Natural gas pipeline leakage detection method based on edge calculation

Abstract

The invention discloses a natural gas pipeline leakage detection method based on edge calculation, relates to the technical field of natural gas pipeline leakage detection, and remarkably improves the stability of edge side leakage detection through double innovations of dynamic multi-scale perception and space-time cooperative verification: a dynamic window mechanism breaks through the limitation of a fixed analysis window; through dual triggering of the energy accumulation rate and the pulse density, feature adaptability extraction of leakage events with different intensities is realized, and the missing report rate of tiny leakage is reduced; spatio-temporal fingerprint verification is combined with matching of pipeline stress wave propagation characteristics and sensor types, a physical conduction chain credibility evaluation system is constructed, and false alarm diffusion caused by regional interference is effectively inhibited; the environment entropy dynamic correction and the sensitivity strategy pool form closed-loop optimization, so that the system keeps stable detection performance under complex working conditions such as construction vibration and meteorological sudden change.

Description

Technical Field

[0001] The present invention relates to the technical field of natural gas pipeline leakage detection, in particular to a natural gas pipeline leakage detection method based on edge computing. Background Art

[0002] Currently, the natural gas pipeline leakage monitoring system is accelerating its evolution towards edge intelligence, relying on computing nodes deployed along the pipeline to achieve local real-time detection; these edge devices integrate multi-dimensional sensors such as pressure, gas concentration, and vibration, and capture leakage characteristics by continuously collecting physical signals; compared with the traditional centralized monitoring mode, edge computing significantly reduces data transmission latency and shows significant advantages in long-distance pipelines and harsh outdoor environments; the mainstream industry solutions generally adopt lightweight AI models for signal feature extraction and combine threshold triggering mechanisms to achieve rapid response, forming a closed-loop link of "perception - judgment - execution".

[0003] The existing solutions face two core bottlenecks: First, the lack of scale adaptability in signal recognition; leakage events include two types of characteristics, slow changes such as concentration drift caused by small leaks and sudden changes such as vibration shocks caused by pipe bursts, and a fixed analysis window is difficult to balance both - an overly large window smooths high-frequency details and leads to missed alarms, while an overly small window amplifies low-frequency noise and causes false alarms; Second, the lack of credibility collaboration between edge nodes; when each node makes independent decisions, it is easily affected by regional interferences such as construction vibrations and meteorological mutations, triggering chain false actions; although some studies have tried to introduce a weighted voting mechanism, they do not consider the spatio-temporal propagation characteristics of environmental interferences, resulting in the spread of false signals within the cluster.

[0004] Recent technical improvements focus on multi-scale signal processing and cluster linkage: Some solutions adopt a dual-window convolution algorithm to capture high-frequency vibrations and low-frequency concentration changes respectively, and then improve the recognition rate through feature fusion; other studies construct a lightweight neural network to output a local event credibility score (LECS) and correct the judgment based on voting among adjacent nodes; however, the former depends on a preset window size and is difficult to cope with sudden signal intensity fluctuations; the latter uses static weight allocation and cannot dynamically filter out false signals generated by continuous interferences such as construction. Summary of the Invention

[0005] In view of the above existing problems, the present invention is proposed.

[0006] The present invention provides a natural gas pipeline leakage detection method based on edge computing to solve the problems that existing edge detection methods are unable to adapt to the multi-scale characteristics of leakage signals due to a fixed analysis window, and node independent decision-making is easily affected by regional interferences, resulting in frequent alternation of false alarms and missed alarms.

[0007] To solve the above technical problems, the present invention provides the following technical solutions: An embodiment of the present invention provides a method for detecting natural gas pipeline leakage based on edge computing, which includes: Step S1, synchronously collect pipeline pressure, gas concentration, and vibration signals at each edge node, and perform multi-scale feature extraction on the signals through a dynamic window generation module; Step S2, input the extracted features into a lightweight neural network model to generate a local event credibility score, where the credibility score includes double evaluation results of signal anomaly intensity and environmental interference probability; Step S3, periodically exchange credibility data packets containing spatio-temporal fingerprints between adjacent edge nodes, and verify the physical consistency of the event propagation path based on the characteristics of pipeline stress wave conduction; Step S4, dynamically adjust the local detection sensitivity strategy according to the credibility data of neighboring nodes and the environmental entropy value, and trigger a leakage warning or enter a delay confirmation mode.

[0008] As a preferred solution of the method for detecting natural gas pipeline leakage based on edge computing according to the present invention, wherein: the dynamic window generation module performs the following operations: Trigger window scale switching according to the real-time signal energy accumulation rate. When the low-frequency signal energy accumulation reaches the first threshold, enable a large-scale window. When the high-frequency pulse signal density exceeds the second threshold, enable a microsecond-level sampling window; In a mixed event scenario, run dual-scale windows synchronously and output a fused feature vector; The triggering condition of the first threshold is: the low-frequency signal energy change rate continuously remains lower than the set slope within N consecutive sampling periods, and the baseline drift amount exceeds the threshold of the pipeline pressure rated value.

[0009] As a preferred solution of the method for detecting natural gas pipeline leakage based on edge computing according to the present invention, wherein: the multi-scale feature extraction includes: Perform baseline drift trend fitting on the low-frequency signal to generate a first feature group including the drift slope and the periodic fluctuation amplitude; Extract the peak interval, duration, and energy decay rate of the pulse cluster from the high-frequency signal to generate a second feature group; Align the first feature group and the second feature group in time and splice them into a joint feature vector.

[0010] As a preferred solution of the method for detecting natural gas pipeline leakage based on edge computing according to the present invention, wherein: in step S1, during the process of performing multi-scale feature extraction on the signal, dynamic window triggering and parallel output are performed. The dynamic window generation module calculates the energy and pulse density of the low-frequency and high-frequency signals in two windows in real time, and switches or runs in parallel according to the triggering conditions: Perform low-frequency energy rate calculation: , When consecutive cycles meet , that is, the large-scale window is enabled , to calculate the high-frequency pulse density: , When , the micro-scale window is enabled , wherein, represents the low-frequency signal of the th sampling point, represents the cumulative energy within the large-scale window, represents the energy accumulation rate, represents the sampling period, represents the length of the large-scale window, represents the number of sampling periods for continuous judgment, represents the rate threshold, represents the th cycle's pressure baseline drift amount, represents the pressure offset threshold, represents the number of pulse clusters detected within the micro-scale window, represents the length of the micro-scale window, represents the pulse density threshold; If both sets of the above conditions are met, the dual-scale windows run in parallel, and two original signal segments are output for subsequent feature extraction.

[0011] As a preferred solution of the natural gas pipeline leakage detection method based on edge computing described in the present invention, wherein: in step S1, during the process of performing multi-scale feature extraction on the signal, for the low-frequency signal segment within the large-scale window perform least squares linear fitting to obtain the drift slope and the fluctuation amplitude: , , wherein, , , , represents the baseline drift slope, represents the periodic fluctuation amplitude; Detect pulse clusters for the high-frequency signal segment within the micro-scale window , and record the The peak time of the cluster is , and the energy within the cluster is: , Cluster duration , peak interval , and perform exponential fitting on the energy decay: , where represents the high-frequency signal samples, are respectively the cluster start and end sampling indices, represents the cluster energy, represents the cluster duration, represents the peak interval between adjacent clusters, represents the number of clusters taken, represents the energy decay rate; Align and splice the low-frequency feature with the high-frequency feature , in chronological order to obtain the combined feature vector: .

[0012] As a preferred embodiment of the natural gas pipeline leakage detection method based on edge computing according to the present invention, wherein: the generation of the local event credibility score includes: Output the original anomaly probability value through the neural network; Superimpose the historical false alarm rate correction coefficient of the current node and the real-time intensity value of the environmental interference source to generate the final credibility score; The environmental interference source includes the wind speed sensor reading and the surface vibration sensor reading; The historical false alarm rate correction coefficient is obtained by calculating through a sliding window: count the false alarm ratio confirmed manually in the warning events triggered by the current node in the past period of time, and generate the correction coefficient according to 1 - false alarm ratio.

[0013] As a preferred embodiment of the natural gas pipeline leakage detection method based on edge computing according to the present invention, wherein: the spatio-temporal fingerprint includes the first peak timestamp of the vibration signal, the sensor type identifier, and the signal propagation direction encoding; The spatio-temporal fingerprint verification includes: When the node detects an abnormal vibration signal, record the arrival timestamp of the first peak; Receive the peak timestamp data of adjacent nodes, calculate the time difference and perform a matching degree check with the stress wave propagation speed corresponding to the pipeline material; If the time difference matching degree exceeds the set ratio, it is marked as a physically conductive valid event.

[0014] As a preferred solution of the natural gas pipeline leakage detection method based on edge computing described in the present invention, wherein: in step S3, during the process of performing matching degree verification, adjacent nodes are received and 's spatio-temporal fingerprints, and in combination with the pipeline stress wave propagation characteristics, calculate the matching degree and perform verification according to the following process: Calculate the time difference and the theoretical conduction time: , wherein, represents the arrival time stamp of the first wave peak of node , represents the distance between nodes measured along the pipeline center line between nodes, represents the stress wave propagation speed corresponding to the pipeline material, Calculate the time matching degree : , wherein, represents the time error, represents the maximum allowable time deviation; Determine the type and the direction matching degree : If , then , otherwise ; , wherein, are respectively the sensor type identifiers of two nodes, are respectively the signal propagation direction coding angles; Define the total matching degree and the verification standard as: , Verification standard: , wherein, is the weight coefficient, is the high matching degree threshold, is the low matching degree threshold.

[0015] As a preferred solution of the natural gas pipeline leakage detection method based on edge computing described in the present invention, wherein: the calculation of the environmental entropy value includes: Collect the wind speed, temperature gradient and mechanical vibration signals around the current node; The standard deviation ratio of each interference source is calculated using a sliding window to generate an environment disorder index normalized to 0-1. The environment disorder index is weighted and averaged with the same type of data of neighboring nodes to obtain the final environment entropy value.

[0016] As a preferred solution of the natural gas pipeline leakage detection method based on edge computing according to the present invention, wherein: the dynamic adjustment of the sensitivity strategy includes: A variety of sensitivity strategy templates are preset, including a night static mode, a strong wind interference mode, and a construction protection mode; According to the detection accuracy rate in the recent 24 hours and the change trend of the environment entropy, the current optimal strategy is automatically selected or mixedly generated; When the credibility scores of more than half of the neighboring nodes continuously fall below the warning value, the high-sensitivity inspection mode is forcibly switched; The calculation method of the detection accuracy rate is: a normal distribution model is established for the time difference between the first report time of this node and the actual occurrence time of the event in the confirmed leakage events, and the distribution function value is taken as the accuracy rate index.

[0017] The beneficial effects of the present invention are: through the dual innovations of dynamic multi-scale perception and spatio-temporal collaborative verification, the stability of leakage detection on the edge side is significantly improved: the dynamic window mechanism breaks through the limitation of the fixed analysis window, and through the dual triggering of the energy accumulation rate and the pulse density, the characteristic adaptive extraction of leakage events of different intensities is realized, reducing the missed report rate of minor leaks; the spatio-temporal fingerprint verification combines the pipeline stress wave propagation characteristics and the sensor type matching to construct a physical conduction chain credibility evaluation system, effectively suppressing the false alarm diffusion caused by regional interference; the dynamic correction of the environment entropy and the sensitivity strategy pool form a closed-loop optimization, enabling the system to maintain stable detection performance under complex working conditions, such as construction vibration and meteorological mutation. Description of the Drawings

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention, and those of ordinary skill in the art can obtain other drawings according to these drawings without creative efforts.

[0019] Figure 1 It is a schematic flow chart of the natural gas pipeline leakage detection method based on edge computing in Embodiment 1. Detailed Embodiments

[0020] In order to make the above objects, features, and advantages of the present invention more obvious and understandable, the detailed embodiments of the present invention will be described below with reference to the drawings of the specification.

[0021] In the following description, many specific details are set forth in order to provide a thorough understanding of the present invention. However, the present invention may be practiced in other ways than those specifically described herein, and those skilled in the art may make similar extensions without departing from the spirit of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0022] Secondly, as used herein, an "embodiment" or "embodiments" refers to specific features, structures, or characteristics that may be included in at least one implementation of the present invention. The phrase "in an embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment that is mutually exclusive of other embodiments.

[0023] Embodiment 1, referring to Figure 1 , this embodiment provides a method for detecting natural gas pipeline leakage based on edge computing, including the following steps: Step S1, synchronously collect pipeline pressure, gas concentration, and vibration signals at each edge node, and perform multi-scale feature extraction on the signals through a dynamic window generation module; The dynamic window generation module performs the following operations: Trigger window scale switching according to the real-time signal energy accumulation rate. When the low-frequency signal energy accumulation reaches the first threshold, enable a large-scale window. When the high-frequency pulse signal density exceeds the second threshold, enable a microsecond-level sampling window; In a mixed event scenario, run dual-scale windows synchronously and output a fused feature vector; The trigger condition for the first threshold is: the low-frequency signal energy change rate continuously remains below the set slope within N consecutive sampling periods, and the baseline drift amount exceeds the threshold of the pipeline pressure rated value; The multi-scale feature extraction includes: Perform baseline drift trend fitting on the low-frequency signal to generate a first feature group including the drift slope and the periodic fluctuation amplitude; Extract the peak interval, duration, and energy decay rate of the pulse cluster from the high-frequency signal to generate a second feature group; Align the first feature group and the second feature group in time and splice them into a joint feature vector; In step S1, during the process of performing multi-scale feature extraction on the signal, dynamic window triggering and parallel output are performed. The dynamic window generation module calculates the energy and pulse density of the low-frequency and high-frequency signals in two windows in real time, and switches or runs in parallel according to the trigger conditions: Perform low-frequency energy rate calculation: , When continuous cycles satisfy , Immediately enable the large-scale window , perform high-frequency pulse density calculation: , When enable the micro-scale window , wherein, represents the low-frequency signal of the th sampling point, represents the cumulative energy within the large-scale window, represents the energy accumulation rate, represents the sampling period, represents the large-scale window length, represents the number of sampling periods for continuous judgment, represents the rate threshold, represents the period pressure baseline drift amount of the th period, represents the pressure offset threshold, represents the number of pulse clusters detected within the micro-scale window, represents the micro-scale window length, represents the pulse density threshold; If the above two sets of conditions are simultaneously satisfied, the dual-scale window runs in parallel, and two original signal segments are output for subsequent feature extraction; In step S1, during the process of multi-scale feature extraction of the signal, perform least-squares linear fitting on the low-frequency signal segment within the large-scale window , , wherein, , , , represents the baseline drift slope, represents the periodic fluctuation amplitude; Detect pulse clusters in the high-frequency signal segment within the micro-scale window , and record the peak time of the th cluster as and the energy within the cluster as: , the cluster duration , the peak interval , and perform exponential fitting on the energy decay: , Among them, represents the high-frequency signal samples, are respectively the start and end sampling indices of the cluster, represents the cluster energy, represents the cluster duration, represents the peak interval between adjacent clusters, represents the number of clusters taken, represents the energy decay rate; The low-frequency features and high-frequency features , are aligned and concatenated in chronological order to obtain a joint feature vector: ; Specifically, through adaptive window switching and parallel operation, multi-scale and composite scenario perception of signals in natural gas pipelines is achieved. The large-scale window targets low-frequency drift, and uses least squares fitting to extract the stable trend slope and periodic fluctuation amplitude, which can effectively filter short-term noise. The micro-scale window targets high-frequency pulses, calculates the peak interval, duration, and exponential decay rate, and captures the transient leakage impact characteristics. By fusing the two-way features, both long-term trend information and instantaneous event details are retained, improving the discrimination ability of the downstream classifier for leakage events. The innovation lies in continuous multi-period rate determination triggering and quadratic fitting of the exponential decay rate, which not only ensures the stability of the features but also enhances the recognition diversity for different leakage intensities and patterns; Step S2: Input the extracted features into a lightweight neural network model to generate a local event credibility score, which includes the dual evaluation results of signal anomaly intensity and environmental interference probability; The generation of the local event credibility score includes: Output the original anomaly probability value through the neural network; Overlay the historical false alarm rate correction coefficient of the current node and the real-time intensity value of the environmental interference source to generate the final credibility score; The environmental interference source includes the readings of the wind speed sensor and the ground vibration sensor; The historical false alarm rate correction coefficient is obtained by calculating through a sliding window: Statistics the false alarm ratio confirmed manually among the warning events triggered in the past period of time at the current node, and generate the correction coefficient according to 1 - false alarm ratio;

[0024] Step S3: Periodically exchange credibility data packets containing spatio-temporal fingerprints between adjacent edge nodes, and verify the physical consistency of the event propagation path based on the characteristics of pipeline stress wave conduction; The spatio-temporal fingerprint includes the timestamp of the first peak of the vibration signal, the sensor type identifier, and the signal propagation direction encoding; The spatio-temporal fingerprint verification includes: When a node detects an abnormal vibration signal, record the timestamp of the arrival of the first peak; Receive the peak timestamp data of adjacent nodes, calculate the time difference, and perform a matching degree check with the stress wave propagation speed corresponding to the pipeline material; If the time difference matching degree exceeds the set ratio, mark it as a physically conductive valid event.

[0025] In step S3, during the matching degree check, receive adjacent nodes and After the spatio-temporal fingerprints of, combine with the stress wave propagation characteristics of the pipeline, and calculate the matching degree and perform a check according to the following process: Calculate the time difference and the theoretical conduction time: , where, represents the timestamp of the arrival of the first peak of node , represents the distance between nodes measured along the center line of the pipeline, represents the stress wave propagation speed corresponding to the pipeline material, Calculate the time matching degree : , where, represents the time error, represents the maximum allowable time deviation; Determine the type and the direction matching degree : If , then , otherwise ; , where, are the sensor type identifiers of the two nodes respectively, are the signal propagation direction encoding angles respectively; Define the total matching degree and the verification standard as: , Verification standard: , where, is the weight coefficient, is the high matching degree threshold, is the low matching degree threshold; Specifically, through the multi-dimensional matching degree design, strict verification of the physical propagation path of natural gas leakage events is achieved. The time matching degree can accurately reflect the deviation degree between the arrival time difference of the first wave peak and the theoretical propagation time. The type matching degree ensures that only data from the same type of sensors participates in the verification. The direction matching degree combines the coding angle to evaluate the consistency of the propagation direction. Finally, the total matching degree is calculated in a weighted manner, and high and low double thresholds are set to implement a quasi-real-time physical consistency screening and delay confirmation mechanism. This method takes into account both rigor and flexibility, achieving a balance between the false alarm risk and the missed alarm risk; Step S4: Dynamically adjust the local detection sensitivity strategy according to the neighborhood node credibility data and the environmental entropy value, trigger a leakage warning or enter the delay confirmation mode; The calculation of the environmental entropy value includes: Collect the wind speed, temperature gradient, and mechanical vibration signals around the current node; Use a sliding window to calculate the standard deviation ratio of each interference source, and generate an environmental disorder index normalized to 0-1; Perform a weighted average of the environmental disorder index and the same type of data of the neighborhood nodes to obtain the final environmental entropy value; The dynamic adjustment of the sensitivity strategy includes: Pre-set multiple sensitivity strategy templates, including the night static mode, strong wind interference mode, and construction protection mode; Automatically select or mix and generate the current optimal strategy according to the detection accuracy rate in the recent 24 hours and the change trend of the environmental entropy; When the credibility scores of more than half of the neighborhood nodes continue to be lower than the warning value, forcibly switch to the high-sensitivity inspection mode; The calculation method of the detection accuracy rate is: perform a normal distribution modeling on the time difference between the first report time of this node and the actual occurrence time of the event in the confirmed leakage events, and take the distribution function value as the accuracy rate index.

[0026] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. A method for detecting natural gas pipeline leakage based on edge computing, characterized in that, including Step S1: Synchronously collect pipeline pressure, gas concentration, and vibration signals at each edge node, and perform multi-scale feature extraction on the signals through a dynamic window generation module; Step S2: Input the extracted features into a lightweight neural network model to generate a local event credibility score, where the credibility score includes dual evaluation results of signal anomaly intensity and environmental interference probability; Step S3: Periodically exchange credibility data packets containing spatio-temporal fingerprints between adjacent edge nodes, and verify the physical consistency of the event propagation path based on the pipeline stress wave conduction characteristics; Step S4: Dynamically adjust the local detection sensitivity strategy according to the credibility data of neighboring nodes and the environmental entropy value, and trigger a leakage warning or enter a delay confirmation mode.

2. The method for detecting natural gas pipeline leakage based on edge computing according to claim 1, characterized in that, The dynamic window generation module performs the following operations: Trigger window scale switching according to the real-time signal energy accumulation rate. When the low-frequency signal energy accumulates to reach the first threshold, enable a large-scale window. When the high-frequency pulse signal density exceeds the second threshold, enable a microsecond-level sampling window; In a mixed event scenario, run dual-scale windows synchronously and output a fused feature vector; The trigger condition for the first threshold is that the low-frequency signal energy change rate continuously remains lower than the set slope within N consecutive sampling periods, and the baseline drift amount exceeds the threshold of the pipeline pressure rated value.

3. The method for detecting natural gas pipeline leakage based on edge computing according to claim 2, characterized in that, The multi-scale feature extraction includes: Perform baseline drift trend fitting on the low-frequency signal to generate a first feature group containing the drift slope and the periodic fluctuation amplitude; Extract the peak interval, duration, and energy decay rate of the pulse cluster from the high-frequency signal to generate a second feature group; Align the first feature group and the second feature group in time and splice them into a joint feature vector.

4. The method for detecting natural gas pipeline leakage based on edge computing according to claim 3, wherein, In Step S1, during the process of performing multi-scale feature extraction on the signals, dynamic window triggering and parallel output are performed. The dynamic window generation module calculates the energy and pulse density of the low-frequency and high-frequency signals in two windows in real time, and switches or runs in parallel according to the trigger conditions: Perform low-frequency energy rate calculation: ， When consecutive cycles are satisfied ， That is, enable the large-scale window , Perform high-frequency pulse density calculation: ， Enable the microscale window when ,​ Among them, represents the low-frequency signal of the th sampling point, represents the cumulative energy within the large-scale window, represents the energy accumulation rate, represents the sampling period, represents the large-scale window length, represents the number of sampling periods for continuous judgment, represents the rate threshold, represents the periodic pressure baseline drift amount of the represents the pressure offset threshold, represents the number of pulse clusters detected within the micro-scale window, represents the micro-scale window length, represents the pulse density threshold; If both sets of the above conditions are satisfied, the dual-scale windows run in parallel, and output two segments of original signals for subsequent feature extraction.

5. The method for detecting natural gas pipeline leakage based on edge computing according to claim 4, characterized in that, In step S1, during the process of performing multi-scale feature extraction on the signal, for the low-frequency signal segment within the large-scale window perform least squares linear fitting to obtain the drift slope and the fluctuation amplitude: ， ， Among them, ， ， ， represents the baseline drift slope, represents the periodic fluctuation amplitude; For the high-frequency signal segment within the micro-scale window Detect the pulse cluster, and denote the peak time of the cluster as , and the energy within the cluster is: ， Cluster duration , Peak interval , Exponential fitting for energy decay: ， Among them, represents the high-frequency signal sample, are respectively the start and end sampling indices of the cluster, represents the energy of the cluster, represents the duration of the cluster, represents the interval between adjacent cluster peaks, represents the number of clusters taken; The low-frequency features and the high-frequency features ， Align and splice in chronological order to obtain a joint feature vector: 。 6. The natural gas pipeline leakage detection method based on edge computing according to claim 1, characterized in that, The generation of the local event credibility score includes: Output the original anomaly probability value through the neural network; Superimpose the historical false alarm rate correction coefficient of the current node and the real-time intensity value of the environmental interference source to generate the final credibility score; The environmental interference source includes the readings of the wind speed sensor and the surface vibration sensor; The historical false alarm rate correction coefficient is obtained through a sliding window calculation: Statistically calculate the false alarm ratio confirmed by manual confirmation among the warning events triggered by the current node in the past period of time, and generate a correction coefficient according to 1 - false alarm ratio.

7. The natural gas pipeline leakage detection method based on edge computing according to claim 1, characterized in that, The spatio-temporal fingerprint includes the first peak timestamp of the vibration signal, the sensor type identifier, and the signal propagation direction encoding; The spatio-temporal fingerprint verification includes: When a node detects an abnormal vibration signal, record the first peak arrival timestamp; Receive the peak timestamp data of adjacent nodes, calculate the time difference, and perform a matching check with the stress wave propagation speed corresponding to the pipeline material; If the time difference matching degree exceeds the set ratio, it is marked as a valid physical conduction event.

8. The method for detecting natural gas pipeline leakage based on edge computing according to claim 7, wherein, In step S3, during the process of performing the matching degree verification, the adjacent nodes are received and After the spatio-temporal fingerprints, combined with the characteristics of pipeline stress wave propagation, calculate the matching degree and perform verification according to the following process: Calculate the time difference and the theoretical conduction time: ， Among them, represents the node first peak arrival timestamp, represents the distance between nodes measured along the centerline of the pipeline between nodes, represents the stress wave propagation speed corresponding to the pipeline material, Calculate time matching degree : ， Among them, represents the time error, represents the maximum allowable time deviation; Determine the type Degree of match with the direction : If , then , otherwise ; ， wherein, are respectively the sensor type identifiers of two nodes, are respectively the encoding angles of the signal propagation directions; Define the total matching degree and the verification standard as: ， Verification standard: ， Among them, is the weight coefficient, is the high matching degree threshold, is the low matching degree threshold.

9. The method for detecting natural gas pipeline leakage based on edge computing according to claim 1, characterized in that The calculation of the environmental entropy value includes: Collect the wind speed, temperature gradient and mechanical vibration signals around the current node; Use a sliding window to calculate the standard deviation ratio of each interference source and generate an environmental disorder index normalized to 0-1; Perform a weighted average of the environmental disorder index and the similar data of neighboring nodes to obtain the final environmental entropy value.

10. A natural gas pipeline leakage detection method based on edge computing according to claim 1, characterized in that, The dynamic adjustment of the sensitivity strategy includes: Pre-set multiple sensitivity strategy templates, including the night static mode, strong wind interference mode and construction protection mode; Automatically select or mix and generate the current optimal strategy according to the detection accuracy rate in the recent 24 hours and the change trend of the environmental entropy; When the credibility scores of more than half of the neighboring nodes continue to be lower than the warning value, forcibly switch to the high-sensitivity inspection mode; The calculation method of the detection accuracy rate is: perform a normal distribution modeling on the time difference between the first report time of this node and the actual occurrence time of the event in the confirmed leakage events, and take the distribution function value as the accuracy rate index.

Images