention relates to the field of intelligent control technology, and in particular to an intelligent control system and method for coal mine gas extraction pipeline networks. Background Technology

[0002] Gas drainage is the fundamental means of controlling gas disasters in coal mines. By pumping gas from the coal seam to the surface through drainage pipelines, the gas content in the coal seam can be reduced, the amount of gas emitted underground can be decreased, and gas resources can be recovered and utilized. As the mining depth of coal mines continues to increase, the scale of gas drainage pipeline networks is becoming increasingly large, with intersecting branch pipelines and dispersed drainage areas, making the operating conditions of the pipeline network system increasingly complex.

[0003] To achieve precise control of gas extraction pipeline networks, numerous technical solutions have been proposed in recent years. For example, one solution involves analyzing the resistance coefficients of each pipeline segment and the characteristic parameters of the extracted gas source through data acquisition, constructing a mixed gas flow model for the pipeline network, and establishing an intelligent decision-making and control model with the goal of maximizing the pure gas volume in the pipeline network and valve opening as the decision variable. Another solution uses multi-parameter sensing data to invert the distribution of friction resistance along the pipeline and the local resistance increment, calculates the estimated value of the deep effective negative pressure, and generates pump station settings and branch valve position settings accordingly. Yet another solution employs a "pipeline network solution-genetic algorithm optimization" approach, constructing a graph theory model of the gas extraction pipeline network based on graph theory principles, and optimizing the pipe diameter with the goal of minimizing pipeline investment. Furthermore, existing research has proposed establishing a multi-objective optimization mathematical model, using an improved genetic algorithm for initial solution, and then performing secondary optimization based on an improved BP neural network.

[0004] While the above-mentioned technical solutions have improved the level of intelligence in the control of gas drainage pipeline networks to varying degrees, they still have the following shortcomings:

[0005] First, the control model operates in a fixed manner and cannot be automatically corrected based on the actual control effect. In existing solutions, after the control command is issued, the system usually does not track and evaluate the control effect, and the control model cannot automatically correct its parameters based on the deviation between the actual control effect and the expected effect. When the resistance coefficient of the pipeline changes due to coal dust deposition or pipeline corrosion, the control commands based on the original model gradually deviate from the actual operating conditions, and the control accuracy decreases over time.

[0006] Secondly, the distribution of negative pressure during gas extraction relies on a static model and cannot be automatically redistributed based on changes in the resistance coefficient. The rational distribution of negative pressure during gas extraction is crucial for ensuring the efficient operation of the extraction system. While existing solutions have constructed pipeline network calculation models for optimizing negative pressure distribution, these models are static once established. When the resistance coefficient of a certain pipe section changes, the system cannot automatically recalculate the resistance coefficient of the entire pipeline network and redistribute the target negative pressure values ​​for each branch. Existing control systems struggle to differentiate and control different boreholes, making it difficult to achieve a rational and accurate distribution of negative pressure for each branch. Summary of the Invention

[0007] To address the above issues, this invention achieves online correction of the control model through effect evaluation and model modification, and realizes dynamic balance of negative pressure in each branch by real-time recalculation of the resistance coefficient of the entire pipeline network and proportional distribution of negative pressure.

[0008] The technical solution is an intelligent control system for coal mine gas drainage pipelines, comprising:

[0009] The data acquisition module collects real-time operating parameters of the gas extraction pipeline network, including gas concentration, gas flow rate, extraction negative pressure, pipeline temperature, and valve opening at each pipeline node.

[0010] The pipeline analysis module establishes a pipeline topology model, calculates the resistance coefficient of each pipe segment and the total resistance coefficient of the pipeline network, determines the correlation value between the deployment location of each pipeline node and the characteristics of operating parameters and constructs a correlation matrix, and classifies the real-time operating parameters based on the correlation matrix to obtain the classification results.

[0011] The intelligent control module formulates control instructions based on the classification results. These instructions include the adjustment amount of each valve opening, the target extraction negative pressure value of each branch, and the target operating parameters of the extraction pump station. The module adjusts the operating status of the gas extraction pipeline network and jointly controls the operating parameters of the extraction pump station according to these instructions. The intelligent control module uses the minimization of the weighted sum of the absolute values ​​of the deviations between the gas concentration at each pipeline node and the target gas concentration, the absolute values ​​of the deviations between the gas flow rate at each pipeline segment and the target gas flow rate, and the real-time power of the extraction pump station as its control objective. The optimal control parameters are obtained by solving for this objective. The three weighting coefficients of the control objective automatically adjust according to the change in the total resistance coefficient of the pipeline network.

[0012] The effect evaluation module continuously monitors the actual gas concentration and actual gas flow of each pipeline node after the control command is executed, and calculates the control deviation value by comprehensively calculating the deviation between the actual value and the expected value. When the control deviation value exceeds the set threshold, a model correction signal is issued.

[0013] The pipeline analysis module receives the model correction signal and adjusts the correlation degree value in the correlation matrix and the calculation parameters of the resistance coefficient according to the model correction signal.

[0014] Furthermore, the pipeline analysis module establishes a pipeline topology model and calculates the resistance coefficient of each pipe segment and the total pipeline resistance coefficient, including:

[0015] Construct a directed graph of the pipeline network topology based on the pipeline connection relationship, with pipeline intersections as nodes and each pipeline segment as an edge; for any edge, take the quotient of the extraction negative pressure difference between the two nodes of the edge and the square of the gas flow rate through the edge as the resistance coefficient of the edge, and add the resistance coefficients of all edges to obtain the total resistance coefficient of the pipeline network.

[0016] The pipeline analysis module determines the correlation value between the deployment location and operating parameter characteristics of each pipeline node and constructs a correlation matrix, including:

[0017] Determine the correlation value between the deployment location of each pipeline node and its operating parameter characteristics, and construct an association matrix with pipeline nodes as rows and operating parameter characteristics as columns. The value of each element in the association matrix represents the strength of the association between the corresponding pipeline node and the corresponding operating parameter characteristics.

[0018] The pipeline network analysis module categorizes the real-time operating parameters based on the correlation matrix to obtain the categorization results, including:

[0019] The real-time operating parameters are categorized based on the correlation matrix to obtain the categorization results.

[0020] 3. The intelligent control system for coal mine gas extraction pipeline network according to claim 1, characterized in that the amplitude of the model correction signal is positively correlated with the magnitude of the control deviation value, and the adjustment amount of the correlation degree value in the correlation matrix is ​​positively correlated with the amplitude of the model correction signal; the monitoring duration of the effect evaluation module after the execution of the control command is negatively correlated with the magnitude of the control deviation value;

[0021] When the change in the real-time resistance coefficient of any pipe segment compared to the previous moment exceeds a set threshold, the pipeline analysis module initiates a recalculation of the resistance coefficient of the entire pipeline network and redistributes the target extraction negative pressure value of each branch based on the recalculated resistance coefficient of each pipe segment. The redistributed target extraction negative pressure value of each branch is allocated according to the proportion of the current resistance coefficient of each branch to the total resistance coefficient, with branches having higher resistance coefficients receiving higher target extraction negative pressure values.

[0022] Furthermore, when the intelligent control module jointly controls the operating parameters of the extraction pump station, there are linkage constraints between the various control parameters:

[0023] When the valve opening of a certain branch is adjusted, the target extraction negative pressure value of that branch is recalculated based on the current resistance coefficient of that branch. The valve opening adjustment amount and the target extraction negative pressure value adjustment amount satisfy the resistance characteristic curve constraint of that branch.

[0024] When the target operating frequency of the extraction pump station is adjusted, the target value of the liquid supply flow rate of the extraction pump station is adjusted in conjunction with the fitting relationship between the target operating frequency and the liquid supply flow rate. The adjustment amount of the operating frequency and the adjustment amount of the liquid supply flow rate satisfy the liquid supply characteristic curve constraint of the extraction pump station.

[0025] The deviation between the sum of the target extraction negative pressure values ​​after the redistribution of each branch and the actual supply extraction negative pressure value of the extraction pump station is used as a constraint condition in solving the control target.

[0026] Furthermore, when the intelligent control module solves for the control target, the control target is the sum of the absolute values ​​of the deviations between the gas concentration of each pipeline node and the target gas concentration multiplied by a first weighting coefficient, the sum of the absolute values ​​of the deviations between the gas flow rate of each pipeline segment and the target gas flow rate multiplied by a second weighting coefficient, and the real-time power of the extraction pump station multiplied by a third weighting coefficient. The minimum value obtained by adding these three values ​​is taken. Solving for this minimum value yields the optimal valve opening combination and the optimal operating power of the extraction pump station. When the total resistance coefficient of the pipeline increases, the weighting coefficients of the gas flow rate deviation and the weighting coefficients of the extraction pump station power automatically increase.

[0027] Furthermore, the pipeline analysis module determines the correlation value between the deployment location and operating parameter characteristics of each pipeline node, including:

[0028] The historical gas extraction pipeline monitoring data were preprocessed, and outliers were treated as missing values ​​and filled using sliding Lagrange interpolation.

[0029] For each pipeline node, an autoregressive integral moving average model is constructed to analyze the time series characteristics of historical gas concentration data and historical gas flow data. The autoregressive integral moving average model uses a linear combination of historical observation values ​​and historical prediction errors to describe the current value.

[0030] The sequential learning algorithm is applied to adjust the model parameters of the autoregressive integral moving average model based on the rate of change of gas concentration, the fluctuation range of gas flow, the rate of change of extraction negative pressure, the rate of change of pipeline temperature, and the valve opening response characteristics.

[0031] The correlation value between each pipeline node and each operating parameter characteristic is determined based on the adjusted autoregressive integral moving average model and filled into the correlation matrix.

[0032] Furthermore, the pipeline network analysis module categorizes the real-time operating parameters based on the correlation matrix, including:

[0033] For each pipeline node, the corresponding operating parameter characteristics of the pipeline node in the correlation matrix are used as transactions to construct a frequent pattern tree. The nodes of the frequent pattern tree represent operating parameter characteristics, and the node weights represent the frequency of the operating parameter characteristics in the transactions.

[0034] By mining common combinations of operating parameter characteristics among multiple pipeline nodes through the frequent pattern tree, pipeline nodes with the same combination of operating parameter characteristics are assigned the same operating category identifier.

[0035] The real-time operating parameters of each pipeline node are compared with the extracted operating parameter characteristics, and the real-time operating parameters are classified according to the comparison results.

[0036] Furthermore, the intelligent control module jointly controls the operating parameters of the extraction pumping station, including:

[0037] The target operating frequency of the pumping station is obtained by multiplying the ratio of the target operating power of the pumping station to the rated power of the pumping station in the control command by the rated operating frequency of the pumping station.

[0038] The target value of the liquid supply flow rate is calculated based on the linear relationship between the target operating frequency of the extraction pump station and the liquid supply flow rate.

[0039] The regulating device generates control commands based on the target operating frequency of the extraction pump station and the target value of the liquid supply flow rate, and sends them to the extraction pump station.

[0040] Furthermore, it also includes an early warning response module, which sets the following early warning thresholds: gas concentration threshold range, extraction negative pressure threshold range, pipeline temperature threshold range, carbon monoxide concentration safety threshold, cumulative pure gas quantity predetermined value, and normal operating parameters range of the extraction pump station; when any real-time operating parameter is detected to meet the corresponding early warning condition, the early warning mechanism is activated and emergency response measures are executed, including: automatically stopping extraction operation when the carbon monoxide concentration exceeds the standard or the cumulative pure gas quantity reaches the predetermined value, and performing fault diagnosis and prompting maintenance when the extraction pump station is not operating normally;

[0041] There is a linkage between the model correction signal issued by the effect evaluation module and the early warning response module: when the control deviation value exceeds the set threshold and the duration exceeds the set duration, the effect evaluation module sends an early warning trigger signal to the early warning response module, and the early warning response module lowers the trigger condition of the early warning threshold.

[0042] 10. A method for intelligent control of coal mine gas drainage pipeline network, comprising the following steps:

[0043] S1, collect real-time operating parameters of the gas extraction pipeline network, including gas concentration, gas flow rate, extraction negative pressure, pipeline temperature and valve opening at each pipeline node;

[0044] S2. Construct a directed graph of the pipeline network topology based on the pipeline connection relationship, with the pipeline intersection points as nodes and each pipeline segment as an edge; for any edge, take the quotient of the extraction negative pressure difference between the two nodes of the edge and the square of the gas flow rate through the edge as the resistance coefficient of the edge, and add the resistance coefficients of all edges to obtain the total resistance coefficient of the pipeline network.

[0045] S3, determine the correlation value between the deployment location of each pipeline node and the operating parameter characteristics, construct an association matrix with pipeline nodes as rows and operating parameter characteristics as columns, where each element value in the association matrix represents the strength of the association between the corresponding pipeline node and the corresponding operating parameter characteristics; classify the real-time operating parameters based on the association matrix to obtain the classification results;

[0046] S4. Based on the classification results, control instructions are formulated, including the adjustment amount of each valve opening, the target extraction negative pressure value of each branch, and the target operating parameters of the extraction pump station. The operating status of the gas extraction pipeline network and the operating parameters of the extraction pump station are adjusted according to the control instructions. The optimal control parameters are solved by minimizing the weighted sum of the absolute values ​​of the deviations between the gas concentration of each pipeline node and the target gas concentration, the absolute values ​​of the deviations between the gas flow rate of each pipeline section and the target gas flow rate, and the real-time power of the extraction pump station. The three weighting coefficients of the control objective are automatically adjusted according to the change of the total resistance coefficient of the pipeline network.

[0047] S5. After the control command is executed, the actual gas concentration and actual gas flow of each pipeline node are continuously monitored. The average of the square of the deviation between the actual gas concentration and the expected gas concentration of each pipeline node and the average of the square of the deviation between the actual gas flow and the expected gas flow of each pipeline segment are added together and the square root is taken as the control deviation value. When the control deviation value exceeds the set threshold, a model correction signal is issued.

[0048] S6, adjust the correlation degree value in the correlation matrix and the calculation parameters of the resistance coefficient according to the model correction signal;

[0049] S7. When the real-time operating parameters meet the early warning conditions, the early warning mechanism is activated and emergency response measures are executed.

[0050] Due to the adoption of the above technical solutions, the present invention has the following advantages compared with the prior art;

[0051] 1. The effect evaluation module calculates the deviation between actual and expected values ​​to generate a control deviation value. When the deviation exceeds a threshold, the amplitude of the correction signal is positively correlated with the magnitude of the deviation. Based on this, the pipeline analysis module adjusts the correlation degree value of the corresponding nodes in the correlation matrix, and simultaneously adjusts the window size of the sliding Lagrange interpolation and the order of the autoregressive integral moving average model. This ensures that the model continuously matches the actual operating conditions of the pipeline network, and the control accuracy does not decrease over time. This achieves online correction of the control model based on the actual control effect, solving the problem of control accuracy decreasing over time due to fixed model operation. 2. When the change in the resistance coefficient of any pipe section compared to the previous moment exceeds a set threshold, the resistance coefficient of the entire pipeline network is automatically recalculated. After recalculation, the target extraction negative pressure value of each branch is redistributed according to the proportion of the current resistance coefficient of each branch to the total resistance coefficient. The branch with the larger resistance coefficient receives a higher target extraction negative pressure value. The sum of the target negative pressure values ​​of all branches is equal to the actual supply negative pressure value of the pumping station. At the same time, the valve opening adjustment and the target extraction negative pressure value adjustment satisfy the resistance characteristic curve constraint, and the pumping station operating frequency adjustment and the liquid supply flow adjustment satisfy the liquid supply characteristic curve constraint, ensuring that the negative pressure is allocated to each branch as needed, and achieving dynamic balance of negative pressure in the entire pipeline network. Attached Figure Description

[0052] Figure 1 This is a flowchart of the system modules of the present invention. Detailed Implementation

[0053] The foregoing and other technical contents, features and effects of the present invention are described in conjunction with the appendix below. Figure 1 The detailed description of the embodiments will make this clear. All structural details mentioned in the following embodiments are based on the accompanying drawings.

[0054] Based on existing technology, in Embodiment 1, this embodiment provides an intelligent control system for coal mine gas drainage pipeline networks, including a data acquisition module, a pipeline network analysis module, an intelligent control module, an effect evaluation module, and an early warning response module. The data acquisition module is used to collect real-time operating parameters of the gas drainage pipeline network, including gas concentration, gas flow rate, drainage negative pressure, pipeline temperature, and valve opening at each pipeline node.

[0055] The data acquisition module includes multiple sensors distributed at various nodes of the gas extraction pipeline network. Among them, the gas concentration sensor is an infrared methane sensor with a range of 0 to 100% CH4; the gas flow sensor is a V-cone flow meter with a range of 0 to 100 cubic meters per minute; the extraction negative pressure sensor is a pressure transmitter with a range of -100 to 0 kPa; the pipeline temperature sensor is a platinum resistance thermometer with a range of -50 to 150 degrees Celsius; and the valve opening is obtained through feedback from the valve positioner. The data acquisition module transmits the collected real-time operating parameters to the pipeline analysis module and the intelligent control module, respectively, and transmits the actual gas concentration and actual gas flow after the control command is executed to the effect evaluation module.

[0056] The pipeline analysis module is used to establish a pipeline topology model, calculate the resistance coefficient of each pipeline segment and the total resistance coefficient of the pipeline network, determine the correlation value between the deployment location of each pipeline node and the characteristics of the operating parameters and construct a correlation matrix, and classify the real-time operating parameters based on the correlation matrix to obtain the classification results.

[0057] The pipeline analysis module constructs a directed graph of the pipeline topology based on pipeline connections, using pipeline intersections as nodes and each pipeline segment as an edge. For any edge, the resistance coefficient is calculated as the quotient of the negative pressure difference between the two nodes of that edge and the square of the gas flow rate passing through that edge. The basis for calculating the resistance coefficient by dividing the negative pressure difference by the square of the flow rate is that in gas extraction pipelines, resistance loss is proportional to the square of the flow rate. Then, the resistance coefficients of all edges are summed to obtain the total pipeline resistance coefficient, which quantifies the overall resistance level of the entire pipeline system.

[0058] When the change in the real-time resistance coefficient of any pipe segment compared to the previous moment exceeds a set threshold, the resistance coefficient of the entire pipeline network is recalculated. The set threshold is determined as follows: when the mine has accumulated no less than thirty sets of historical resistance coefficient change data, the 90th percentile of the historical resistance coefficient change sequence is taken as the threshold; when there are fewer than thirty sets of historical data, 10% of the resistance coefficient of the pipe segment under pure water conditions is taken as the initial threshold.

[0059] The method for obtaining the resistance coefficient under pure water conditions is as follows: During the shutdown and maintenance of the gas extraction pump station, the drag-reducing fluid injection system of the extraction pump station is shut down, and only pure water is used as the working fluid. After the extraction pump station is running stably (running time of not less than 30 minutes and the negative pressure fluctuation of extraction does not exceed ±5%), the resistance coefficient of the pipe section is collected and calculated according to the above resistance coefficient calculation formula. Five consecutive collections are taken and the average value is taken. When the historical data accumulates to 30 sets, the threshold determination method is automatically switched from the initial threshold to the 90ths percentile based on the historical data. A smooth transition method is adopted during the switch, that is, the new threshold is the weighted average of the initial threshold and the historical data threshold. Initially, the historical data threshold weight is 10%, and thereafter, the historical data threshold weight increases by 10% for every 10 additional sets of data, until the historical data threshold weight reaches 100%.

[0060] The redistributed target extraction negative pressure values ​​for each branch are allocated according to the proportion of each branch's current resistance coefficient to the sum of all branch resistance coefficients. Specifically, the target extraction negative pressure value for each branch is obtained by dividing the branch's resistance coefficient by the total network resistance coefficient and then multiplying by the actual negative pressure supplied by the extraction pump station. A branch with a higher resistance coefficient means that airflow is more difficult to pass through, requiring a higher extraction negative pressure to maintain normal extraction flow; conversely, a branch with a lower resistance coefficient requires a lower negative pressure, thus achieving a dynamic balance of negative pressure across the entire network.

[0061] The pipeline network analysis module determines the correlation between the deployment location of each pipeline node and its operational parameter characteristics and constructs a correlation matrix. These operational parameter characteristics include the rate of change of gas concentration, the amplitude of gas flow fluctuation, the rate of change of extraction negative pressure, the rate of change of pipeline temperature, and the valve opening response characteristics. The rate of change of gas concentration is defined as the derivative of gas concentration with respect to time; the amplitude of gas flow fluctuation is defined as the difference between the maximum and minimum gas flow rates per unit time; the rate of change of extraction negative pressure is defined as the derivative of extraction negative pressure with respect to time; the rate of change of pipeline temperature is defined as the derivative of pipeline temperature with respect to time; and the valve opening response characteristics are defined as the rate of response of valve opening with respect to time.

[0062] The process for determining the correlation coefficient is as follows: First, the historical gas extraction pipeline monitoring data is preprocessed, outliers are identified as missing values, and sliding Lagrange interpolation is applied to fill them in. The outlier identification method is as follows: calculate the historical mean and standard deviation of each monitoring parameter, and mark data points exceeding the mean plus or minus three times the standard deviation as outliers. The specific method of sliding Lagrange interpolation is as follows: for a missing value at a certain moment, take data from five time points before and after that moment, for a total of ten data points, construct a ninth-order Lagrange interpolation polynomial, and use the value of this polynomial at the missing moment as the estimated value of the missing value.

[0063] Then, for each pipeline node, an autoregressive integral moving average model is constructed to analyze the time series characteristics of historical gas concentration and historical gas flow data. The autoregressive integral moving average model uses a linear combination of historical observations and historical prediction errors to describe the current value. For the historical gas concentration series of each node, the stationarity of the series is determined by a unit root test, with a significance level of 0.05. If the series is non-stationary, differencing is performed until stationarity is achieved. The number of autoregressive terms and moving average terms are determined through autocorrelation and partial autocorrelation function analysis, and the model parameters are estimated using maximum likelihood estimation.

[0064] Then, a sequential learning algorithm is applied to adjust the model parameters of the autoregressive integral moving average model based on the rate of change of gas concentration, the fluctuation range of gas flow, the rate of change of extraction negative pressure, the rate of change of pipeline temperature, and the valve opening response characteristics. The sequential learning algorithm uses recursive least squares to update the model parameters online: the gain vector is calculated using the regression vector and covariance matrix at the current time, and then the model parameters are updated using the gain vector and the prediction error at the current time, while simultaneously updating the covariance matrix according to the forgetting factor. The forgetting factor ranges from 0.95 to 0.99, initially set at 0.98. When the rate of change of gas concentration increases, the forgetting factor is decreased to accelerate the model's response to new data; when the fluctuation range of gas flow increases, the forgetting factor is increased to enhance the model's ability to suppress noise.

[0065] Then, based on the adjusted autoregressive integral moving average model, the correlation degree between each pipeline node and each operating parameter characteristic is determined and filled into the correlation matrix. Specifically, the determination method is as follows: the autoregressive coefficient sequence of the adjusted autoregressive integral moving average model is used as the feature sequence of the node, and the sequences of each operating parameter characteristic are used as comparison sequences to calculate the grey correlation degree. The calculation steps are as follows: first, the feature sequence and comparison sequence are dimensionless (using mean value processing); then, the correlation coefficient at each time point is calculated, with the resolution coefficient set to 0.5; finally, the average of the correlation coefficients at each time point is taken as the correlation degree value. The correlation degree value is between zero and one; a value greater than 0.8 indicates a strong correlation, a value between 0.6 and 0.8 indicates a moderate correlation, and a value less than 0.6 indicates a weak correlation. All calculated correlation degree values ​​are filled into the correlation matrix with pipeline nodes as rows and operating parameter characteristics as columns.

[0066] The pipeline analysis module classifies the real-time operating parameters based on the correlation matrix as follows: For each pipeline node, all element values ​​in the corresponding row of the correlation matrix are extracted, and the operating parameter characteristics corresponding to element values ​​greater than 0.6 are taken as transaction items for that node. Then, for each pipeline node, the corresponding operating parameter characteristics in the correlation matrix are taken as transactions, and a frequent pattern tree is constructed. The process of constructing the frequent pattern tree is as follows: First, all node transactions are scanned, and the frequency of each operating parameter characteristic in all transactions is counted, and the characteristics are sorted from high to low frequency; then, all node transactions are scanned again, and for each transaction, the characteristics are inserted into the frequent pattern tree in order of frequency. The root node of the frequent pattern tree is an empty node, and each characteristic is taken as a node. The node records the number of times the characteristic appears on the path. If multiple transactions have the same prefix characteristic sequence, they share the same path.

[0067] Then, the common operating parameter characteristic combinations among multiple pipeline nodes are mined using the frequent pattern tree. A frequent pattern growth algorithm is used for mining, with a preset minimum support threshold of three. Finally, all characteristic combinations with support not lower than this threshold are obtained. Then, pipeline nodes with the same operating parameter characteristic combination are assigned the same operating category identifier: each mined frequent characteristic combination is considered a category; for each pipeline node, the frequent characteristic combinations contained in its transactions are found, and the node is assigned to the corresponding category. If a node's transactions contain multiple frequent characteristic combinations, the characteristic combination with the highest support is selected as the node's category. Then, the real-time operating parameters of each pipeline node are compared with the mined operating parameter characteristic combinations, and the real-time operating parameters are categorized according to the comparison results. The real-time operating parameters of all pipeline nodes in the same category are grouped into the same data set.

[0068] The intelligent control module is used to formulate control instructions based on the classification results. The control instructions include the adjustment amount of the opening of each valve, the target extraction negative pressure value of each branch, and the target operating parameters of the extraction pump station. Based on the control instructions, the module adjusts the operating status of the gas extraction pipeline network and jointly controls the operating parameters of the extraction pump station.

[0069] The intelligent control module aims to minimize the weighted sum of three factors: the sum of the absolute deviations of the gas concentration at each pipeline node from the target gas concentration, the sum of the absolute deviations of the gas flow rate at each pipeline segment from the target gas flow rate, and the real-time power of the extraction pump station. Optimal control parameters are obtained by solving for these control objectives. The particle swarm optimization algorithm is used to solve for these objectives, with a particle swarm size of one hundred particles, an inertia weight linearly decreasing from 0.9 to 0.4, an acceleration constant of 2.0, and a maximum of five hundred iterations. Each dimension of the particle position is constrained to be within the range of the corresponding decision variable, and the velocity is constrained to be ten percent of the range of each dimension. The termination condition is that the change in the objective function value over fifty consecutive iterations is less than one part per million.

[0070] The selection of the above algorithm parameters is based on the characteristics of the optimization problem in this application: the dimension of the decision variable is the sum of the number of nodes and the number of pipe segments in the pipeline network. For a typical mine pipeline network, the number of nodes is generally ten to thirty, the number of pipe segments is generally eight to twenty-five, and the dimension of the decision variable is generally eighteen to fifty-five. The particle swarm size is two to three times the dimension of the decision variable, and one hundred particles can cover an optimization space of eighteen to fifty dimensions. The inertia weight is decreased from 0.9 to 0.4, which gives the algorithm a strong global exploration capability in the early stage and a strong local development capability in the later stage. The acceleration constant of 2.0 is a typical recommended value in particle swarm optimization algorithms, balancing the learning tendency of particles towards individual optimal positions and global optimal positions. The maximum number of iterations is set to 500. Within 500 iterations, the change in the objective function value is less than one part per million, and further increasing the number of iterations will improve the quality of the solution by less than 0.5%. The control target is specifically calculated by multiplying the sum of the absolute values ​​of the deviations between the gas concentration at each pipeline node and the target gas concentration by a first weighting coefficient, the sum of the absolute values ​​of the deviations between the gas flow rate at each pipeline section and the target gas flow rate by a second weighting coefficient, and the real-time power of the extraction pump station by a third weighting coefficient. The minimum value obtained by adding these three values ​​is taken. The target gas concentration is set according to coal mine safety regulations and extraction compliance requirements, while the target gas flow rate is determined based on the extraction design and the number of boreholes. Solving for this minimum value yields the optimal valve opening combination and the optimal operating power of the extraction pump station.

[0071] The three weighting coefficients of the control target automatically adjust according to the change of the total resistance coefficient of the pipeline network: Let the baseline value of the total resistance coefficient of the pipeline network be the total resistance coefficient under pure water conditions. Initially, the first weighting coefficient is 0.5, the second weighting coefficient is 0.3, and the third weighting coefficient is 0.2, with the sum of the three initial weighting coefficients being 1.0. When the total resistance coefficient of the pipeline network is less than 1.1 times the baseline value, the weighting coefficients remain unchanged from their initial values. When the total resistance coefficient of the pipeline network is greater than 1.1 times the baseline value, the second and third weighting coefficients automatically increase as follows: the second weighting coefficient is calculated by multiplying the initial second weighting coefficient by the adjustment coefficient (within parentheses) by the portion of the total resistance coefficient exceeding 1.1 times the baseline value, divided by the value of 1.1 times the baseline value; the third weighting coefficient is calculated by multiplying the initial third weighting coefficient by the adjustment coefficient (within parentheses) by the portion of the total resistance coefficient exceeding 1.1 times the baseline value, divided by the value of 1.1 times the baseline value. The two adjustment coefficients were determined as follows: using the gas flow rate compliance rate and the power consumption per unit of pure gas extraction at the extraction pumping station as evaluation indicators, an experiment was conducted using a three-factor, three-level orthogonal array. The three factors were: the multiple by which the total resistance coefficient exceeded the benchmark value (at levels of 1.2, 1.5, and 2.0); the adjustment coefficient of the second weighting coefficient (at levels of 0.5, 1.0, and 1.5); and the adjustment coefficient of the third weighting coefficient (at levels of 0.4, 0.8, and 1.2). Nine experiments were conducted, recording the flow rate compliance rate and unit power consumption values ​​for each experiment. Range analysis (calculating the average value of the indicators at each level for each factor, and selecting the optimal level for that factor) was used to obtain the parameter combination that maximized the flow rate compliance rate and minimized the unit power consumption. The value of the first weighting coefficient was correspondingly reduced to ensure that the sum of the three weighting coefficients remained at 1.0. When the intelligent control module jointly controls the operating parameters of the extraction pumping station, there are linkage constraints between the control parameters. When the valve opening of a certain branch is adjusted, the target extraction negative pressure value of that branch is recalculated based on the current resistance coefficient of that branch. The valve opening adjustment amount and the target extraction negative pressure value adjustment amount satisfy the resistance characteristic curve constraint of that branch.

[0072] The method for obtaining the resistance characteristic curve is as follows: During the shutdown and maintenance of the gas extraction pump station, all valves except those on the branch are closed. At seven valve opening levels (0°, 15°, 30°, 45°, 60°, 75°, and 90°), the negative pressure difference and flow rate at both ends of the branch are measured. After stable operation for ten minutes at each opening level, five sets of data are continuously recorded and averaged. The scattered data of the negative pressure difference and flow rate are then fitted using a quadratic polynomial to obtain the resistance characteristic equation. The least squares method is used to minimize the sum of squared residuals, and the goodness of fit must be greater than 0.95. If the goodness of fit does not meet the requirement, the number of opening levels is increased to eleven for recalibration. After calibration, the fitting coefficients are stored in the system database. The calibration validity period is three months; recalibration is required after the validity period expires or after a major system overhaul. When the valve opening of a branch is adjusted, the resistance coefficients corresponding to the opening before and after the adjustment are first determined based on the resistance characteristic curve of that branch. The adjustment amount of the target extraction negative pressure value is calculated by dividing the adjusted resistance coefficient by the resistance coefficient before the adjustment, multiplying by the current target extraction negative pressure value, and then subtracting the current target extraction negative pressure value. This achieves a linkage and matching between the valve opening adjustment amount and the target negative pressure value adjustment amount. When the target operating frequency of the extraction pump station is adjusted, the target value of the liquid supply flow rate of the extraction pump station is adjusted in linkage based on the fitting relationship between the target operating frequency and the liquid supply flow rate. The adjustment amount of the operating frequency and the adjustment amount of the liquid supply flow rate satisfy the constraints of the liquid supply characteristic curve of the extraction pump station.

[0073] The method for obtaining the liquid supply characteristic curve is as follows: Under the rated operating conditions of the pumping station, the operating frequency is set to seven levels: 40%, 50%, 60%, 70%, 80%, 90%, and 100% of the rated frequency. After stable operation for 15 minutes at each frequency level, the liquid supply flow rate is measured using an electromagnetic flowmeter. Five sets of data are continuously recorded at each frequency level, and the average value is taken. After obtaining the scatter plot data of frequency and flow rate, linear regression is used to fit and obtain the liquid supply characteristic equation. The least squares method is used to minimize the sum of squares of the fitting residuals, and the goodness of fit must be greater than 0.95. If the goodness of fit does not meet the requirements, the number of frequency levels is increased to eleven for recalibration. After calibration, the fitting coefficients are stored in the system database. The calibration validity period is six months. Recalibration is required after the validity period expires or after the liquid supply pump is replaced. The deviation between the sum of the target negative pressure values ​​of each branch after redistribution and the actual supply negative pressure value of the pumping station is used as a constraint in solving the control target. This constraint is that the absolute value of the difference between the sum of the target negative pressure values ​​of each branch and the actual supply negative pressure value of the pumping station does not exceed five percent of the actual supply negative pressure value of the pumping station. If this constraint cannot be met during the optimization process, the target negative pressure values ​​of each branch are reduced proportionally, prioritizing branches with lower negative pressures to meet the minimum extraction requirements. The specific process of the intelligent control module jointly controlling the operating parameters of the pumping station is as follows: the ratio of the target operating power of the pumping station to the rated power of the pumping station in the control command is multiplied by the rated operating frequency of the pumping station to obtain the target operating frequency of the pumping station. Then, the target value of the supply flow rate is calculated based on the linear relationship between the target operating frequency of the pumping station and the supply flow rate, which is obtained by fitting a supply characteristic curve. Finally, a control command is generated based on the target operating frequency of the pumping station and the target value of the supply flow rate and sent to the regulating device of the pumping station. The effect evaluation module is used to continuously monitor the actual gas concentration and actual gas flow of each pipeline node after the control command is executed, and to calculate the control deviation value by comprehensively calculating the deviation between the actual value and the expected value. When the control deviation value exceeds the set threshold, a model correction signal is issued.

[0074] After the control command is executed, the effect evaluation module continuously monitors the actual gas concentration and actual gas flow rate of each pipeline node at a sampling period of one minute. It internally stores the expected gas concentration and expected gas flow rate based on the control command. The control deviation value is calculated as follows: the average of the squared difference between the actual and expected concentrations at each node, and the average of the squared difference between the actual and expected flow rates at each pipeline segment, are added together and the square root is taken as the control deviation value. The threshold value is determined as follows: thirty historically effective control periods (two hours each) are selected from the mine's data. The control deviation value for each period is calculated, and the 95th percentile of these thirty control deviation values ​​is used as the threshold value. If there are fewer than thirty historical data sets, the initial threshold value is set to 5% of the target gas concentration value. Once thirty historical data sets are accumulated, the threshold value determination method automatically switches from the initial value to the 95th percentile based on historical data.

[0075] A smooth transition is adopted during the switching process, that is, the new threshold is the weighted average of the initial threshold and the historical data threshold. Initially, the historical data threshold has a weight of 10%, and thereafter, the weight of the historical data threshold increases by 10% for every ten additional sets of data, until the weight of the historical data threshold reaches 100%. When the control deviation exceeds the set threshold, the effect evaluation module issues a model correction signal. The amplitude of the model correction signal is positively correlated with the magnitude of the control deviation: when the control deviation does not exceed the threshold, the amplitude of the correction signal is zero; when the control deviation exceeds the threshold, the amplitude of the correction signal is the preset proportional coefficient multiplied by the value of the control deviation exceeding the threshold. The preset proportional coefficient is 2.0, and it is determined by selecting twenty time periods in the mine's history where the deviation exceeded the threshold, testing the number of control cycles required for the system to recover stability under six values ​​of proportional coefficient: 0.5, 1.0, 1.5, 2.0, 2.5, and 3.0, respectively, and calculating the average recovery cycle for each of the twenty time periods under each value. The proportional coefficient that minimizes the average recovery cycle and prevents oscillations in all time periods is selected.

[0076] The effect evaluation module observes a negative correlation between the monitoring duration after the execution of the control command and the magnitude of the control deviation value. Specifically, when the control deviation value is small, the monitoring duration is the preset standard monitoring duration of thirty minutes; when the control deviation value exceeds a threshold, the monitoring duration is extended based on the standard duration by multiplying the result of a preset extension coefficient by the relative amount of the deviation value exceeding the threshold. The preset extension coefficient is set to 1.0, determined by conducting five sets of comparative tests with extension coefficients of 0.5, 0.75, 1.0, 1.25, and 1.5 in three different mines. The reduction in the control deviation value in the next control cycle after model correction is used as the evaluation index, and the extension coefficient that best achieves the correction effect is selected. The pipeline analysis module receives the model correction signal and adjusts the correlation degree value in the correlation matrix and the calculation parameters of the resistance coefficient based on the model correction signal. The pipeline analysis module uses the pipeline operation status data at the time of deviation occurrence as an adjustment sample to adjust the correlation degree value in the correlation matrix. The correlation value is adjusted as follows: For nodes where the deviation value at the time of the deviation occurrence is greater than 1.5 times the average deviation value of that node, the correlation value between that node and the gas concentration change rate characteristic is increased by 0.05 (but not exceeding the upper limit of 1.0), and the correlation value between that node and the gas flow fluctuation amplitude characteristic is increased by 0.03 (but not exceeding the upper limit of 1.0). For nodes where the deviation value at the time of the deviation occurrence is less than 0.5 times the average deviation value of that node, the correlation value between that node and the gas concentration change rate characteristic is decreased by 0.03 (but not lower than the lower limit of zero), and the correlation value between that node and the gas flow fluctuation amplitude characteristic is decreased by 0.02 (but not lower than the lower limit of zero).

[0077] The adjustment amount of the correlation degree value in the correlation matrix is ​​positively correlated with the amplitude of the model correction signal: when the amplitude of the correction signal is less than 0.5, the adjustment amount is the above-mentioned baseline adjustment amount multiplied by 0.5; when the amplitude is greater than 0.5 and less than 1.0, the adjustment amount is the above-mentioned baseline adjustment amount multiplied by 1.0; when the amplitude is greater than 1.0, the adjustment amount is the above-mentioned baseline adjustment amount multiplied by 1.5. Simultaneously, the pipeline analysis module also adjusts the calculation parameters of the resistance coefficient based on the model correction signal. Specifically, the calculation parameters of the resistance coefficient refer to the window size of the sliding Lagrange interpolation and the order of the autoregressive integral moving average model (the number of autoregressive terms and the number of moving average terms).

[0078] The specific method for adjusting the window size is as follows: when the amplitude of the model correction signal is less than 0.5, the window size is adjusted from five time points before and after to six time points before and after; when the amplitude is greater than 0.5 but less than 1.0, the window size is adjusted to seven time points before and after; when the amplitude is greater than 1.0, the window size is adjusted to eight time points before and after. The window size should not exceed ten time points before and after, as an excessively large window will cause the interpolation results to become overly smooth and distorted. After adjusting the window size, the interpolated flow rate data is substituted into the resistance coefficient calculation formula, which is the negative pressure difference between the two ends of the pipe section divided by the square of the flow rate, thereby changing the value of the resistance coefficient.

[0079] The order adjustment method for the autoregressive integral moving average model is as follows: when the amplitude of the correction signal is greater than 0.5, the number of autoregressive terms and the number of moving average terms are each increased by one (but neither exceeds the upper limit of five), and the model parameters are re-estimated after adjustment. After the order adjustment, the change in model parameters is transmitted to the calculation of the resistance coefficient through the following chain: the adjusted model parameters are used to determine the correlation value; the change in the correlation value changes the classification result; the change in the classification result affects the formulation of control instructions; the change in control instructions affects the actual gas flow; and the change in the actual gas flow is ultimately reflected in the recalculation of the resistance coefficient. The early warning response module sets the following early warning thresholds: gas concentration threshold range, extraction negative pressure threshold range, pipeline temperature threshold range, carbon monoxide concentration safety threshold, cumulative pure gas quantity predetermined value, and normal operating parameters range of the extraction pump station. The gas concentration threshold range is set to 0.2% to 0.8%, and is taken as 80% according to the regulation in the coal mine safety regulations that the gas concentration in the return air flow of the mining face shall not exceed 1%. The carbon monoxide concentration safety threshold is set to 24 parts per million. The cumulative net gas volume is predetermined based on the gas reserves and extraction plan of the extraction area. When any real-time operating parameter is detected to meet the corresponding early warning conditions, the early warning mechanism is activated and emergency response measures are implemented.

[0080] The emergency response measures include: automatically stopping extraction when carbon monoxide concentration exceeds the standard or cumulative pure gas volume reaches a predetermined value; and diagnosing and prompting maintenance for abnormal operation of the extraction pump station. There is a linkage between the model correction signal issued by the effect evaluation module and the early warning response module: when the control deviation value exceeds a set threshold and the duration exceeds a set duration (60 minutes), the effect evaluation module sends an early warning trigger signal to the early warning response module. The 60-minute duration is based on the following: a control deviation lasting more than 60 minutes indicates that the system has been unable to control the gas concentration and flow rate to the expected values ​​after multiple control cycles, suggesting that the current model deviation or pipeline status change has exceeded the system's normal adjustment capacity.

[0081] Upon receiving an early warning trigger signal, the early warning response module lowers the trigger conditions for each early warning threshold according to the following rules: the reduction ratio of the threshold is calculated by multiplying the reduction coefficient by the relative amount of the deviation exceeding the threshold; when the reduction ratio exceeds 50% of the upper limit, the upper limit is used. The reduction coefficient is 0.2, meaning that for every doubling of the relative amount of the deviation exceeding the threshold, the reduction ratio increases by 20%. When the control deviation is at the threshold, the reduction ratio is 0%; when the control deviation is twice the threshold, the reduction ratio is 20%; and when the control deviation reaches 3.5 times the threshold, the reduction ratio reaches the upper limit of 50%. Lowering the trigger conditions expands the range of gas concentration early warning thresholds, triggering early warnings within a wider concentration range. When the control deviation returns to below the set threshold, the early warning response module restores the trigger conditions of the early warning threshold to the original set value. Example 2: Based on Example 1, the specific process by which the intelligent control module solves for the control target is as follows: The control target is the sum of the absolute values ​​of the deviations between the gas concentration of each pipeline node and the target gas concentration multiplied by a first weighting coefficient; the sum of the absolute values ​​of the deviations between the gas flow rate of each pipeline section and the target gas flow rate multiplied by a second weighting coefficient; and the real-time power of the extraction pump station multiplied by a third weighting coefficient. The minimum value obtained by adding these three values ​​is taken. The target gas concentration is set according to the coal mine safety regulations and extraction compliance requirements (for example, 35% for this mine). The target gas flow rate is determined based on the pre-extraction volume and the number of boreholes connected to the pipeline section. The decision variables are the valve opening degree and the operating power of the extraction pump station.

[0082] The constraints include: the valve opening degree must be within the range of 0 to 90 degrees; the operating power of the extraction pump station must be within the range of 20% to 100% of its rated power; and the deviation between the sum of the target extraction negative pressure values ​​of each branch and the actual extraction negative pressure value supplied by the extraction pump station must not exceed 5%. A particle swarm optimization algorithm is used to solve this optimization problem. The particle swarm size is 100 particles, the inertia weight decreases linearly from 0.9 to 0.4, the acceleration constant is 2.0, and the maximum number of iterations is 500. The optimization solution yields the optimal valve opening combination and the optimal operating power of the extraction pump station. Based on the optimal operating power, the target operating frequency of the extraction pump station is obtained by multiplying the ratio of this operating power to the rated power of the extraction pump station by the rated operating frequency. The target value of the liquid supply flow rate is calculated based on the linear relationship between the target operating frequency and the liquid supply flow rate. Then, control commands are generated based on the target operating frequency and the target liquid supply flow rate and sent to the frequency converters of the extraction pump station and the liquid supply pump.

[0083] In this embodiment, when the pipeline analysis module detects that the change in the real-time resistance coefficient of any pipe segment compared to the previous moment exceeds a set threshold, it initiates a recalculation of the entire pipeline network resistance coefficient. The recalculation process is as follows: traverse all pipe segments, re-collect the real-time extraction negative pressure difference and real-time gas flow rate at both ends of each pipe segment, recalculate the resistance coefficient of each pipe segment by dividing the negative pressure difference by the square of the flow rate, and then sum the resistance coefficients of all segments to obtain the recalculated total pipeline network resistance coefficient. Based on the recalculated resistance coefficients of each pipe segment, redistribute the target extraction negative pressure value of each branch: divide the resistance coefficient of the branch by the recalculated total pipeline network resistance coefficient, and then multiply by the actual supply negative pressure value of the extraction pump station to obtain the target extraction negative pressure value of the branch. The sum of the target extraction negative pressure values ​​of all branches after redistribution and the actual supply extraction negative pressure value of the extraction pump station achieve a dynamic balance of the negative pressure of the entire pipeline network.

[0084] Example 3: Based on Example 1, the specific process for calculating the control deviation value by the effect evaluation module is as follows: After the control command is executed, the actual gas concentration and actual gas flow rate of each pipeline node are continuously monitored with a sampling period of one minute. The control deviation value is calculated as follows: the difference between the actual concentration and the expected concentration of each node is squared and averaged; the difference between the actual flow rate and the expected flow rate of each pipeline segment is squared and averaged; the two averages are added together and the square root is taken as the control deviation value. When the control deviation value is small (e.g., 0.02), it indicates that the control effect is good; when the control deviation value is large (e.g., 0.15), it indicates that the control effect is not ideal. The method for determining the threshold is as follows: select thirty periods with good control effects in the history of the mine, calculate the control deviation value of each period, and take the 95th percentile as the threshold value. When the control deviation value exceeds the threshold value, the effect evaluation module sends a model correction signal, the signal amplitude of which is the value of the control deviation value exceeding the threshold value multiplied by a preset proportional coefficient. The effect evaluation module shows a negative correlation between the monitoring duration after the execution of the control command and the magnitude of the control deviation: when the control deviation is small, the monitoring duration is the standard duration of thirty minutes; when the control deviation exceeds the threshold, the monitoring duration is extended based on the standard duration by multiplying the result of the preset extension coefficient by the relative amount of the deviation exceeding the threshold.

[0085] The linkage mechanism between the effect evaluation module and the early warning response module is as follows: The effect evaluation module has an internal timer that starts timing when the control deviation exceeds a set threshold. If the control deviation exceeds the threshold for sixty minutes, the effect evaluation module sends an early warning trigger signal to the early warning response module. Upon receiving the signal, the early warning response module lowers the trigger conditions for each early warning threshold by a reduction ratio of 0.2 multiplied by the relative amount of the deviation exceeding the threshold, not exceeding fifty percent. When the control deviation returns to below the set threshold, the timer is reset to zero, and the early warning response module restores the trigger conditions for the early warning thresholds to their original settings.

[0086] In practical use, based on existing technology, the on-site engineering technicians first input the pipeline connection relationships, pipe section lengths, and pipe diameter parameters from the mine gas drainage design drawings into the system database. They then determine the target gas flow rate for each branch and the total target gas concentration for the mine based on the drainage design. Simultaneously, following system prompts, sensors are installed at each pipeline node. Gas concentration sensors, gas flow sensors, drainage negative pressure sensors, and pipeline temperature sensors are fixed to pre-drilled holes in the pipe wall. Valve positioners are installed above the electric valves of each branch. Power is connected, and signal debugging is completed. After confirming that the sensor readings are normal and the data transmission is error-free, the system enters standby mode.

[0087] After the system starts, operators set initial parameters through the system's human-machine interface, including the target gas concentration (set according to coal mine safety regulations and extraction compliance requirements), the target gas flow rate for each branch (determined based on the number of boreholes and the pre-extraction volume), the sampling period (one minute), and the pure water calibration instructions. After clicking the "Start Run" button, the system starts running automatically. Operators can view the gas concentration, gas flow rate, extraction negative pressure, pipeline temperature, valve opening data, and system-defined control instructions for each pipeline node in real time through the human-machine interface. Each operating parameter is displayed in the form of a trend chart.

[0088] For systems put into use for the first time or after major overhaul, operators must perform a "pure water condition calibration" once the gas extraction pump station has been running stably with pure water as the working fluid. The system automatically records the resistance coefficient of each pipe section under this condition as the benchmark value for subsequent threshold selection. During normal operation, the system automatically reminds operators to recalibrate the resistance characteristic curves of each branch every three months and the fluid supply characteristic curve of the extraction pump station every six months. During calibration, operators should follow the system prompts to adjust the valve opening or pump station operating frequency to the specified values. The system automatically records the data and completes curve fitting. After calibration, the system prompts "Calibration Complete" and automatically updates the fitting coefficients in the database. If the goodness of fit does not meet the requirements during calibration, the system interface displays "Insufficient goodness of fit, it is recommended to add calibration points." Operators should add calibration points according to the system prompts and recalibrate until the requirements are met.

[0089] During system operation, the human-machine interface uses different colors to indicate the operating status of each module: green indicates normal operation, yellow indicates parameters are out of tolerance but have not reached the warning level, and red indicates a warning has been triggered. When the warning response module triggers a warning, a warning prompt window pops up on the interface, displaying the warning type and the current value of the warning parameter, while the audible and visual alarm emits an audible and visual signal. If the warning is for excessive carbon monoxide concentration or accumulated pure gas reaching a predetermined value, the system automatically executes a stop extraction operation and displays "Emergency Stop Extraction in Progress" on the interface. The operator can only manually reset the system after confirming safety. If the warning is for abnormal operation of the extraction pump station, the interface displays specific fault diagnosis results and maintenance suggestions, which the operator then follows to perform repairs.

[0090] When the effect evaluation module sends an early warning trigger signal to the early warning response module, the operator can observe on the interface that the displayed value of the early warning threshold automatically decreases, and the interface flashes to indicate "The early warning threshold has been dynamically tightened". When the control deviation value returns to normal, the early warning threshold automatically returns to the original set value, the interface stops flashing and displays "The early warning threshold has been restored".

[0091] Throughout the gas extraction operation, the system operates continuously without manual intervention; operators only need to observe the displayed parameters and warning prompts. If significant changes occur in the pipeline network structure due to changes in the underground mining face, the addition or closure of new boreholes, the operator must manually enter the updated pipeline connections. The system will automatically reconstruct the pipeline topology and continue operation. When the gas extraction operation is complete, the operator clicks the "Stop" button, and the system automatically executes the shutdown procedure, stopping the extraction pumps and all equipment.

[0092] The above description is a further detailed explanation in conjunction with specific embodiments, and it should not be considered that the specific embodiments of the present invention are limited to this. For those skilled in the art to which this invention pertains and related fields, any extensions, operation methods, and data substitutions made based on the technical solution concept of this invention should fall within the protection scope of this invention.

Claims

1. An intelligent control system for coal mine gas drainage pipelines, characterized in that, include: The data acquisition module collects real-time operating parameters of the gas extraction pipeline network, including gas concentration, gas flow rate, extraction negative pressure, pipeline temperature, and valve opening at each pipeline node. The pipeline analysis module establishes a pipeline topology model, calculates the resistance coefficient of each pipe segment and the total resistance coefficient of the pipeline network, determines the correlation value between the deployment location of each pipeline node and the characteristics of operating parameters and constructs a correlation matrix, and classifies the real-time operating parameters based on the correlation matrix to obtain the classification results. The intelligent control module formulates control instructions based on the classification results. These instructions include the adjustment amount of each valve opening, the target extraction negative pressure value of each branch, and the target operating parameters of the extraction pump station. The module adjusts the operating status of the gas extraction pipeline network and jointly controls the operating parameters of the extraction pump station according to these instructions. The intelligent control module uses the minimization of the weighted sum of the absolute values ​​of the deviations between the gas concentration at each pipeline node and the target gas concentration, the absolute values ​​of the deviations between the gas flow rate at each pipeline segment and the target gas flow rate, and the real-time power of the extraction pump station as its control objective. The optimal control parameters are obtained by solving for this objective. The three weighting coefficients of the control objective automatically adjust according to the change in the total resistance coefficient of the pipeline network. The effect evaluation module continuously monitors the actual gas concentration and actual gas flow of each pipeline node after the control command is executed, and calculates the control deviation value by comprehensively calculating the deviation between the actual value and the expected value. When the control deviation value exceeds the set threshold, a model correction signal is issued. The pipeline analysis module receives the model correction signal and adjusts the correlation degree value in the correlation matrix and the calculation parameters of the resistance coefficient according to the model correction signal.

2. The intelligent control system for coal mine gas drainage pipelines according to claim 1, characterized in that, The pipeline analysis module establishes a pipeline topology model and calculates the resistance coefficient of each pipe segment and the total pipeline resistance coefficient, including: Construct a directed graph of the pipeline network topology based on the pipeline connection relationship, with pipeline intersections as nodes and each pipeline segment as an edge; for any edge, take the quotient of the extraction negative pressure difference between the two nodes of the edge and the square of the gas flow rate through the edge as the resistance coefficient of the edge, and add the resistance coefficients of all edges to obtain the total resistance coefficient of the pipeline network. The pipeline analysis module determines the correlation value between the deployment location and operating parameter characteristics of each pipeline node and constructs a correlation matrix, including: Determine the correlation value between the deployment location of each pipeline node and its operating parameter characteristics, and construct an association matrix with pipeline nodes as rows and operating parameter characteristics as columns. The value of each element in the association matrix represents the strength of the association between the corresponding pipeline node and the corresponding operating parameter characteristics. The pipeline network analysis module categorizes the real-time operating parameters based on the correlation matrix to obtain the categorization results, including: The real-time operating parameters are categorized based on the correlation matrix to obtain the categorization results.

3. The intelligent control system for coal mine gas drainage pipelines according to claim 1, characterized in that, The amplitude of the model correction signal is positively correlated with the magnitude of the control deviation value, and the adjustment amount of the correlation degree value in the correlation matrix is ​​positively correlated with the amplitude of the model correction signal; the monitoring duration of the effect evaluation module after executing the control command is negatively correlated with the magnitude of the control deviation value. When the change in the real-time resistance coefficient of any pipe segment compared to the previous moment exceeds a set threshold, the pipeline analysis module initiates a recalculation of the resistance coefficient of the entire pipeline network and redistributes the target extraction negative pressure value of each branch based on the recalculated resistance coefficient of each pipe segment. The target extraction negative pressure value of each branch after redistribution is allocated according to the proportion of the current resistance coefficient of each branch to the total resistance coefficient. The branch with the larger resistance coefficient will receive a higher target extraction negative pressure value.

4. The intelligent control system for coal mine gas drainage pipelines according to claim 1, characterized in that, When the intelligent control module jointly controls the operating parameters of the extraction pump station, there are linkage constraints between the control parameters: When the valve opening of a certain branch is adjusted, the target extraction negative pressure value of that branch is recalculated based on the current resistance coefficient of that branch. The valve opening adjustment amount and the target extraction negative pressure value adjustment amount satisfy the resistance characteristic curve constraint of that branch. When the target operating frequency of the extraction pump station is adjusted, the target value of the liquid supply flow rate of the extraction pump station is adjusted in conjunction with the fitting relationship between the target operating frequency and the liquid supply flow rate. The adjustment amount of the operating frequency and the adjustment amount of the liquid supply flow rate satisfy the liquid supply characteristic curve constraint of the extraction pump station. The deviation between the sum of the target extraction negative pressure values ​​after the redistribution of each branch and the actual supply extraction negative pressure value of the extraction pump station is used as a constraint condition in solving the control target.

5. The intelligent control system for coal mine gas drainage pipelines according to claim 1, characterized in that, When the intelligent control module solves for the control target, the control target is the sum of the absolute values ​​of the deviations between the gas concentration of each pipeline node and the target gas concentration multiplied by a first weighting coefficient, the sum of the absolute values ​​of the deviations between the gas flow rate of each pipeline section and the target gas flow rate multiplied by a second weighting coefficient, and the real-time power of the extraction pump station multiplied by a third weighting coefficient. The minimum value obtained by adding these three values ​​is taken. Solving for this minimum value yields the optimal valve opening combination and the optimal operating power of the extraction pump station. When the total resistance coefficient of the pipeline increases, the weighting coefficients of the gas flow rate deviation and the weighting coefficients of the extraction pump station power automatically increase.

6. The intelligent control system for coal mine gas drainage pipelines according to claim 1, characterized in that, The pipeline analysis module determines the correlation value between the deployment location and operating parameter characteristics of each pipeline node, including: The historical gas extraction pipeline monitoring data were preprocessed, and outliers were treated as missing values ​​and filled using sliding Lagrange interpolation. For each pipeline node, an autoregressive integral moving average model is constructed to analyze the time series characteristics of historical gas concentration data and historical gas flow data. The autoregressive integral moving average model uses a linear combination of historical observation values ​​and historical prediction errors to describe the current value. The sequential learning algorithm is applied to adjust the model parameters of the autoregressive integral moving average model based on the rate of change of gas concentration, the fluctuation range of gas flow, the rate of change of extraction negative pressure, the rate of change of pipeline temperature, and the valve opening response characteristics. The correlation value between each pipeline node and each operating parameter characteristic is determined based on the adjusted autoregressive integral moving average model and filled into the correlation matrix.

7. The intelligent control system for coal mine gas drainage pipelines according to claim 6, characterized in that, The pipeline network analysis module categorizes the real-time operating parameters based on the correlation matrix, including: For each pipeline node, the corresponding operating parameter characteristics of the pipeline node in the correlation matrix are used as transactions to construct a frequent pattern tree. The nodes of the frequent pattern tree represent operating parameter characteristics, and the node weights represent the frequency of the operating parameter characteristics in the transactions. By mining common combinations of operating parameter characteristics among multiple pipeline nodes through the frequent pattern tree, pipeline nodes with the same combination of operating parameter characteristics are assigned the same operating category identifier. The real-time operating parameters of each pipeline node are compared with the extracted operating parameter characteristics, and the real-time operating parameters are classified according to the comparison results.

8. The intelligent control system for coal mine gas drainage pipelines according to claim 1, characterized in that, The intelligent control module jointly controls the operating parameters of the extraction pump station, including: The target operating frequency of the pumping station is obtained by multiplying the ratio of the target operating power of the pumping station to the rated power of the pumping station in the control command by the rated operating frequency of the pumping station. The target value of the liquid supply flow rate is calculated based on the linear relationship between the target operating frequency of the extraction pump station and the liquid supply flow rate. The regulating device generates control commands based on the target operating frequency of the extraction pump station and the target value of the liquid supply flow rate, and sends them to the extraction pump station.

9. The intelligent control system for coal mine gas drainage pipelines according to claim 1, characterized in that, It also includes an early warning response module, which sets the following early warning thresholds: gas concentration threshold range, extraction negative pressure threshold range, pipeline temperature threshold range, carbon monoxide concentration safety threshold, cumulative pure gas quantity predetermined value, and normal operating parameters of the extraction pump station; when any real-time operating parameter is detected to meet the corresponding early warning condition, the early warning mechanism is activated and emergency response measures are executed, including: automatically stopping extraction operation when the carbon monoxide concentration exceeds the standard or the cumulative pure gas quantity reaches the predetermined value, and performing fault diagnosis and prompting maintenance when the extraction pump station is not operating normally; There is a linkage between the model correction signal issued by the effect evaluation module and the early warning response module: when the control deviation value exceeds the set threshold and the duration exceeds the set duration, the effect evaluation module sends an early warning trigger signal to the early warning response module, and the early warning response module lowers the trigger condition of the early warning threshold.

10. A method for intelligent control of coal mine gas drainage pipeline network, applied to the intelligent control system of coal mine gas drainage pipeline network as described in any one of claims 1 to 9, characterized in that, Includes the following steps: S1, collect real-time operating parameters of the gas extraction pipeline network, including gas concentration, gas flow rate, extraction negative pressure, pipeline temperature and valve opening at each pipeline node; S2. Construct a directed graph of the pipeline network topology based on the pipeline connection relationship, with the pipeline intersection points as nodes and each pipeline segment as an edge; for any edge, take the quotient of the extraction negative pressure difference between the two nodes of the edge and the square of the gas flow rate through the edge as the resistance coefficient of the edge, and add the resistance coefficients of all edges to obtain the total resistance coefficient of the pipeline network. S3, determine the correlation value between the deployment location of each pipeline node and the operating parameter characteristics, and construct an association matrix with pipeline nodes as rows and operating parameter characteristics as columns. The value of each element in the association matrix represents the strength of the association between the corresponding pipeline node and the corresponding operating parameter characteristics. The real-time operating parameters are categorized based on the correlation matrix to obtain the categorization results. S4. Based on the classification results, control instructions are formulated, including the adjustment amount of each valve opening, the target extraction negative pressure value of each branch, and the target operating parameters of the extraction pump station. The operating status of the gas extraction pipeline network and the operating parameters of the extraction pump station are adjusted according to the control instructions. The optimal control parameters are solved by minimizing the weighted sum of the absolute values ​​of the deviations between the gas concentration of each pipeline node and the target gas concentration, the absolute values ​​of the deviations between the gas flow rate of each pipeline section and the target gas flow rate, and the real-time power of the extraction pump station. The three weighting coefficients of the control objective are automatically adjusted according to the change of the total resistance coefficient of the pipeline network. S5. After the control command is executed, the actual gas concentration and actual gas flow of each pipeline node are continuously monitored. The average of the square of the deviation between the actual gas concentration and the expected gas concentration of each pipeline node and the average of the square of the deviation between the actual gas flow and the expected gas flow of each pipeline segment are added together and the square root is taken as the control deviation value. When the control deviation value exceeds the set threshold, a model correction signal is issued. S6, adjust the correlation degree value in the correlation matrix and the calculation parameters of the resistance coefficient according to the model correction signal; S7. When the real-time operating parameters meet the early warning conditions, the early warning mechanism is activated and emergency response measures are executed.

Images