Gas extraction pipe network anomaly library construction and multi-dimensional quantitative diagnosis method

Abstract

This invention belongs to the field of mine gas drainage safety monitoring technology. It discloses a method for constructing an anomaly database and multi-dimensional quantitative diagnosis of gas drainage pipeline networks, aiming to solve the problems of traditional diagnosis relying on experience, lacking quantitative standards, insufficient precise positioning capabilities, and low automation levels. By collecting and preprocessing multi-source monitoring data from key nodes of the pipeline network, a multi-dimensional anomaly feature database is constructed by extracting feature tags, establishing a system-level, comprehensive diagnostic indicator system. A three-level progressive diagnostic model is designed, and through quantitative calculation and feature matching, it achieves overall anomaly perception of the pipeline network, fine differentiation of four types of anomalies, precise delineation of abnormal pipe sections, and severity classification, forming a closed-loop diagnostic process. This invention overcomes the limitations of experience-based diagnosis, achieves quantitative collaborative diagnosis, improves the standardization, automation, and accuracy of diagnosis, and provides precise basis for on-site fault handling. It is applicable to various types of mine gas drainage pipeline networks, effectively ensuring the stable operation of the drainage system and mine production safety.

Description

Technical Field

[0001] This invention relates to the field of mine gas drainage safety monitoring technology, and in particular to a method for constructing an anomaly database of gas drainage pipelines and a multi-dimensional quantitative diagnosis method. Background Technology

[0002] Mine gas drainage is a core technical means for the prevention and control of coal mine gas disasters. As the core fluid transmission carrier of the drainage system, the gas drainage pipeline network is a key link in ensuring the stable operation of the gas drainage system and preventing underground gas accumulation accidents by real-time sensing, accurate diagnosis and rapid handling of typical abnormal conditions such as main pipeline blockage, branch pipeline blockage, main pipeline leakage and branch pipeline leakage.

[0003] Currently, the diagnosis of anomalies in underground gas drainage pipelines in coal mines largely relies on the industry experience of on-site maintenance personnel, combined with manual judgment based on single-point pressure and flow velocity monitoring data collected by conventional sensors. A systematic and quantitative diagnostic system has not yet been established. Furthermore, diagnostic methods generally depend on experience-based feature matching, failing to extract specific characterization labels for the essential features of various abnormal operating conditions in the pipeline network. A multi-dimensional anomaly feature library encompassing negative pressure response, velocity response, physical mechanisms, and identification features has not been constructed. A comprehensive diagnostic indicator system coupling parameter response and spatial distribution characteristics is lacking. This makes it impossible to transform the qualitative characteristics of anomalies into quantitative judgment logic containing variable relationships, threshold boundaries, and spatial correlations, resulting in highly subjective and inconsistent diagnostic results, making it difficult to establish unified diagnostic standards.

[0004] Meanwhile, traditional diagnostic methods lack precise localization capabilities, making it difficult to achieve hierarchical diagnosis from overall system anomaly perception to anomaly type differentiation, and then to precise delineation of specific pipe sections. Furthermore, existing diagnostic methods largely rely on single sensor data for judgment, failing to achieve complementarity and cross-validation of multi-source parameters, making them susceptible to misjudgments due to sensor noise, data distortion, and other factors. In addition, traditional diagnostic methods do not quantify and classify the severity of anomalies, making it impossible for on-site maintenance personnel to conduct tiered responses based on anomaly levels. This can easily lead to untimely or excessive responses, impacting the operational efficiency of the gas extraction system and the economic efficiency of fault handling.

[0005] Therefore, there is an urgent need to design a method for constructing an anomaly database and multi-dimensional quantitative diagnosis of gas drainage pipeline networks, so as to achieve fine differentiation of anomaly types, accurate delineation of anomaly range, and quantitative classification of severity of anomalies in gas drainage pipeline networks, thereby improving the reliability and automation level of anomaly diagnosis and making it applicable to the safety monitoring and fault handling of gas drainage pipeline networks in various mines. Summary of the Invention

[0006] To address the aforementioned technical problems, this invention provides a method for constructing an anomaly database and performing multi-dimensional quantitative diagnosis of gas drainage pipeline networks. This method enables precise differentiation of anomaly types, accurate delineation of anomaly ranges, and quantitative classification of severity, improving the reliability, standardization, and automation of anomaly diagnosis. It provides precise technical support for safety monitoring and fault handling of mine gas drainage pipeline networks, ensuring the stable operation of the gas drainage system.

[0007] The technical solution adopted by this invention to solve its technical problem is: a method for constructing an anomaly database and performing multi-dimensional quantitative diagnosis of gas extraction pipeline networks, comprising the following steps: S1: Collect multi-source monitoring data of key nodes in the gas extraction pipeline network. The multi-source monitoring data includes real-time monitoring data and historical monitoring data. Based on the multi-source monitoring data, establish a multi-source monitoring data sample library for the pipeline network. S2: Preprocess the multi-source monitoring data in the sample library to complete data quality control, missing value verification, outlier labeling and normalization, and determine the stable range and dynamic change characteristics of each parameter; S3: Collect multi-source real-time monitoring data of key nodes in the gas extraction pipeline network as the real-time input data source for the diagnostic model; S4: Based on the preprocessed multi-source monitoring data from the sample library, calibrate the parameter benchmarks, fluctuation thresholds, feature correlation patterns, and judgment systems under normal operating conditions, and construct a three-level progressive diagnostic model; input the real-time multi-source monitoring data collected in step S3 into the three-level progressive diagnostic model, and perform feature extraction, quantitative calculation, and trend analysis on the real-time multi-source monitoring data through the three-level progressive diagnostic model, and perform multi-dimensional comparison and matching with the normal operating condition benchmarks and abnormal feature patterns in the sample library to perform three-level progressive diagnosis; S5: Outputs diagnostic results including the anomaly types, pipeline ranges, and severity levels of the entire gas extraction pipeline network, forming a complete closed-loop process of multi-source data acquisition and database construction, data preprocessing, real-time data acquisition, model diagnosis, and result output.

[0008] Furthermore, in step S4, the three-level progressive diagnosis specifically includes: Level 1: Extract the overall monitoring values ​​of pressure and flow velocity from all measuring points in the pipeline network, calculate the mean, rate of change, and overall fluctuation range of pressure / flow velocity at all measuring points, compare the calculation results with the stable pressure and flow velocity range and parameter fluctuation threshold under normal operating conditions in the sample library, and comprehensively determine whether the pipeline network operating parameters exceed the normal fluctuation range or deviate from the normal operating conditions based on whether the overall pressure rise and fall exceeds the threshold, whether the overall trend of flow velocity changes deviates from the normal pattern, and whether the fluctuation of parameters at multiple measuring points shows consistent abnormality. Level 2: For real-time data that is determined to deviate from normal operating conditions, calculate the coordinated changes in pressure and flow velocity at each measuring point, spatial distribution characteristics, and correlation degree of main and branch parameters. Compare and match the calculated characteristic patterns with the typical characteristic spectra of four types of anomalies in the sample library: main blockage, branch blockage, main leakage, and branch leakage. Based on the characteristic matching degree, complete the fine distinction of anomaly types. Combined with the spatial distribution pattern of parameter anomalies, preliminarily determine the main / branch to which the anomaly belongs and the corresponding pipeline section. Level 3: For the initially identified abnormal pipeline sections, calculate the parameter sensitivity, pipeline anomaly correlation, and multi-parameter change consistency coefficient of each measuring point within the section. Compare the calculation results with the sensitivity threshold and correlation grading standards calibrated in the sample library. Using the measuring point with the highest anomaly correlation as the core, identify and delineate the specific pipeline section range where the anomaly is located by combining the spatial correlation of the measuring points. Based on the quantitative results of sensitivity and correlation, complete the graded assessment of the severity of the anomaly.

[0009] Furthermore, the multi-source monitoring data includes extraction negative pressure, gas mixing flow rate, methane concentration, oxygen concentration, carbon monoxide concentration, carbon dioxide concentration, pump station power, and pipeline temperature, which are used to comprehensively characterize the operating status and safety situation of key nodes in the pipeline network.

[0010] Furthermore, in step S2, the preprocessing of multi-source monitoring data in the sample library specifically includes: labeling and tracing missing, distorted, or sensor-failed data; normalizing the data to the [0,1] interval using the Min-Max standardization formula; defining the parameter stability interval based on historical normal operating conditions; and marking data exceeding the threshold as potential anomalies. The standardized calculation formula is as follows: In the formula, and These are the upper and lower limits of the threshold values ​​for each parameter, obtained based on statistical data from historical normal operating conditions.

[0011] Furthermore, in the second-level diagnosis, the synergistic relationship between pressure and flow velocity is calculated and compared with the four types of abnormal feature spectra in the sample library to distinguish between blockage and leakage anomalies; based on the spatial distribution pattern of parameter anomalies and the matching of the main and branch characteristic patterns calibrated in the sample library, it is preliminarily determined whether the anomaly is located in the main or branch road. The specific mathematical determination formula is defined as follows: Regarding the blockage of a branch road, this situation occurs if the following conditions are met: Regarding main road congestion, if the following conditions are met: Regarding branch circuit leakage, the following conditions must be met: Regarding branch circuit leakage, the following conditions must be met: In the formula, For a set of branches, each branch measurement point Each has a corresponding associated main road measuring point. ; The branch index variable represents the set. Zhongde Di A side road; For set A specific branch in the middle, determined by the existential quantifier Definition: A target branch road suspected of being blocked; branch road The change in negative pressure, relative to normal operating conditions, is expressed as follows: a negative value indicates an increase in the absolute value of negative pressure, and a positive value indicates a decrease in the absolute value of negative pressure. This represents the change in negative pressure at the main road measuring point corresponding to branch i. branch road The change in flow velocity, a positive value indicates an increase in velocity, and a negative value indicates a decrease, and so on. This represents the change in flow velocity at the main road measuring point corresponding to branch i. To retrieve a set The maximum value of the corresponding parameter for all branches; , Representing branches The collection of upstream and downstream branches.

[0012] Furthermore, in the third-level diagnosis, the correlation between the negative pressure sensitivity of the measuring point and the pipeline abnormality is calculated using the following formula: The severity of the anomaly is graded based on the correlation between pipeline anomalies, and the grading formula is as follows: In the formula: The negative pressure sensitivity at the i-th measuring point; This represents the real-time negative pressure monitoring value at the i-th measuring point; The historical negative pressure reference value under normal operating conditions for the i-th measuring point; Let be the degree of abnormal correlation of the j-th pipeline segment; Let j be the set of monitoring points corresponding to the j-th pipeline segment; Weighting of the spatial influence of the measuring points on the pipeline section; This represents the severity level of the abnormality. and The grading thresholds determined by historical operating condition statistics must meet the following conditions. ; , to The degree of abnormality represents an increasing trend.

[0013] Furthermore, the third-level diagnostic process integrates multi-source parameter complementarity and cross-validation. Collaborative discrimination is achieved by calculating the consistency coefficient of multi-parameter changes through a model, and the calculation results are compared with the consistency threshold calibrated in the sample library. This reduces the tendency for misjudgment caused by noise from a single sensor and maintains the robustness of the diagnostic process. The consistency coefficient of multi-parameter changes is calculated using the following formula: In the formula: The consistency coefficient for multiple parameter variations; The number of parameter types involved in the diagnosis; This represents the real-time change of the k-th type of parameter; The steady-state reference value is calibrated in the parameter sample library of type k.

[0014] Furthermore, the three-level progressive diagnostic model has adaptive expansion capabilities, and can iteratively optimize indicator thresholds and discrimination logic based on pipeline topology changes and historical diagnostic feedback, combined with newly added effective diagnostic cases in the sample library. The adaptive iterative correction formula for the threshold is: ; In the formula: The diagnostic threshold is updated iteratively. The original diagnostic threshold before iteration; The learning coefficient; The observation threshold is obtained by statistically analyzing historical valid diagnostic cases in the sample database.

[0015] Furthermore, in step S5, the output diagnostic results are geared towards engineering operation and maintenance design. Standardized results are generated after quantitative calculation by the model, providing direct evidence for on-site fault handling. The specific expression of the standardized results is as follows: ; In the formula, For comprehensive diagnostic output; This refers to the pipeline network operation status determination result obtained from the first level of the three-level progressive diagnosis, which includes both normal and abnormal states. If the result is normal, no further results will be output. This is a detailed differentiation result of the abnormality type obtained from the second level of diagnosis in a three-level progressive diagnosis; The result of identifying and delineating the abnormal pipeline range obtained from the third level of diagnosis in the three-level progressive diagnosis; This represents the severity grading result of the abnormality obtained from the third-level diagnosis in the third-level progressive diagnosis.

[0016] The beneficial effects of this invention are: 1. This invention constructs a multi-dimensional anomaly feature library and a system-level parameter response-spatial distribution feature coupling global diagnostic index system. It transforms the qualitative features of pipeline anomalies into quantitative judgment logic containing variable relationships, threshold boundaries, and spatial correlations, and provides matching judgment formulas and quantitative calculation indicators. This changes the traditional diagnostic mode that relies on manual experience matching, realizes the standardization and normalization of anomaly diagnosis, and effectively avoids the subjectivity of manual diagnosis and the inconsistency of judgment results.

[0017] 2. This invention follows the mechanism-driven principle and integrates the fluid dynamics characteristics and topology of the pipeline network. It designs a three-level progressive diagnostic model that progresses from overall anomaly perception and fine differentiation of anomaly types to precise delineation of specific pipe sections and severity classification. This achieves full-dimensional diagnosis and solves the technical problem that traditional diagnosis can only determine the existence of anomalies but is difficult to distinguish types and accurately locate them, thus improving the accuracy and pertinence of anomaly diagnosis.

[0018] 3. This invention collects multi-source monitoring data such as negative pressure, flow rate, and gas concentration. During the diagnosis process, it integrates the complementarity and cross-validation of multi-source parameters and achieves collaborative discrimination by calculating the consistency coefficient of multi-parameter changes, which effectively reduces the tendency of misjudgment. At the same time, it performs standardized preprocessing and quality control on the monitoring data to further ensure the reliability of the diagnostic results.

[0019] 4. The three-level progressive diagnostic model of this invention can optimize the index thresholds and discrimination logic by adaptively iteratively correcting the formula based on changes in pipeline topology, historical diagnostic feedback, and newly added valid cases in the sample library. It can adapt to the diagnostic needs of gas drainage pipelines in different mines and under different working conditions without major reconstruction of the model. It solves the problem that traditional diagnostic criteria are fixed and cannot adapt to dynamic changes in pipelines. It is applicable to the safety monitoring of gas drainage pipelines in various mines and has outstanding versatility and practicality.

[0020] 5. This invention constructs a complete closed-loop process of multi-source data acquisition and database construction, data preprocessing, real-time data acquisition, model diagnosis, and result output. The entire process is completed by the model, including data calculation, feature matching, anomaly judgment, and result output. No manual intervention is required in the core diagnostic process, which improves the automation level of anomaly diagnosis in gas drainage pipeline networks and meets the development needs of intelligent and digital safe production in modern coal mines.

[0021] 6. This invention outputs standardized diagnostic results including anomaly type, specific pipe section range, and severity level, directly clarifying the core information for fault handling. At the same time, it classifies anomalies into three levels of severity, facilitating on-site maintenance personnel to carry out graded and targeted handling, shortening the fault investigation and handling time, reducing the manpower and time costs of pipeline network operation and maintenance, and improving the efficiency and economy of gas extraction pipeline network fault handling.

[0022] 7. This invention enables timely detection of potential pipeline operation hazards and guidance for on-site handling through real-time, accurate, and rapid diagnosis of four typical anomalies. It effectively avoids problems such as decreased gas extraction efficiency, gas accumulation, and pipeline failure caused by pipeline anomalies, ensuring the safety of underground workers and the order of mine production. It provides important technical support for the stable and efficient operation of mine gas extraction systems. Attached Figure Description

[0023] Figure 1 This is a schematic diagram of the preliminary identification process for abnormal types in the second-level diagnosis of this invention.

[0024] Figure 2 This is a schematic diagram of the process for determining the location of abnormal pipelines in the third-level diagnosis of this invention. Detailed Implementation

[0025] The present invention will be further described in detail below with reference to the accompanying drawings.

[0026] This invention discloses a method for constructing an anomaly database and performing multi-dimensional quantitative diagnosis of gas extraction pipeline networks.

[0027] This method for constructing an anomaly database and developing a multi-dimensional quantitative diagnosis method for gas drainage pipeline networks is based on a hierarchical diagnostic system built through multi-parameter collaborative sensing. It constructs a multi-dimensional anomaly feature database encompassing negative pressure response, velocity response, physical mechanisms, and identification characteristics. Furthermore, it establishes a comprehensive diagnostic index system coupling system-level parameter response and spatial distribution characteristics to achieve quantitative diagnosis of anomalies in gas drainage pipeline networks. The specific steps include: S1: Collect multi-source monitoring data of key nodes in the gas drainage pipeline network. The multi-source monitoring data includes real-time monitoring data and historical monitoring data, and establish a multi-source monitoring data sample library of the pipeline network based on the multi-source monitoring data. Among them, the key nodes of the gas drainage pipeline network include the main pipeline junction point, the connection node between the branch and the main pipeline, the pipeline diameter change section, the long-distance measuring point, and the inlet and outlet of the pump station. The multi-source monitoring data includes drainage negative pressure, gas mixing flow rate, methane concentration, oxygen concentration, carbon monoxide concentration, carbon dioxide concentration, pump station power, and pipeline temperature, which are used to comprehensively characterize the operating status and safety status of the key nodes of the pipeline network.

[0028] S2: Preprocess the multi-source monitoring data in the sample library, complete data quality control, missing value verification, outlier labeling and normalization, and determine the stable range and dynamic change characteristics of each parameter.

[0029] The preprocessing of multi-source monitoring data specifically includes: labeling and tracing missing, distorted, or sensor-failed data; normalizing the data to the [0,1] interval using the Min-Max standardization formula; defining the parameter stability interval based on historical normal operating conditions and marking data exceeding the threshold as potential anomalies; extracting the instantaneous changes and sliding time window changes of each parameter to form dynamic change characteristics, which serve as stable inputs for the subsequent construction of a three-level progressive diagnostic model.

[0030] The standardized calculation formula is as follows: In the formula, and These are the upper and lower limits of the threshold values ​​for each parameter, obtained based on statistical data from historical normal operating conditions.

[0031] S3: Collect multi-source real-time monitoring data of key nodes in the gas extraction pipeline network as the real-time input data source for the diagnostic model.

[0032] S4: Based on the preprocessed multi-source monitoring data from the sample library, calibrate the parameter benchmarks, fluctuation thresholds, feature correlation patterns, and judgment systems under normal operating conditions, and construct a three-level progressive diagnostic model; input the real-time multi-source monitoring data collected in step S3 into the three-level progressive diagnostic model, and perform feature extraction, quantitative calculation, and trend analysis on the real-time multi-source monitoring data through the three-level progressive diagnostic model, and perform multi-dimensional comparison and matching with the normal operating condition benchmarks and abnormal feature patterns in the sample library to execute the three-level progressive diagnosis: The three-level progressive diagnosis follows a mechanism-driven principle, integrating the fluid dynamics characteristics and topology of the pipeline network. Each level of diagnosis executes a logical process of real-time data quantification calculation, comparison with the sample library benchmark / features, and quantification judgment, realizing multi-layered progressive judgment from anomaly perception and fine-grained type differentiation to location identification and delineation. Specifically: Level 1: Extract the overall monitoring values ​​of pressure and flow velocity from all measuring points in the pipeline network, calculate the average pressure / flow velocity, rate of change, and overall fluctuation range of all measuring points, compare the calculation results with the pressure and flow velocity stability range and parameter fluctuation thresholds under normal operating conditions in the sample library, and comprehensively determine whether the pipeline network operating parameters exceed the normal fluctuation range or deviate from the normal operating conditions based on whether the overall pressure rise and fall exceeds the threshold, whether the overall flow velocity change trend deviates from the normal pattern, and whether the fluctuation of parameters at multiple measuring points shows consistent abnormalities.

[0033] Level 2: For real-time data determined to deviate from normal operating conditions, calculate the coordinated changes in pressure and flow velocity at each measuring point, spatial distribution characteristics, and correlation degree of main and branch parameters. Compare and match the calculated characteristic patterns with the typical characteristic spectra of four types of abnormal operating conditions in the sample library: main blockage, branch blockage, main leakage, and branch leakage, as shown in Table 1. Based on the characteristic matching degree, complete the fine distinction of abnormal types, and combine the spatial distribution pattern of parameter abnormalities to preliminarily determine the main / branch to which the abnormality belongs and the corresponding pipeline section, providing a basis for the precise location of the third-level diagnosis.

[0034] Table 1. Typical Characteristics of Four Types of Abnormal Operating Conditions in Gas Drainage Pipeline Networks The specific mathematical formula for main / branch anomaly detection is as follows: Regarding the blockage of a branch road, this situation occurs if the following conditions are met: Regarding main road congestion, if the following conditions are met: Regarding branch circuit leakage, the following conditions must be met: Regarding branch circuit leakage, the following conditions must be met: In the formula, For a set of branches, each branch measurement point Each has a corresponding associated main road measuring point. ; The branch index variable represents the set. Zhongde Di A side road; For set A specific branch in the middle, determined by the existential quantifier Definition: A target branch road suspected of being blocked; branch road The change in negative pressure, relative to normal operating conditions, is expressed as follows: a negative value indicates an increase in the absolute value of negative pressure, and a positive value indicates a decrease in the absolute value of negative pressure. This represents the change in negative pressure at the main road measuring point corresponding to branch i. branch road The change in flow velocity, a positive value indicates an increase in velocity, and a negative value indicates a decrease, and so on. This represents the change in flow velocity at the main road measuring point corresponding to branch i. To retrieve a set The maximum value of the corresponding parameter for all branches; , Representing branches The collection of upstream and downstream branches.

[0035] Reference Figure 1 The specific judgment process for pipeline anomaly detection in this second level is as follows: Based on the total negative pressure of the pipeline network monitored by the pumping station as the core criterion, the situation is divided into three categories according to the trend of negative pressure change: Scenario 1: Pump station negative pressure drop: It is determined that there may be a leak in the pipeline network. The preliminary verification process for leak-related anomalies is initiated to check whether the pump station flow data has increased abnormally at the same time. If the flow rate does not increase synchronously, the pump station equipment should be checked for malfunctions; if the flow rate increases synchronously and abnormally, it is determined that there is a leak in the pipeline, and the location of the leak should be determined by further combining the main road flow velocity data; if the main road flow velocity increases abnormally, it is determined that there is a leak in the main road; otherwise, it is determined that there is a leak in the branch road.

[0036] Scenario 2: If there is no significant change in the negative pressure of the pump station, the pipeline network is determined to be operating normally, and the preliminary diagnosis process ends.

[0037] Scenario 3: The negative pressure of the pump station rises, indicating that there may be a blockage in the pipeline network. The preliminary verification process for blockage-related anomalies is initiated to check whether the power consumption data of the pump station increases synchronously. If power consumption does not increase synchronously, pump station equipment malfunctions need to be investigated; if power consumption increases synchronously, it is determined that there is a blockage in the pipeline network, and the location of the blockage should be determined by combining the negative pressure data of each branch; if only some branches have increased negative pressure, it is determined that there is a blockage in the main pipeline; if all branches have increased negative pressure synchronously, it is determined that there is a blockage in the branch pipeline.

[0038] Level 3: For the initially identified abnormal pipeline sections, calculate the parameter sensitivity, pipeline anomaly correlation, and multi-parameter change consistency coefficient of each measuring point within the section. Compare the calculation results with the sensitivity threshold and correlation grading standards calibrated in the sample library. Using the measuring point with the highest anomaly correlation as the core, identify and delineate the specific pipeline section range where the anomaly is located by combining the spatial correlation of the measuring points. Based on the quantitative results of sensitivity and correlation, complete the graded assessment of the severity of the anomaly.

[0039] The sensitivity of the measuring point to negative pressure and the correlation of pipeline abnormalities are calculated using the following formula: The severity of the anomaly is graded based on the correlation between pipeline anomalies, and the grading formula is as follows: In the formula: The negative pressure sensitivity at the i-th measuring point; This represents the real-time negative pressure monitoring value at the i-th measuring point; The historical negative pressure reference value under normal operating conditions for the i-th measuring point; Let be the degree of abnormal correlation of the j-th pipeline segment; Let j be the set of monitoring points corresponding to the j-th pipeline segment; Weighting of the spatial influence of the measuring points on the pipeline section; This represents the severity level of the abnormality. and The grading thresholds determined by historical operating condition statistics must meet the following conditions. ; , to The degree of abnormality increases, that is... This is a general abnormality. This is a relatively severe abnormality. This is a serious abnormality.

[0040] In the third-level diagnosis process, multi-source parameter complementarity and cross-validation are integrated. The consistency coefficient of multi-parameter changes is calculated by the model to achieve collaborative discrimination. The calculation results are compared with the consistency threshold calibrated in the sample library to reduce the tendency of misjudgment caused by single sensor noise and maintain the robustness of the diagnosis process. The consistency coefficient of multi-parameter variation is calculated using the following formula: In the formula: The consistency coefficient for multiple parameter variations; The number of parameter types involved in the diagnosis; This represents the real-time change of the k-th type of parameter; The steady-state reference value is calibrated in the parameter sample library of type k.

[0041] In addition, the three-level progressive diagnostic model has adaptive expansion capabilities, and can iteratively optimize indicator thresholds and discrimination logic based on changes in pipeline topology and historical diagnostic feedback, combined with newly added effective diagnostic cases in the sample library. The adaptive iterative correction formula for the threshold is: ; In the formula: The diagnostic threshold is updated iteratively. The original diagnostic threshold before iteration; The learning coefficient; The observation threshold is obtained by statistically analyzing historical valid diagnostic cases in the sample database.

[0042] Reference Figure 2 The specific judgment process for accurately locating abnormal pipelines in complex pipe networks during the third-level diagnosis is as follows: Starting with the second-level diagnostic determination of a leak, a more refined assessment is conducted based on the characteristics of changes in flow velocity, negative pressure, and oxygen concentration in the main and branch pipelines. If the flow velocity in all branches decreases, it is determined that the main pipeline is leaking. Further comparison is made between the flow velocity and negative pressure changes of adjacent sensors in the main pipeline. If the changes are inconsistent, it is determined that there is a leak in the interval between the two sensors. If the changes are consistent, it is transferred to the overall judgment process for blockage-related anomalies. If the flow velocity in some branches does not decrease, proceed to branch leak investigation. Traverse all branches to determine whether the flow velocity or oxygen concentration of the i-th branch increases. If it increases, check the flow velocity data near the borehole end of the branch. If the flow velocity decreases, it is determined that there is a leak point in the branch. If the flow velocity does not decrease, it is determined that the borehole sealing has failed. If neither the branch flow velocity nor the oxygen concentration increases, they are all incorporated into the overall judgment process for blockage anomalies. After entering the overall judgment process for blockage, it is first verified whether the negative pressure of all branches in the entire pipeline network has increased. If not all branches have increased negative pressure, it is determined that there is a blockage in the main pipeline. The changes in negative pressure of adjacent sensors in the main pipeline are compared. If the changes are inconsistent, it is determined that there is a blockage point in the interval between the two sensors. If the changes are consistent, it is verified in conjunction with the oxygen concentration. An abnormal rise in concentration indicates a blockage and leak in that section; no abnormality indicates the pipeline is normal. If the negative pressure in all branches increases synchronously, then branch blockage detection is initiated. The branches are traversed to verify the characteristic of decreased flow velocity in the i-th branch and increased flow velocity in adjacent branches. If this characteristic is not met, the branch is directly determined to be blocked. If the characteristic is met, the oxygen concentration is checked again. An abnormal rise in concentration indicates an additional risk of leakage in that section. The flow velocity data near the borehole end of the branch needs to be further checked. If no abnormality is found, the branch is determined to be in normal operating condition, and the false blockage judgment is eliminated.

[0043] S5: Outputs diagnostic results encompassing the anomaly types, pipeline ranges, and severity levels across the entire gas extraction pipeline network, forming a complete closed-loop process of multi-source data acquisition and database construction, data preprocessing, real-time data acquisition, model diagnosis, and result output. The output diagnostic results are geared towards engineering operation and maintenance design. Standardized results are generated through model quantification calculations, providing direct evidence for on-site fault handling. The specific expression for the standardized results is as follows: ; In the formula, For comprehensive diagnostic output; This refers to the pipeline network operation status determination result obtained from the first level of the three-level progressive diagnosis, which includes both normal and abnormal states. If the result is normal, no further results will be output. This is a detailed differentiation result of the abnormality type obtained from the second level of diagnosis in a three-level progressive diagnosis; The result of identifying and delineating the abnormal pipeline range obtained from the third level of diagnosis in the three-level progressive diagnosis; This represents the severity grading result of the abnormality obtained from the third-level diagnosis in the third-level progressive diagnosis.

[0044] The multidimensional anomaly feature library is constructed based on on-site monitoring data of four typical abnormal operating conditions: main road blockage, branch road blockage, main road leakage, and branch road leakage. It extracts unique feature labels that characterize each type of anomaly. The full-domain diagnostic indicator system transforms qualitative features into quantitative judgment logic containing variable relationships, threshold boundaries, and spatial correlations. It achieves anomaly diagnosis through the collaborative analysis of core parameters such as negative pressure difference and speed difference.

[0045] Based on the above-described preferred embodiments of the present invention, and through the foregoing description, those skilled in the art can make various changes and modifications without departing from the inventive concept. The technical scope of this invention is not limited to the contents of the specification, but must be determined according to the scope of the claims.

Claims

1. A method for gas extraction pipe network anomaly library construction and multi-dimensional quantitative diagnosis, characterized in that: Includes the following steps: S1: Collect multi-source monitoring data of key nodes in the gas extraction pipeline network. The multi-source monitoring data includes real-time monitoring data and historical monitoring data. Based on the multi-source monitoring data, establish a multi-source monitoring data sample library for the pipeline network. S2: Preprocess the multi-source monitoring data in the sample library to complete data quality control, missing value verification, outlier labeling and normalization, and determine the stable range and dynamic change characteristics of each parameter; S3: Collect multi-source real-time monitoring data of key nodes in the gas extraction pipeline network as the real-time input data source for the diagnostic model; S4: Based on the preprocessed multi-source monitoring data from the sample library, calibrate the parameter benchmarks, fluctuation thresholds, feature correlation patterns, and judgment systems under normal operating conditions, and construct a three-level progressive diagnostic model; input the real-time multi-source monitoring data collected in step S3 into the three-level progressive diagnostic model, and perform feature extraction, quantitative calculation, and trend analysis on the real-time multi-source monitoring data through the three-level progressive diagnostic model, and perform multi-dimensional comparison and matching with the normal operating condition benchmarks and abnormal feature patterns in the sample library to perform three-level progressive diagnosis; S5: Outputs diagnostic results including the anomaly types, pipeline ranges, and severity levels of the entire gas extraction pipeline network, forming a complete closed-loop process of multi-source data acquisition and database construction, data preprocessing, real-time data acquisition, model diagnosis, and result output.

2. The abnormal library construction and multi-dimensional quantitative diagnosis method for gas extraction pipe network according to claim 1, characterized in that: In step S4, the three-level progressive diagnosis specifically includes: Level 1: Extract the overall monitoring values ​​of pressure and flow velocity from all measuring points in the pipeline network, calculate the mean, rate of change, and overall fluctuation range of pressure / flow velocity at all measuring points, compare the calculation results with the stable pressure and flow velocity range and parameter fluctuation threshold under normal operating conditions in the sample library, and comprehensively determine whether the pipeline network operating parameters exceed the normal fluctuation range or deviate from the normal operating conditions based on whether the overall pressure rise and fall exceeds the threshold, whether the overall trend of flow velocity changes deviates from the normal pattern, and whether the fluctuation of parameters at multiple measuring points shows consistent abnormality. Level 2: For real-time data that is determined to deviate from normal operating conditions, calculate the coordinated changes in pressure and flow velocity at each measuring point, spatial distribution characteristics, and correlation degree of main and branch parameters. Compare and match the calculated characteristic patterns with the typical characteristic spectra of four types of anomalies in the sample library: main blockage, branch blockage, main leakage, and branch leakage. Based on the characteristic matching degree, complete the fine distinction of anomaly types. Combined with the spatial distribution pattern of parameter anomalies, preliminarily determine the main / branch to which the anomaly belongs and the corresponding pipeline section. Level 3: For the initially identified abnormal pipeline sections, calculate the parameter sensitivity, pipeline anomaly correlation, and multi-parameter change consistency coefficient of each measuring point within the section. Compare the calculation results with the sensitivity threshold and correlation grading standards calibrated in the sample library. Using the measuring point with the highest anomaly correlation as the core, identify and delineate the specific pipeline section range where the anomaly is located by combining the spatial correlation of the measuring points. Based on the quantitative results of sensitivity and correlation, complete the graded assessment of the severity of the anomaly.

3. The method for constructing an anomaly database and performing multi-dimensional quantitative diagnosis of a gas extraction pipeline network according to claim 1, characterized in that: The multi-source monitoring data includes extraction negative pressure, gas mixing flow rate, methane concentration, oxygen concentration, carbon monoxide concentration, carbon dioxide concentration, pump station power, and pipeline temperature, which are used to comprehensively characterize the operating status and safety situation of key nodes in the pipeline network.

4. The abnormal library construction and multi-dimensional quantitative diagnosis method for gas extraction pipe network according to claim 1, characterized in that: In step S2, the preprocessing of multi-source monitoring data in the sample library specifically includes: labeling and tracing missing, distorted, or sensor-failed data; normalizing the data to the [0,1] interval using the Min-Max standardization formula; defining the parameter stability interval based on historical normal operating conditions; and marking data exceeding the threshold as potential anomalies. The standardized calculation formula is as follows: In the formula, and These are the upper and lower limits of the threshold values ​​for each parameter, obtained based on statistical data from historical normal operating conditions.

5. The abnormal library construction and multi-dimensional quantitative diagnosis method for gas extraction pipe network according to claim 2, characterized in that: In the second-level diagnosis, the synergistic relationship between pressure and flow velocity is calculated and compared with the four types of abnormal feature spectra in the sample library to distinguish between blockage and leakage anomalies. Based on the spatial distribution pattern of the parameter anomalies and the matching of the main and branch characteristic patterns calibrated in the sample library, it is preliminarily determined whether the anomaly is located in the main or branch road. The specific mathematical determination formula is defined as follows: Regarding the blockage of a branch road, this situation occurs if the following conditions are met: Regarding main road congestion, if the following conditions are met: Regarding branch circuit leakage, the following conditions must be met: Regarding branch circuit leakage, the following conditions must be met: In the formula, For a set of branches, each branch measurement point Each has a corresponding associated main road measuring point. ; The branch index variable represents the set. Zhongde Di A side road; For set A specific branch in the middle, determined by the existential quantifier Definition: A target branch road suspected of being blocked; branch road The change in negative pressure, relative to normal operating conditions, is expressed as follows: a negative value indicates an increase in the absolute value of negative pressure, and a positive value indicates a decrease in the absolute value of negative pressure. This represents the change in negative pressure at the main road measuring point corresponding to branch i. branch road The change in flow velocity, a positive value indicates an increase in velocity, and a negative value indicates a decrease, and so on. This represents the change in flow velocity at the main road measuring point corresponding to branch i. To retrieve a set The maximum value of the corresponding parameter for all branches; , Representing branch roads The collection of upstream and downstream branches.

6. The method for constructing an anomaly database and performing multi-dimensional quantitative diagnosis of a gas extraction pipeline network according to claim 5, characterized in that: In the third-level diagnosis, the sensitivity of the measuring point to negative pressure and the correlation between pipeline abnormalities are calculated using the following formula: The severity of the anomaly is graded based on the correlation between pipeline anomalies, and the grading formula is as follows: In the formula: The negative pressure sensitivity at the i-th measuring point; This represents the real-time negative pressure monitoring value at the i-th measuring point; The historical negative pressure reference value under normal operating conditions for the i-th measuring point; Let be the degree of abnormal correlation of the j-th pipeline segment; Let j be the set of monitoring points corresponding to the j-th pipeline segment; Weighting of the spatial influence of the measuring points on the pipeline section; This represents the severity level of the abnormality. and The grading thresholds determined by historical operating condition statistics must meet the following conditions. ; , to The degree of abnormality represents an increasing trend.

7. The abnormal library construction and multi-dimensional quantitative diagnosis method for gas extraction pipe network according to claim 6, characterized in that: The third-level diagnostic process integrates multi-source parameter complementarity and cross-validation. It achieves collaborative discrimination by calculating the consistency coefficient of multi-parameter changes through a model, and compares the calculation results with the consistency threshold calibrated in the sample library. This reduces the tendency for misjudgment caused by noise from a single sensor and maintains the robustness of the diagnostic process. The consistency coefficient of multi-parameter changes is calculated using the following formula: In the formula: is a multi-parameter consistency coefficient; is the number of parameter types participating in diagnosis; is the real-time change amount of the kth parameter; is the calibrated steady-state reference value in the kth parameter sample library.

8. The abnormal library construction and multi-dimensional quantitative diagnosis method for gas extraction pipe network according to claim 7, characterized in that: The three-level progressive diagnostic model has adaptive expansion capabilities and can iteratively optimize indicator thresholds and discrimination logic based on changes in pipeline topology and historical diagnostic feedback, combined with newly added effective diagnostic cases in the sample library. The threshold adaptive iterative correction formula is: ; In the formula: is the updated diagnosis threshold after iteration; is the original diagnosis threshold before iteration; is the learning coefficient; is the observation threshold obtained by statistical analysis of historical effective diagnosis cases in the sample library.

9. The method for constructing an anomaly database and performing multi-dimensional quantitative diagnosis of a gas extraction pipeline network according to claim 1, characterized in that: In step S5, the output diagnostic result is oriented towards engineering operation and maintenance design, and a standardized result is generated after model quantitative calculation, thereby providing a direct basis for on-site fault disposal. The specific expression of the standardized result is: ; In the formula, For comprehensive diagnostic output; This refers to the pipeline network operation status determination result obtained from the first level of the three-level progressive diagnosis, which includes both normal and abnormal states. If the condition is normal, no further results will be output. This is a detailed differentiation result of the abnormality type obtained from the second level of diagnosis in a three-level progressive diagnosis; The result of identifying and delineating the abnormal pipeline range obtained from the third level of diagnosis in the three-level progressive diagnosis; This represents the severity grading result of the abnormality obtained from the third-level diagnosis in the third-level progressive diagnosis.

Images