A gas leakage intelligent detection and early warning method and system

Abstract

The present application relates to the technical field of gas leakage detection, and particularly relates to a gas leakage intelligent detection and early warning method and system, comprising: according to the pipeline layout and use scene of a target gas pipeline, completing the division of monitoring sub-areas of a target area, and uniquely identifying each monitoring sub-area. The present application divides monitoring sub-areas by combining the layout and use scene of the target gas pipeline, extracts the working condition characteristics of each area and defines the working condition type, adjusts the data collection parameters according to the working condition characteristics, makes the data collection more targeted, effectively reduces the collection of invalid data, and improves the effectiveness of the monitoring data; the initial stability index is corrected for deviation by a working condition characteristic correction model, which eliminates the monitoring error caused by working condition fluctuations, and makes the stability index better reflect the actual operation state of each area.

Description

Technical Field

[0001] This invention relates to the field of gas leak detection technology, specifically to an intelligent gas leak detection and early warning method and system. Background Technology

[0002] Natural gas, as a clean energy source, is widely used in residential and industrial production. However, leaks throughout its pipeline transportation and end-use processes can directly lead to safety accidents such as explosions and poisoning. Therefore, real-time detection and accurate early warning of gas leaks are crucial aspects of gas safety management. With the development of intelligent sensing and algorithm analysis technologies, various gas leak early warning technologies combining multi-source data acquisition and index calculation are gradually being implemented. By acquiring operational data from gas pipelines and surrounding environmental data, and calculating relevant stability indices to assess leak risk, these technologies have become a primary means of improving the safety of gas usage.

[0003] In the prior art, such as the invention patent with publication number CN117037455B, a gas leak alarm system and method based on artificial intelligence, the scheme first divides the gas pipeline area into monitoring sub-areas with equal area, then collects gas usage information and environmental information of each sub-area, calculates three types of stability indices: gas monitoring, gas pipeline status, and environmental status, obtains the gas leak assessment coefficient by directly summing the three types of indices, compares the coefficient with the preset value to complete the leak risk judgment, and stores the historical assessment coefficients of each sub-area in a unified manner.

[0004] In practical applications, the aforementioned technical solutions simply perform a linear superposition of various stability indices when calculating leakage assessment coefficients. They fail to consider the differences in actual operating conditions across different monitoring sub-regions, nor do they account for the varying degrees of influence of gas monitoring, pipeline status, and environmental status indices on leakage risk assessment under different operating conditions. This results in calculated assessment coefficients that cannot accurately match the true leakage risk situation in each sub-region. Under complex actual operating conditions, this calculation method is prone to bias in risk assessment. It may fail to effectively identify low-probability, potential leakage risks in some areas, or misjudge areas without leakage risk as having leakage risk. This directly reduces the accuracy of gas leak early warning and increases the ineffective workload of safety supervision. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides an intelligent detection and early warning method and system for gas leaks, which solves the problems of existing technologies failing to consider differences in operating conditions, using fixed weights and simple index superposition leading to biased gas leak risk assessments, and low early warning accuracy.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a smart detection and early warning method for gas leaks, comprising the following steps:

[0007] S1. Based on the pipeline layout and usage scenario of the target gas pipeline, complete the division of the target area into monitoring sub-areas and assign a unique identifier to each monitoring sub-area.

[0008] S2. Collect real-time operation correlation data of each monitoring sub-area, extract the operating condition characteristics of each monitoring sub-area, and define the operating condition type of each monitoring sub-area based on the operating condition characteristics.

[0009] S3. Adjust the frequency and accuracy of data acquisition according to the operating characteristics and types of each monitoring sub-area, and collect gas alarm data, gas pipeline status data and surrounding environment data of each monitoring sub-area;

[0010] S4. Calculate the initial gas monitoring stability index, gas pipeline status stability index, and environmental status stability index based on the gas alarm data, gas pipeline status data, and surrounding environment data. Then, perform deviation correction on the gas monitoring stability index, gas pipeline status stability index, and environmental status stability index in combination with the operating condition characteristics to obtain the corrected stability index.

[0011] S5. Retrieve historical early warning data and handling results, combine them with the real-time operating conditions of each monitoring sub-area, optimize and adjust the pre-stored initial index weights of each operating condition type, complete the self-learning update of the weights, and obtain the operating condition adaptation weights.

[0012] S6. Perform weighted fusion calculation on the three types of modified stability indices according to the weighted fusion algorithm to obtain the gas leakage assessment coefficient of each monitoring sub-region;

[0013] S7. Retrieve the pre-stored leakage risk classification thresholds for each operating condition type, compare the gas leakage assessment coefficient of each monitoring sub-area with the corresponding threshold, determine the leakage risk level of each monitoring sub-area, and trigger early warning measures according to the risk level.

[0014] Furthermore, in S2, the operating condition characteristics of each monitoring sub-region are extracted, including:

[0015] Extract the characteristics of gas usage load variation, which include the average daily peak gas consumption, the standard deviation of gas consumption fluctuation, and the frequency of instantaneous flow change.

[0016] Extract pipeline aging characteristics, including pipeline years of operation, material corrosion rate, mechanical stress distribution at interface connections, and integrity data of anti-corrosion coating.

[0017] Extract the surrounding environmental interference features, which include the temperature and humidity cycle fluctuation amplitude of the monitoring area, the distribution of ground vibration frequency, the frequency of surrounding construction, and the intensity of interference from third-party activities;

[0018] The definition of the operating conditions includes dividing the monitoring sub-areas into high load fluctuation type, severe pipeline aging type, complex environmental interference type, and normal operation type.

[0019] Furthermore, in step S3, the frequency and accuracy of data acquisition are adjusted according to the operating characteristics and types of each monitoring sub-region, including:

[0020] For monitoring sub-areas defined as high-load fluctuation type, increase the sampling frequency of gas flow sensor and pressure sensor, and increase the filtering accuracy of flow fluctuation;

[0021] For monitoring sub-areas defined as severely aged pipelines, improve the acquisition accuracy of acoustic monitoring data and pipeline deformation data, and extend the online duration of acoustic sensors;

[0022] For monitoring sub-areas defined as having complex and interfering environments, the number of data collection points for temperature and humidity sensors and acceleration sensors is increased, and redundant data from multiple points is used to eliminate environmental noise interference.

[0023] The data transmission bandwidth allocation is dynamically adjusted in real time according to the operating conditions of each monitoring sub-area, and the bandwidth corresponding to the preset high-priority bandwidth threshold is allocated to the monitoring sub-area with complex operating conditions.

[0024] Furthermore, in step S4, deviation corrections are made for the gas monitoring stability index, the gas pipeline status stability index, and the environmental status stability index, including:

[0025] A working condition characteristic correction model is established, and the deviation correction factor for each monitoring sub-region is calculated using the working condition characteristics as input parameters.

[0026] The gas load correction factor is used to compensate the gas monitoring stability index to offset the interference of sudden flow changes caused by normal peak gas consumption.

[0027] The gas pipeline stability index is augmented using a pipeline aging correction factor, and the risk weight of the aging pipeline under minute pressure changes is adjusted by the gain coefficient.

[0028] The environmental stability index is denoised by using an environmental interference correction factor to filter sensor zero-point drift data caused by construction vibration or extreme weather, thereby outputting a corrected stability index after eliminating the influence of operating condition fluctuations.

[0029] Furthermore, the self-learning update of the weights in S5 includes:

[0030] Construct a weight adjustment matrix with each working condition type as the dimension, and set the initial weights as the initial values ​​of the matrix;

[0031] The system receives real-time feedback on the on-site handling results of the early warning, uses the accuracy of the early warning as the objective function, and calculates the sensitivity of the three types of corrected stability indices to the accuracy of the early warning under each working condition.

[0032] If the sensitivity of a certain stability index under the corresponding working condition is higher than a preset threshold, then the value of the stability index in the weight adjustment matrix is ​​increased.

[0033] If historical warning data shows that a certain stability index causes a false alarm frequency higher than a preset false alarm frequency threshold under the corresponding operating condition, then the weight ratio of the stability index in the weight adjustment matrix is ​​reduced, and the dynamic evolution of the weight under each operating condition is realized through feedback closed loop.

[0034] Furthermore, in S6, the gas leakage assessment coefficients for each monitored sub-region are obtained, including:

[0035] The modified stability index is mapped to a unified dimension space to obtain standardized gas monitoring components, pipeline state components, and environmental state components.

[0036] Call the corresponding working condition adaptation weights for the working condition type and perform a weighted summation calculation;

[0037] The calculation results are processed by time series moving average to eliminate the influence of instantaneous data jitter, and a gas leakage assessment coefficient reflecting the actual leakage risk trend of the monitored sub-area is obtained.

[0038] Furthermore, in S7, determining the leakage risk level of each monitoring sub-area includes:

[0039] The leakage risk level is divided into four levels: no risk, low risk, medium risk, and high risk.

[0040] The matching hierarchical threshold array is retrieved from the threshold library based on the operating condition type.

[0041] When the gas leak assessment coefficient is within the first threshold range, it is determined to be low risk;

[0042] When the gas leak assessment coefficient is within the second threshold range, it is determined to be of medium risk;

[0043] When the gas leak assessment coefficient exceeds the third threshold, it is determined to be high risk;

[0044] Specifically, for monitoring sub-areas with severe pipeline aging, the threshold settings for entering the medium-risk and high-risk levels have been lowered.

[0045] Furthermore, the triggering logic for the early warning measures includes:

[0046] For low-risk monitoring sub-areas, system log recording is performed and routine inspection plan suggestions are generated;

[0047] For monitoring sub-areas at medium risk level, trigger the audible and visual alarm terminal and push abnormal data and real-time location to the preset mobile terminal of the supervisor, and start the encrypted monitoring mode.

[0048] For high-risk monitoring sub-areas, the emergency shut-off valve of the corresponding pipeline section is immediately triggered, and the early warning information is simultaneously pushed to the emergency response command system and regulatory departments at all levels. The optimal repair route is also planned based on the relevant on-site data.

[0049] Furthermore, the method also includes a full data storage and traceability step:

[0050] The system stores in real time the monitoring sub-area division data, operating condition characteristic data, multi-dimensional collected basic information, stability index before and after correction, weight self-learning process data, gas leakage assessment coefficient, risk classification results, and early warning and handling results.

[0051] Establish a correlation index between data items based on timestamps and regional identifiers to realize historical retrospective analysis of the entire gas leak detection and early warning process.

[0052] This invention also provides a smart gas leak detection and early warning system, comprising:

[0053] The gas pipeline area division module is used to divide monitoring sub-areas according to pipeline layout and usage scenarios and assign unique identifiers;

[0054] The operating condition feature extraction module is connected to the gas pipeline area division module and is used to collect real-time operation correlation data of each monitoring sub-area and extract operating condition features to define the operating condition type.

[0055] The multi-dimensional information acquisition module is used to adjust the acquisition frequency and accuracy according to the characteristics of the working conditions and to collect multi-dimensional basic data of each monitoring sub-area;

[0056] The index dynamic correction module is connected to the working condition feature extraction module and the multi-dimensional information acquisition module. It is used to calculate the initial stability index and combine the working condition features to correct the deviation, so as to obtain the corrected stability index.

[0057] The working condition weight self-learning module is used to retrieve historical early warning and handling data and combine them with real-time working condition characteristics to complete the self-learning update of weights and output working condition adapted weights.

[0058] The weighted fusion evaluation module, connected to the index dynamic correction module and the working condition weight self-learning module, is used to calculate the gas leakage evaluation coefficient through the weighted fusion algorithm.

[0059] The graded early warning execution module, connected to the weighted fusion evaluation module, is used to perform risk level determination and trigger corresponding early warning measures, while feeding back the handling results to the working condition weight self-learning module.

[0060] The full data storage module is bidirectionally connected to each of the modules and is used to store the entire process data of detection and early warning and to provide historical data retrieval services for the calculation and processing of each module.

[0061] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0062] This invention divides monitoring sub-regions by combining the layout and usage scenarios of the target gas pipeline, extracts the operating characteristics of each region and defines the operating condition type, and adjusts the data acquisition parameters accordingly based on the operating condition characteristics, making data acquisition more targeted, effectively reducing the collection of invalid data, and improving the effectiveness of monitoring data. A condition characteristic correction model is used to correct the initial stability index, eliminating monitoring errors caused by operating condition fluctuations, making the stability index more reflective of the actual operating status of each region. By constructing a feedback loop with self-learning updates of weights, the index weights of each operating condition type are dynamically optimized based on the accuracy of early warnings and the frequency of false alarms, solving the risk assessment bias problem caused by fixed weights and simple linear superposition in existing technologies. A gas leakage assessment coefficient is obtained through a weighted fusion algorithm and moving average processing, and differentiated risk classification thresholds are set according to the operating condition type, achieving accurate classification of leakage risks. Simultaneously, targeted early warning measures are triggered according to different risk levels, making early warning response more scientific and timely. Full data storage and traceability provide data support for the continuous optimization of the entire detection and early warning system and gas safety management. In practical applications, this method can effectively improve the accuracy of gas leak risk assessment, effectively identify potential leak risks, reduce false alarm and false alarm rates, reduce the ineffective workload of gas safety supervision, and solve the technical problem of insufficient accuracy of gas leak early warning in existing technologies. Attached Figure Description

[0063] Figure 1 This is an overall flowchart of the intelligent gas leak detection and early warning method of the present invention;

[0064] Figure 2 This is a flowchart of the working condition feature extraction and data acquisition parameter adjustment process of the present invention;

[0065] Figure 3 This is a flowchart of the dynamic correction process for the stability index of the present invention;

[0066] Figure 4 This is a flowchart of the self-learning update process for the working condition weights of the present invention;

[0067] Figure 5 This is a flowchart illustrating the gas leak risk classification and early warning measures triggering process of the present invention;

[0068] Figure 6 This is a module architecture diagram of the intelligent gas leak detection and early warning system of the present invention. Detailed Implementation

[0069] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0070] Example 1

[0071] Please see Figures 1-5 This invention provides an intelligent detection and early warning method for gas leaks, comprising the following steps:

[0072] S1. Based on the pipeline layout and usage scenario of the target gas pipeline, complete the division of the target area into monitoring sub-areas and assign a unique identifier to each monitoring sub-area.

[0073] In practice, the pipeline layout of the target gas pipeline is sorted out according to the pipe diameter, transmission pressure and pipe segment connection method of the pipeline network. The usage scenarios are divided into residential living scenarios, industrial production scenarios and commercial operation scenarios according to the regional functions. The division of monitoring sub-areas is based on the independent transmission unit of the pipe segment, while taking into account the homogeneity of the usage scenarios. After the division is completed, each monitoring sub-area is assigned a unique character identifier containing geographical information and pipe segment number. This identifier will serve as the association identifier for all subsequent monitoring data and run through the entire detection and early warning process.

[0074] Please see Figure 2 In one embodiment, S2, real-time operation correlation data of each monitoring sub-region is collected, operating condition characteristics of each monitoring sub-region are extracted from them, and the operating condition type of each monitoring sub-region is defined based on the operating condition characteristics.

[0075] Furthermore, the operating condition characteristics of each monitoring sub-region are extracted in S2, including:

[0076] Extract the characteristics of gas usage load variation, which include the average daily peak gas consumption, the standard deviation of gas consumption fluctuation, and the frequency of instantaneous flow rate mutation.

[0077] Extract pipeline aging characteristics, which include pipeline years of operation, material corrosion rate, mechanical stress distribution at interface connections, and integrity data of anti-corrosion coating.

[0078] Extract the surrounding environmental interference features, which include the temperature and humidity cycle fluctuation amplitude of the monitoring area, the distribution of ground vibration frequency, the frequency of surrounding construction, and the intensity of interference from third-party activities;

[0079] The definition of the operating conditions includes dividing the monitoring sub-areas into high load fluctuation type, severe pipeline aging type, complex environmental interference type, and normal operation type.

[0080] Specifically, in actual data acquisition, real-time operational correlation data is obtained through various types of intelligent sensing devices deployed in each monitoring sub-area, including flow sensors, pressure sensors, corrosion sensors, strain gauges, temperature and humidity sensors, and acceleration sensors. This is supplemented by historical gas consumption data from the gas operation platform, pipeline archive data, and construction activity data from regional monitoring. The extraction of gas usage load variation characteristics is achieved through time-series statistical analysis of the flow data collected by the sensors. The daily average peak gas consumption is the maximum instantaneous gas flow rate within a statistical 24-hour period. The standard deviation of gas consumption fluctuations is obtained by calculating the statistical standard deviation of gas consumption over a continuous time period. The instantaneous flow rate mutation frequency is the frequency of changes within a unit of time. The number of times the flow rate exceeded the normal fluctuation range; in the pipeline operation aging characteristics, the pipeline's service life was retrieved from the pipeline archive data, the material corrosion rate was calculated using real-time detection data from corrosion sensors, the mechanical stress distribution at the interface connection was analyzed spatially using stress data collected by strain gauges, and the integrity data of the anti-corrosion coating was determined using coating reflection data obtained by infrared detection equipment; the extraction of surrounding environmental interference characteristics was achieved by trend analysis of detection data from temperature and humidity sensors and acceleration sensors, combined with regional construction reporting data and third-party activity identification data from video surveillance, to calculate the amplitude of temperature and humidity cycle fluctuations, the distribution of ground vibration frequency, the frequency of surrounding construction, and the intensity of third-party activity interference. The definition of operating conditions is achieved by constructing an operating condition feature classification model. The three types of operating condition features extracted are used as input to the model. Through the preset feature threshold range, each monitoring sub-area is defined as high load fluctuation type, severe pipeline aging type, complex environmental interference type, and normal operation type. Among them, areas with large fluctuations in gas demand, such as commercial complexes and catering concentration areas, are usually defined as high load fluctuation type. Areas with old pipeline sections that have been in use for more than 15 years are usually defined as severe pipeline aging type. Pipeline areas near urban construction sections and main traffic arteries are usually defined as complex environmental interference type. Newly built residential pipeline areas with a single use scenario are usually defined as normal operation type.

[0081] Please see Figure 2 In one embodiment, S3, according to the operating characteristics and operating type of each monitoring sub-area, the frequency and accuracy of data acquisition are adjusted, and gas alarm data, gas pipeline status data and surrounding environment data of each monitoring sub-area are collected.

[0082] Furthermore, S3 adjusts the frequency and accuracy of data acquisition based on the operating characteristics and types of each monitoring sub-region, including:

[0083] For monitoring sub-areas defined as high-load fluctuation type, increase the sampling frequency of gas flow sensor and pressure sensor, and increase the filtering accuracy of flow fluctuation;

[0084] For monitoring sub-areas defined as severely aged pipelines, improve the acquisition accuracy of acoustic monitoring data and pipeline deformation data, and extend the online duration of acoustic sensors;

[0085] For monitoring sub-areas defined as having complex and interfering environments, the number of data collection points for temperature and humidity sensors and acceleration sensors is increased, and redundant data from multiple points is used to eliminate environmental noise interference.

[0086] The data transmission bandwidth allocation is dynamically adjusted in real time according to the operating conditions of each monitoring sub-area, and the bandwidth corresponding to the preset high-priority bandwidth threshold is allocated to the monitoring sub-area with complex operating conditions.

[0087] Specifically, in the actual parameter adjustments, for the high-load fluctuation monitoring sub-area, the conventional sampling frequency of the gas flow sensor and pressure sensor was increased from 1 time / minute to 1 time / 10 seconds. The Kalman filter algorithm was used for the flow fluctuation filtering, and the number of filtering iterations was increased to improve the filtering accuracy and more accurately capture subtle changes in flow and pressure. For the monitoring sub-area with severe pipeline aging, the sampling accuracy of the acoustic monitoring data was increased from 0.1kHz to 0.01kHz, the acquisition accuracy of the pipeline deformation data was increased from 0.1mm to 0.01mm, and the online duration of the acoustic sensor was extended from 12 hours / day to 24 hours / day to achieve continuous monitoring of the acoustic characteristics of minor pipeline leaks. For the monitoring sub-area with complex environmental interference, the number of acquisition points of the temperature and humidity sensor was increased from 1 to 3-5, and the number of acquisition points of the acceleration sensor was increased from 2 to 4-6. The redundant data from multiple acquisitions were fused using the arithmetic mean method to effectively eliminate detection errors caused by environmental noise. The dynamic adjustment of data transmission bandwidth is based on the total bandwidth resources of the gas monitoring system. A high-priority bandwidth threshold of 40% of the total bandwidth is preset, prioritizing the allocation of this bandwidth to monitoring sub-areas experiencing high load fluctuations, severe pipeline aging, or complex environmental interference. The remaining bandwidth is allocated to monitoring sub-areas operating under normal conditions, ensuring real-time and stable transmission of monitoring data from sub-areas with complex operating characteristics. After adjusting the data acquisition parameters and transmission bandwidth, gas alarm data, gas pipeline status data, and surrounding environmental data are synchronously collected from each monitoring sub-area through various sensing devices. Gas alarm data includes detection data from gas concentration sensors and abnormal flow and pressure alarm data. Gas pipeline status data includes data on pipeline deformation, stress, corrosion, and coating integrity. Surrounding environmental data includes data on temperature and humidity, ground vibration, atmospheric pressure, and wind speed.

[0088] Please see Figure 3 In one embodiment, S4, the initial gas monitoring stability index, gas pipeline status stability index, and environmental status stability index are calculated based on the gas alarm data, gas pipeline status data, and surrounding environment data. The gas monitoring stability index, gas pipeline status stability index, and environmental status stability index are then corrected for deviations based on the operating condition characteristics to obtain the corrected stability index.

[0089] Furthermore, in S4, deviation corrections are made for the gas monitoring stability index, gas pipeline condition stability index, and environmental condition stability index, including:

[0090] A working condition characteristic correction model is established, and the deviation correction factor for each monitoring sub-region is calculated using the working condition characteristics as input parameters.

[0091] The gas load correction factor is used to compensate the gas monitoring stability index to offset the interference of sudden flow changes caused by normal peak gas consumption.

[0092] The gas pipeline stability index is augmented using a pipeline aging correction factor, and the risk weight of the aging pipeline under minute pressure changes is adjusted by the gain coefficient.

[0093] The environmental stability index is denoised by using an environmental interference correction factor to filter sensor zero-point drift data caused by construction vibration or extreme weather, thereby outputting a corrected stability index after eliminating the influence of operating condition fluctuations.

[0094] Specifically, in the actual calculation, the collected gas alarm data, gas pipeline status data, and surrounding environmental data are first normalized, mapping all data to a unified dimension range of [0,1]. Values ​​closer to 1 indicate a more stable state, while values ​​closer to 0 indicate an anomaly. Based on the normalized data, a weighted average method is used to calculate the initial gas monitoring stability index. Gas pipeline condition stability index Environmental stability index The calculation formula is: ,in These represent gas monitoring, pipeline status, and environmental status, respectively. The number of indicators for the corresponding type of monitoring data; For the first The weights of each indicator are set according to the degree of influence of the indicator on leakage risk, and must meet the following conditions: ; For the first The normalized values ​​of the indicators.

[0095] After calculating the initial stability index, a neural network-based operating condition characteristic correction model is established. This model takes as input an operating condition feature vector composed of extracted gas usage load variation characteristics, pipeline aging characteristics, and surrounding environmental disturbance characteristics. Through model training, it outputs deviation correction factors for each monitoring sub-region, including a gas load correction factor. Pipeline aging correction factor Environmental interference correction factor The value of the correction factor is dynamically adjusted according to the complexity of the operating conditions, such as in high-load fluctuation monitoring sub-regions. Values ​​between 0.8 and 1.2 are used for monitoring sub-areas with severe pipeline aging. Values ​​between 1.1 and 1.5 are used for monitoring sub-regions with complex and interfering environments. The value ranges from 0.7 to 1.0. The initial stability index is corrected using a deviation correction factor to obtain the corrected stability index, calculated using the following formula: , , ,in To correct the gas monitoring stability index, To correct the gas pipeline condition stability index, This is the corrected environmental stability index. This deviation correction process effectively offsets monitoring data interference caused by peak gas consumption, improves the risk identification sensitivity of aging pipelines, and filters sensor data drift caused by environmental noise. This allows the corrected stability index to more accurately reflect the actual state of each monitoring sub-area, solving the problem of index calculation deviation caused by neglecting differences in operating conditions in existing technologies.

[0096] Please see Figure 4 In one embodiment, S5, historical early warning data and handling results are retrieved, and the pre-stored initial index weights of each working condition type are optimized and adjusted in combination with the real-time working condition characteristics of each monitoring sub-region. The weights are then updated through self-learning to obtain the working condition adaptation weights.

[0097] The self-learning update of weights in S5 includes:

[0098] Construct a weight adjustment matrix with each working condition type as the dimension, and set the initial weights as the initial values ​​of the matrix;

[0099] The system receives real-time feedback on the on-site handling results of the early warning, uses the accuracy of the early warning as the objective function, and calculates the sensitivity of the three types of corrected stability indices to the accuracy of the early warning under each working condition.

[0100] If the sensitivity of a certain stability index under the corresponding working condition is higher than a preset threshold, then the value of the stability index in the weight adjustment matrix is ​​increased.

[0101] If historical warning data shows that a certain stability index causes a false alarm frequency higher than a preset false alarm frequency threshold under the corresponding operating condition, then the weight ratio of the stability index in the weight adjustment matrix is ​​reduced, and the dynamic evolution of the weight under each operating condition is realized through feedback closed loop.

[0102] Specifically, in practical implementation, a weight adjustment matrix is ​​first constructed with operating condition type as the row and three types of corrected stability indices as the column. The matrix form is: The matrix consists of rows corresponding to four operating conditions: high load fluctuation, severe pipeline aging, complex environmental interference, and normal operation. The columns correspond to the weights of the gas monitoring stability index, pipeline condition stability index, and environmental condition stability index. The initial values ​​of the matrix are set based on engineering experience and historical data in the gas industry; for example, the initial weight for the normal operation type is set to... , , The initial weight for severely aged pipelines is set to... , , .

[0103] Historical early warning data and on-site handling results for each monitoring sub-area are retrieved from the full data storage module. On-site handling results include information such as the actual leakage situation after early warning verification, the cause of false alarms, and missed alarms. Early warning accuracy is used as the objective function, defined as the ratio of the number of times an actual leak was correctly warned to the total number of early warnings. For each operating condition type, the sensitivity of the three types of corrected stability indices to early warning accuracy is calculated. ,in , Representing four operating condition types, sensitivity is calculated using partial derivatives, i.e. , To ensure accurate early warning, a preset sensitivity threshold is set. This threshold is obtained through training with historical early warning data and has a value of 0.8. The sensitivity of a certain stability index... This indicates that the stability index has a significant impact on the accuracy of early warnings for the corresponding operating conditions. Therefore, its value in the weight adjustment matrix is ​​increased using a weight adjustment formula, which is as follows: ,in The weighting adjustment coefficient is calibrated based on actual operating conditions and ranges from 0.1 to 0.3. Simultaneously, the false alarm frequency of each stability index under the corresponding operating condition type is statistically analyzed. The false alarm frequency is the ratio of the number of false alarms to the total number of warnings, with a preset false alarm frequency threshold. The value is 0.3. This indicates that the stability index is prone to false alarms for the corresponding operating conditions. Therefore, its weight in the weight adjustment matrix can be reduced using a weight adjustment formula. The adjustment formula is as follows: ,in This is the weight adjustment coefficient, with a value between 0.1 and 0.3. After weight adjustment, the weights in each row of the matrix need to be normalized to ensure that the sum of the weights in each row is 1. This refers to the working condition adaptation weights for each working condition type. The self-learning update process of these weights constructs a feedback loop, using the on-site handling results and the accuracy of the early warning as the basis for weight adjustment. This allows the weights for each working condition type to dynamically evolve with the actual monitoring situation, solving the problem that fixed weights in existing technologies cannot adapt to different working conditions. This makes subsequent risk assessments more closely aligned with the actual situation of each sub-area.

[0104] Please see Figure 5 In one embodiment, S6, the three types of modified stability indices are weighted and fused according to the weighted fusion algorithm to obtain the gas leakage assessment coefficient of each monitoring sub-region.

[0105] Furthermore, in S6, the gas leakage assessment coefficients for each monitoring sub-area are obtained, including:

[0106] The modified stability index is mapped to a unified dimension space to obtain standardized gas monitoring components, pipeline state components, and environmental state components.

[0107] Call the corresponding working condition adaptation weights for the working condition type and perform a weighted summation calculation;

[0108] The calculation results are processed by time series moving average to eliminate the influence of instantaneous data jitter, and a gas leakage assessment coefficient reflecting the actual leakage risk trend of the monitored sub-area is obtained.

[0109] Specifically, in actual calculations, the corrected stability index is first... , , The system is then standardized again, mapped to a unified dimensional space of [0,1], eliminating minor dimensional differences in the calculation of different indices, and obtaining standardized gas monitoring components. Pipeline state components Environmental state components Based on the operating condition type of each monitoring sub-region, the weight adjustment matrix is ​​used. Retrieve the corresponding working condition adaptation weights , , The original evaluation coefficients are obtained by performing a weighted summation calculation. The calculation formula is: Because the data collected by the sensors has instantaneous fluctuations, the original assessment coefficients will fluctuate slightly and cannot truly reflect the changing trend of leakage risk. Therefore, the original assessment coefficients are processed by time series moving average, and the size of the moving window is determined. Based on the data collection frequency setting, the value is 5-10 data points, and the formula for calculating the moving average is: ,in This is the gas leak assessment coefficient. For the first The original assessment coefficients at each time point. The gas leakage assessment coefficient ranges from [0,1]. The closer the value is to 0, the higher the gas leakage risk in the monitored sub-area; the closer the value is to 1, the more stable the operating status of the monitored sub-area and the lower the leakage risk. Through weighted fusion and moving average processing, the monitoring results of the three types of stability indices can be integrated, and targeted risk assessment can be achieved by combining the operating condition adaptation weights. At the same time, the influence of instantaneous data fluctuations is eliminated, allowing the assessment coefficients to truly reflect the leakage risk trend of each monitored sub-area.

[0110] In one embodiment, S7, retrieve the pre-stored leakage risk classification thresholds for each operating condition type, compare the gas leakage assessment coefficient of each monitoring sub-area with the corresponding threshold, determine the leakage risk level of each monitoring sub-area, and trigger early warning measures according to the risk level.

[0111] Furthermore, S7 determines the leakage risk level of each monitoring sub-area, including:

[0112] The leakage risk level is divided into four levels: no risk, low risk, medium risk, and high risk.

[0113] The matching hierarchical threshold array is retrieved from the threshold library based on the operating condition type.

[0114] When the gas leak assessment coefficient is within the first threshold range, it is determined to be low risk;

[0115] When the gas leak assessment coefficient is within the second threshold range, it is determined to be of medium risk;

[0116] When the gas leak assessment coefficient exceeds the third threshold, it is determined to be high risk;

[0117] Specifically, for monitoring sub-areas with severe pipeline aging, the threshold settings for entering the medium-risk and high-risk levels have been lowered.

[0118] Furthermore, the triggering logic for early warning measures includes:

[0119] For low-risk monitoring sub-areas, system log recording is performed and routine inspection plan suggestions are generated;

[0120] For monitoring sub-areas at medium risk level, trigger the audible and visual alarm terminal and push abnormal data and real-time location to the preset mobile terminal of the supervisor, and start the encrypted monitoring mode.

[0121] For high-risk monitoring sub-areas, the emergency shut-off valve of the corresponding pipeline section is immediately triggered, and the early warning information is simultaneously pushed to the emergency response command system and regulatory departments at all levels. The optimal repair route is also planned based on the relevant on-site data.

[0122] Specifically, in practical implementation, a leakage risk classification threshold library is first constructed. The classification threshold array in the threshold library is obtained by training historical early warning data, leakage accident data, and machine learning models for various operating conditions. Different threshold ranges are set for different operating conditions. Among them, the classification threshold ranges for normal operation, high load fluctuation, and complex environmental interference are: no risk. Low risk (First threshold interval), medium risk (Second threshold interval), high risk (Third Threshold); For monitoring sub-areas with severe pipeline aging, to identify potential leakage risks earlier, the threshold settings for risk and high risk are lowered. The tiered threshold range is: No Risk Low risk (First threshold interval), medium risk (Second threshold interval), high risk (Third threshold).

[0123] Based on the operating conditions of each monitoring sub-area, the system retrieves a matching tiered threshold array from the threshold database. The calculated gas leak assessment coefficient is compared with each threshold range to determine the leak risk level of each monitoring sub-area as no risk, low risk, medium risk, or high risk. Corresponding early warning measures are triggered for different risk levels. For low-risk monitoring sub-areas, the system automatically logs the assessment coefficient, operating condition characteristics, and monitoring data. Simultaneously, it generates personalized routine inspection plan suggestions based on the operating condition type of the sub-area. For example, for low-risk sub-areas with high load fluctuations, it suggests increasing the frequency of flow and pressure inspections; for low-risk sub-areas with severely aged pipelines, it suggests increasing the frequency of pipeline flaw detection and anti-corrosion coating inspections. For medium-risk monitoring sub-areas, the system immediately triggers the audible and visual alarm terminals deployed on-site in that sub-area, providing on-site early warning through audible and visual signals. Simultaneously, it pushes abnormal monitoring data, gas leak assessment coefficients, and real-time geographical location information for that sub-area to preset mobile terminals for supervisors, including mobile apps and handheld terminals, and initiates encrypted monitoring. The system further increases the data collection frequency of the sub-area by 50% and activates multi-sensor linkage monitoring to achieve key monitoring of the area. For high-risk monitoring sub-areas, the system immediately triggers the emergency shut-off valve of the corresponding pipeline section through the IoT communication module to cut off gas supply and prevent the leakage accident from escalating. At the same time, the system pushes the early warning information to the city's gas emergency response command system, local gas regulatory departments, fire and rescue departments, gas repair teams, and other relevant departments at all levels through the early warning information release platform. The early warning information includes sub-area identification, geographical location, assessment coefficient, abnormal data, leakage risk level, etc. In addition, the system combines the geographical information of the gas pipeline network layout, urban traffic data, and on-site related data such as the location data of the repair team to plan the optimal repair route through path planning algorithm, providing guidance for the rapid implementation of repair work.

[0124] In one embodiment, during the execution of the entire detection and early warning process, a full data storage and traceability step is also performed simultaneously. The method further includes the full data storage and traceability step:

[0125] The system stores in real time the monitoring sub-area division data, operating condition characteristic data, multi-dimensional collected basic information, stability index before and after correction, weight self-learning process data, gas leakage assessment coefficient, risk classification results, and early warning and handling results.

[0126] Establish a correlation index between data items based on timestamps and regional identifiers to realize historical retrospective analysis of the entire gas leak detection and early warning process.

[0127] Specifically, in practical implementation, a distributed cloud database is used to store all data. This database has the characteristics of large capacity, high fault tolerance, and scalability, which can meet the storage needs of massive amounts of gas monitoring data. The stored data includes geographical information, unique identifiers, and division criteria for monitoring sub-regions; raw data, extraction results, and operational condition type definition results for operating condition characteristics; basic information such as gas alarm data, pipeline status data, and environmental data collected by multiple sensors; calculation process and results of initial stability index and corrected stability index; process data such as initial weights, weight adjustment matrix, sensitivity and false alarm frequency calculation data, and operational condition adaptation weights for weight self-learning; raw values ​​and moving average values ​​of gas leakage assessment coefficients; threshold arrays and risk level determination results for risk classification; and early warning and handling results data such as early warning measure trigger records, on-site handling results, and emergency repair execution status. All stored data carries a unified timestamp and region identifier, with the timestamp accurate to the second and the region identifier serving as a unique identifier for each monitoring sub-region. A tree-structured association index is established based on the timestamp and region identifier. This index enables rapid querying and retrieval of the entire detection and early warning process data for any time interval and any monitoring sub-region, allowing for historical retrospective analysis of the entire gas leak detection and early warning process. Through full data storage and traceability, comprehensive and accurate historical data support is provided for the daily maintenance of gas pipeline networks, the analysis of the causes of leak accidents, and the continuous optimization of early warning models and weight adjustment models. Simultaneously, it provides data evidence for tracing responsibility in gas safety supervision.

[0128] The intelligent gas leak detection and early warning method in this embodiment divides the monitoring sub-areas by combining the layout and usage scenarios of the target gas pipeline, extracts the operating characteristics of each area and defines the operating condition type, and adjusts the data acquisition parameters in a targeted manner according to the operating condition characteristics, making data acquisition more targeted, effectively reducing the collection of invalid data, and improving the effectiveness of monitoring data. The initial stability index is corrected by a working condition characteristic correction model, eliminating monitoring errors caused by operating condition fluctuations, and making the stability index more reflective of the actual operating status of each area. By constructing a feedback loop of weight self-learning update, the index weights of each operating condition type are dynamically optimized according to the accuracy of the early warning and the frequency of false alarms, solving the risk assessment bias problem caused by fixed weights and simple linear superposition in existing technologies. A gas leak assessment coefficient is obtained through a weighted fusion algorithm and moving average processing, and differentiated risk classification thresholds are set according to the operating condition type, achieving accurate classification of leak risks. At the same time, targeted early warning measures are triggered according to different risk levels, making early warning response more scientific and timely. Full data storage and traceability provide data support for the continuous optimization of the entire detection and early warning system and gas safety management. In practical applications, this method can effectively improve the accuracy of gas leak risk assessment, effectively identify potential leak risks, reduce false alarm and false alarm rates, reduce the ineffective workload of gas safety supervision, and solve the technical problem of insufficient accuracy of gas leak early warning in existing technologies.

[0129] Example 2

[0130] Please see Figure 6 The present invention also provides a gas leak intelligent detection and early warning system for performing the above-described method, comprising:

[0131] The gas pipeline area division module is used to divide monitoring sub-areas according to pipeline layout and usage scenarios and assign unique identifiers;

[0132] The operating condition feature extraction module is connected to the gas pipeline area division module and is used to collect real-time operation correlation data of each monitoring sub-area and extract operating condition features to define the operating condition type.

[0133] The multi-dimensional information acquisition module is used to adjust the acquisition frequency and accuracy according to the characteristics of the working conditions and to collect multi-dimensional basic data of each monitoring sub-area;

[0134] The index dynamic correction module is connected to the working condition feature extraction module and the multi-dimensional information acquisition module. It is used to calculate the initial stability index and combine the working condition features to correct the deviation, so as to obtain the corrected stability index.

[0135] The working condition weight self-learning module is used to retrieve historical early warning and handling data and combine them with real-time working condition characteristics to complete the self-learning update of weights and output working condition adapted weights.

[0136] The weighted fusion evaluation module, connected to the index dynamic correction module and the working condition weight self-learning module, is used to calculate the gas leakage evaluation coefficient through the weighted fusion algorithm.

[0137] The graded early warning execution module, connected to the weighted fusion evaluation module, is used to perform risk level determination and trigger corresponding early warning measures, while feeding back the handling results to the working condition weight self-learning module.

[0138] The full data storage module is bidirectionally connected to each of the modules and is used to store the entire process data of detection and early warning and to provide historical data retrieval services for the calculation and processing of each module.

[0139] Specifically, in the actual operation of the system, the gas pipeline area division module first receives gas pipeline layout data and area usage scenario data uploaded by the gas operation platform. It then uses a built-in area division algorithm to divide the target area into monitoring sub-areas and assigns a unique identifier to each sub-area. The division results and identifier data are synchronously transmitted to the operating condition feature extraction module and the full data storage module for storage. The operating condition feature extraction module establishes a communication connection with the intelligent sensing devices deployed in each monitoring sub-area, collecting real-time operational correlation data for each sub-area. Simultaneously, it retrieves supplementary data such as pipeline archives and historical gas consumption from the full data storage module. Through the built-in feature extraction algorithm and operating condition classification model, it extracts the gas usage load change characteristics, pipeline aging characteristics, and surrounding environmental interference characteristics of each sub-area, and completes the definition of the operating condition type. The operating condition feature data and the operating condition type definition results are synchronously transmitted to the multi-dimensional information acquisition module, the index dynamic correction module, and the full data storage module.

[0140] The multi-dimensional information acquisition module, based on the received operating condition characteristic data and operating condition type definition results, uses a built-in parameter adjustment algorithm to specifically adjust the acquisition frequency and accuracy of the sensors in each monitoring sub-area. Simultaneously, it dynamically allocates the system's data transmission bandwidth. After adjustment, it collects multi-dimensional basic data such as gas alarm data, gas pipeline status data, and surrounding environmental data from each sub-area through the sensors. This collected multi-dimensional basic data is then transmitted to the index dynamic correction module and the full data storage module. The index dynamic correction module receives the basic data transmitted from the multi-dimensional information acquisition module and the operating condition characteristic data transmitted from the operating condition feature extraction module. First, it calculates the initial gas monitoring stability index, gas pipeline status stability index, and environmental status stability index using a built-in index calculation algorithm. Then, it calculates the deviation correction factor using a built-in operating condition feature correction model to correct the initial stability index, obtaining the corrected stability index. The stability index data before and after correction are then transmitted to the weighted fusion evaluation module and the full data storage module.

[0141] The operating condition weight self-learning module retrieves historical early warning data and on-site handling results for each monitoring sub-area from the full data storage module. Combined with real-time operating condition features transmitted by the operating condition feature extraction module, it constructs a weight adjustment matrix using a built-in weight adjustment algorithm and completes the self-learning update of the initial weights to obtain the operating condition adaptation weights for each operating condition type. These operating condition adaptation weights are then transmitted to the weighted fusion evaluation module and the full data storage module. The weighted fusion evaluation module receives the corrected stability index transmitted by the index dynamic correction module and the operating condition adaptation weights transmitted by the operating condition weight self-learning module. Using a built-in weighted fusion algorithm and a moving average algorithm, it calculates the gas leakage assessment coefficient for each monitoring sub-area. The assessment coefficient data is then transmitted to the graded early warning execution module and the full data storage module.

[0142] The tiered early warning execution module retrieves the leakage risk grading threshold library for each operating condition type from the full data storage module. Combined with the received gas leakage assessment coefficients, it determines the leakage risk level of each monitored sub-area using a built-in risk grading algorithm. Based on the risk level, it triggers corresponding early warning measures. The triggering records of these measures and the on-site feedback of the handling results are synchronously transmitted to the operating condition weight self-learning module and the full data storage module, providing a basis for subsequent self-learning updates of the operating condition weights. The full data storage module maintains bidirectional communication with all modules in the system. On the one hand, it stores the entire detection and early warning process data transmitted by each module in real time. On the other hand, it provides historical data retrieval and query services for each module according to their computational needs. Simultaneously, through a built-in index building algorithm, it establishes a correlation index based on timestamps and area identifiers, supporting historical backtracking analysis of the entire process data.

[0143] This intelligent gas leak detection and early warning system achieves full-process intelligent operation of gas leak detection and early warning, from regional division, data collection, index calculation, weight optimization, risk assessment to early warning handling, through the coordinated operation of various modules. The functions of each module are interconnected and the data is integrated to form a closed-loop intelligent detection and early warning system.

[0144] Example 3

[0145] To enable those skilled in the art to fully understand and implement this invention, the specific implementation principle of this invention is further explained below in conjunction with a specific application scenario.

[0146] This embodiment selects a gas transmission pipeline network area in the core urban area as the application verification scenario. This area covers approximately 5 square kilometers and includes old residential areas, newly built residential areas, commercial complex areas, small industrial workshop areas, and pipeline areas near urban construction sections. It is a gas pipeline usage scenario with multiple operating conditions. The gas pipeline diameter spans a large range, the transmission pressure varies, and the operating conditions of different sub-areas differ significantly, making it a typical verification representative. In this verification process, the above-mentioned intelligent gas leak detection and early warning method and system are applied to this area, and implemented step by step according to the method's execution flow to verify the actual applicability and effectiveness of the method and system.

[0147] First, the gas pipeline layout data of the region and the usage scenario data of each area are imported into the gas pipeline area division module of the intelligent gas leak detection and early warning system. Based on the pipe diameter, transmission pressure, pipe segment connection method, and functional attributes of each area, the module divides the region into 28 independent monitoring sub-regions and assigns a unique identifier containing geographical information and pipe segment number to each sub-region. After the area division is completed, the division results are stored in the full data storage module. Subsequently, the system's operating condition feature extraction module collects real-time operational data from sensors in each monitoring sub-area, while also retrieving supplementary data such as pipeline archives, historical gas usage, and regional construction reports. This extracts the gas load variation characteristics, pipeline aging characteristics, and surrounding environmental interference characteristics of each sub-area. Using an operating condition classification model, the 28 monitoring sub-areas are categorized into 6 high-load fluctuation types, 7 severely aging pipeline types, 5 complex environmental interference types, and 10 normal operation types. Specifically, 6 sub-areas in concentrated commercial complex areas are categorized as high-load fluctuation types, 7 sub-areas in old residential areas are categorized as severely aging pipeline types, 5 sub-areas near construction sites are categorized as complex environmental interference types, and 10 sub-areas in newly built residential areas are categorized as normal operation types. The operating condition feature and type classification results are simultaneously transmitted to the relevant modules.

[0148] The system's multi-dimensional information acquisition module dynamically adjusts the acquisition parameters of the sensing devices and the system transmission bandwidth according to the operating conditions of each sub-region. For six high-load fluctuation sub-regions, the sampling frequency of flow and pressure sensors is increased, enhancing the accuracy of flow filtering. For seven sub-regions with severe pipeline aging, the acquisition accuracy of acoustic wave and pipeline deformation data is improved, extending the online duration of acoustic sensors. For five sub-regions with complex environmental interference, the number of acquisition points for temperature, humidity, and acceleration sensors is increased, utilizing redundant data to eliminate noise interference. 40% of the system's high-priority bandwidth is allocated to the aforementioned 18 sub-regions with complex operating conditions, with the remaining bandwidth allocated to the 10 normally operating sub-regions. After parameter adjustments, gas alarm data, pipeline status data, and surrounding environmental data from each sub-region are collected, and the multi-dimensional basic data is transmitted to the exponential dynamic correction module.

[0149] The index dynamic correction module normalizes the collected basic data and calculates the initial gas monitoring stability index, pipeline condition stability index, and environmental condition stability index for each sub-region. Then, combined with the operating condition characteristic data, the deviation correction factor for each sub-region is calculated through the operating condition characteristic correction model to correct the deviation of the initial stability index and obtain the corrected stability index. Among them, the pipeline aging correction factor for the sub-region with severe pipeline aging is between 1.2 and 1.5, which effectively improves the risk identification sensitivity of this type of area. The environmental interference correction factor for the sub-region with complex environmental interference is between 0.7 and 0.9, which effectively filters out sensor data drift caused by construction vibration. After the corrected stability index is transmitted to the weighted fusion evaluation module, the system's operating condition weight self-learning module retrieves the historical early warning data and handling results of the region for the past three years from the full data storage module. Combining this with the real-time operating condition characteristics of each sub-region, a weight adjustment matrix is ​​constructed. The sensitivity and false alarm frequency of each stability index to the accuracy of early warnings are calculated, and the initial weights are updated through self-learning to obtain the operating condition adaptation weights for each operating condition type. Specifically, the pipeline stability index weight for the severely aging pipeline sub-region is adjusted from the initial 0.5 to 0.62, the gas monitoring stability index weight for the high load fluctuation sub-region is adjusted from the initial 0.45 to 0.53, the environmental stability index weight for the complex and disturbed environment sub-region is adjusted from the initial 0.2 to 0.28, and the weight for the normal operation sub-region remains slightly adjusted. After the weight adjustment, normalization is performed to obtain the final operating condition adaptation weights.

[0150] The weighted fusion assessment module maps the corrected stability index to a unified dimension space to obtain standardized components. It then retrieves the corresponding operating condition adaptation weights for each sub-region, performs weighted summation to obtain the original assessment coefficients, and uses a sliding window of 8 data points for time-series moving average processing to eliminate the impact of instantaneous data fluctuations, resulting in gas leakage assessment coefficients for 28 monitoring sub-regions. The system's graded early warning execution module retrieves the leakage risk grading threshold array for each operating condition type from the threshold library, compares the assessment coefficients of each sub-region with the corresponding thresholds, and determines the risk level. Two sub-regions with severe pipeline aging are classified as medium risk, three sub-regions with high load fluctuations and two sub-regions with complex environmental interference are classified as low risk, and the remaining 21 sub-regions are classified as risk-free. No sub-regions were classified as high risk.

[0151] Based on the determined risk level, the system triggers corresponding early warning measures. For the two medium-risk sub-areas with severe pipeline aging, on-site audible and visual alarm terminals are immediately triggered, pushing abnormal data and real-time location to the mobile terminals of supervisors. At the same time, an encrypted monitoring mode is activated, increasing the data collection frequency by 50%. For the seven low-risk sub-areas, system log recording is performed, and personalized routine inspection plan suggestions are generated based on the operating conditions of each sub-area. For the three low-risk sub-areas with high load fluctuations, it is recommended to increase the frequency of flow and pressure inspections. For the two low-risk sub-areas with complex environmental interference, it is recommended to increase the linkage frequency of environmental monitoring and pipeline inspections. For the 21 risk-free sub-areas, only routine system data recording and status monitoring are performed. All trigger records of early warning measures are synchronously transmitted to the full data storage module. Based on the medium-risk early warning information pushed by the system, the supervisors went to the site for verification and found that the pipelines in the two medium-risk sub-areas had slight interface corrosion and slight pressure fluctuations, which posed a potential leakage risk. They then took anti-corrosion treatment and interface reinforcement maintenance measures according to the system's recommendations, which eliminated the safety hazards in a timely manner. After the maintenance and handling results were fed back to the system, they were transmitted to the operating condition weight self-learning module, which provided on-site data support for subsequent weight updates.

[0152] Throughout the application verification process, the system synchronously executes full data storage and traceability steps, storing all process data, including monitoring sub-area division data, operating condition characteristic data, multi-dimensional collected data, stability index before and after correction, weight self-learning process data, gas leak assessment coefficient, risk classification results, early warning measure trigger records, and on-site handling results, into a distributed cloud database. A correlation index based on timestamps and area identifiers is established, allowing regulatory personnel to quickly retrieve detection and early warning data for any sub-area and any time interval through the system's query function, thus realizing full-process historical retrospective analysis.

[0153] During this application verification process, the intelligent gas leak detection and early warning method and system were able to accurately divide the monitoring sub-areas and define the operating conditions based on the multi-condition characteristics of the verification scenario, and adjust the data acquisition parameters accordingly, effectively improving the effectiveness of the monitoring data. Through stability index deviation correction and weight self-learning update, it achieved accurate assessment of the leakage risk of each sub-area, accurately identifying medium-risk areas with potential leakage risks, without any false alarms or missed alarms. The early warning measures triggered by different risk levels are targeted and timely, providing scientific guidance for the maintenance and emergency response of gas pipeline networks. The full data storage and traceability provide comprehensive data support for the subsequent gas safety management of the area. The entire verification process shows that the method and system can adapt to complex multi-condition gas pipeline usage scenarios, effectively solving the problem of insufficient accuracy in gas leak early warning in existing technologies. It has good practical applicability and effectiveness in intelligent detection and accurate early warning of gas leaks, and can provide reliable technical support for gas safety management.

[0154] The intelligent gas leak detection and early warning method and system of this invention overcomes the limitations of fixed weights and simple index superposition in existing technologies. It integrates operating condition characteristics into the entire detection and early warning process, achieving targeted data collection, accurate index calculation, precise risk assessment, and scientific early warning response. This effectively improves the ability to identify gas leak risks, reduces false alarm and false alarm rates, and reduces the ineffective workload of gas safety supervision. At the same time, the modular design of the system and the process-oriented implementation of the method allow it to adapt to gas pipeline monitoring scenarios of different scales and operating conditions, possessing broad application prospects and significant practical value in the field of gas leak detection technology.

[0155] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0156] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for intelligent detection and early warning of gas leaks, characterized in that, Includes the following steps: S1. Based on the pipeline layout and usage scenario of the target gas pipeline, complete the division of the target area into monitoring sub-areas and assign a unique identifier to each monitoring sub-area. S2. Collect real-time operation correlation data of each monitoring sub-area, extract the operating condition characteristics of each monitoring sub-area, and define the operating condition type of each monitoring sub-area based on the operating condition characteristics. The operating condition characteristics of each monitoring sub-region are extracted in S2, including: Extract the characteristics of gas usage load variation, which include the average daily peak gas consumption, the standard deviation of gas consumption fluctuation, and the frequency of instantaneous flow rate mutation. Extract pipeline aging characteristics, which include pipeline years of operation, material corrosion rate, mechanical stress distribution at interface connections, and integrity data of anti-corrosion coating. Extract the surrounding environmental interference features, which include the temperature and humidity cycle fluctuation amplitude of the monitoring area, the distribution of ground vibration frequency, the frequency of surrounding construction, and the intensity of interference from third-party activities; The definition of the operating condition type includes dividing the monitoring sub-area into high load fluctuation type, severe pipeline aging type, complex environmental interference type, and normal operation type. S3. Adjust the frequency and accuracy of data acquisition according to the operating characteristics and types of each monitoring sub-area, and collect gas alarm data, gas pipeline status data and surrounding environment data of each monitoring sub-area; S4. Calculate the initial gas monitoring stability index, gas pipeline status stability index, and environmental status stability index based on the gas alarm data, gas pipeline status data, and surrounding environment data. Then, perform deviation correction on the gas monitoring stability index, gas pipeline status stability index, and environmental status stability index in combination with the operating condition characteristics to obtain the corrected stability index. S4 involves deviation corrections for the gas monitoring stability index, gas pipeline condition stability index, and environmental condition stability index, including: A working condition characteristic correction model is established, and the deviation correction factor for each monitoring sub-region is calculated using the working condition characteristics as input parameters. The gas load correction factor is used to compensate the gas monitoring stability index to offset the interference of sudden flow changes caused by normal peak gas consumption. The gas pipeline stability index is augmented using a pipeline aging correction factor, and the risk weight of the aging pipeline under minute pressure changes is adjusted by the gain coefficient. The environmental stability index is denoised by using an environmental interference correction factor to filter sensor zero-point drift data caused by construction vibration or extreme weather, thereby outputting the corrected stability index after eliminating the influence of operating condition fluctuations. S5. Retrieve historical early warning data and handling results, combine them with the real-time operating conditions of each monitoring sub-area, optimize and adjust the pre-stored initial index weights of each operating condition type, complete the self-learning update of the weights, and obtain the operating condition adaptation weights. The self-learning update of the weights in S5 includes: Construct a weight adjustment matrix with each working condition type as the dimension, and set the initial weights as the initial values ​​of the matrix; The system receives real-time feedback on the on-site handling results of the early warning, uses the accuracy of the early warning as the objective function, and calculates the sensitivity of the three types of corrected stability indices to the accuracy of the early warning under each working condition. If the sensitivity of a certain stability index under the corresponding working condition is higher than a preset threshold, then the value of the stability index in the weight adjustment matrix is ​​increased. If historical early warning data shows that a certain stability index causes a false alarm frequency higher than a preset false alarm frequency threshold under the corresponding working condition, then the weight ratio of the stability index in the weight adjustment matrix is ​​reduced, and the dynamic evolution of the weight under each working condition is realized through feedback closed loop. S6. Perform weighted fusion calculation on the three types of modified stability indices according to the weighted fusion algorithm to obtain the gas leakage assessment coefficient of each monitoring sub-region; The gas leakage assessment coefficients for each monitoring sub-area obtained in S6 include: The modified stability index is mapped to a unified dimension space to obtain standardized gas monitoring components, pipeline state components, and environmental state components. Call the corresponding working condition adaptation weights for the working condition type and perform a weighted summation calculation; The calculation results are processed by time series moving average to eliminate the influence of instantaneous data jitter, and a gas leakage assessment coefficient reflecting the actual leakage risk trend of the monitored sub-area is obtained. S7. Retrieve the pre-stored leakage risk classification thresholds for each operating condition type, compare the gas leakage assessment coefficient of each monitoring sub-area with the corresponding threshold, determine the leakage risk level of each monitoring sub-area, and trigger early warning measures according to the risk level.

2. The intelligent detection and early warning method for gas leaks according to claim 1, characterized in that, In step S3, the frequency and accuracy of data acquisition are adjusted according to the operating characteristics and types of each monitoring sub-region, including: For monitoring sub-areas defined as high-load fluctuation type, increase the sampling frequency of gas flow sensor and pressure sensor, and increase the filtering accuracy of flow fluctuation; For monitoring sub-areas defined as severely aged pipelines, improve the acquisition accuracy of acoustic monitoring data and pipeline deformation data, and extend the online duration of acoustic sensors; For monitoring sub-areas defined as having complex and interfering environments, the number of data collection points for temperature and humidity sensors and acceleration sensors is increased, and redundant data from multiple points is used to eliminate environmental noise interference. The data transmission bandwidth allocation is dynamically adjusted in real time according to the operating conditions of each monitoring sub-area, and the bandwidth corresponding to the preset high-priority bandwidth threshold is allocated to the monitoring sub-area with complex operating conditions.

3. The intelligent detection and early warning method for gas leaks according to claim 1, characterized in that, The S7 step of determining the leakage risk level of each monitoring sub-area includes: The leakage risk level is divided into four levels: no risk, low risk, medium risk, and high risk. The matching hierarchical threshold array is retrieved from the threshold library based on the operating condition type. When the gas leak assessment coefficient is within the first threshold range, it is determined to be low risk; When the gas leak assessment coefficient is within the second threshold range, it is determined to be of medium risk; When the gas leak assessment coefficient exceeds the third threshold, it is determined to be high risk; Specifically, for monitoring sub-areas with severe pipeline aging, the threshold settings for entering the medium-risk and high-risk levels have been lowered.

4. The intelligent detection and early warning method for gas leaks according to claim 3, characterized in that, The triggering logic for the early warning measures includes: For low-risk monitoring sub-areas, system log recording is performed and routine inspection plan suggestions are generated; For monitoring sub-areas at medium risk level, trigger the audible and visual alarm terminal and push abnormal data and real-time location to the preset mobile terminal of the supervisor, and start the encrypted monitoring mode. For high-risk monitoring sub-areas, the emergency shut-off valve of the corresponding pipeline section is immediately triggered, and the early warning information is simultaneously pushed to the emergency response command system and regulatory departments at all levels. The optimal repair route is also planned based on the relevant on-site data.

5. The intelligent detection and early warning method for gas leaks according to claim 1, characterized in that, The method also includes a full data storage and traceability step: The system stores in real time the monitoring sub-area division data, operating condition characteristic data, multi-dimensional collected basic information, stability index before and after correction, weight self-learning process data, gas leakage assessment coefficient, risk classification results, and early warning and handling results. Establish a correlation index between data items based on timestamps and regional identifiers to realize historical retrospective analysis of the entire gas leak detection and early warning process.

6. A gas leak intelligent detection and early warning system, used to execute the gas leak intelligent detection and early warning method according to any one of claims 1 to 5, characterized in that, include: The gas pipeline area division module is used to divide monitoring sub-areas according to pipeline layout and usage scenarios and assign unique identifiers; The operating condition feature extraction module is connected to the gas pipeline area division module and is used to collect real-time operation correlation data of each monitoring sub-area and extract operating condition features to define the operating condition type. The multi-dimensional information acquisition module is used to adjust the acquisition frequency and accuracy according to the characteristics of the working conditions and to collect multi-dimensional basic data of each monitoring sub-area; The index dynamic correction module is connected to the working condition feature extraction module and the multi-dimensional information acquisition module. It is used to calculate the initial stability index and combine the working condition features to correct the deviation, so as to obtain the corrected stability index. The working condition weight self-learning module is used to retrieve historical early warning and handling data and combine them with real-time working condition characteristics to complete the self-learning update of weights and output working condition adapted weights. The weighted fusion evaluation module, connected to the index dynamic correction module and the working condition weight self-learning module, is used to calculate the gas leakage evaluation coefficient through the weighted fusion algorithm. The graded early warning execution module, connected to the weighted fusion evaluation module, is used to perform risk level determination and trigger corresponding early warning measures, while feeding back the handling results to the working condition weight self-learning module. The full data storage module is bidirectionally connected to the gas pipeline area division module, operating condition feature extraction module, multi-dimensional information acquisition module, index dynamic correction module, operating condition weight self-learning module, weighted fusion evaluation module, and graded early warning execution module. It is used to store the entire process data of detection and early warning and provide historical data retrieval services for the operation and processing of each module.

Images