Method for detecting leakage of natural gas pipeline through double-side optical fibers

Abstract

The invention provides a method for detecting natural gas pipeline leakage through double-side optical fibers. According to the method, single-mode optical fibers are arranged on the two sides of a pipeline in parallel, optical pulses are sent through an optical time domain reflectometer, and back scattering light signals are received. Data preprocessing, feature extraction and leakage detection algorithms are adopted, and the method comprises signal smoothing, baseline correction, differential analysis, local anomaly detection, temporal correlation analysis, spatial gradient analysis and bilateral optical fiber comparison. A detection threshold is dynamically adjusted by combining a CFAR algorithm, spatial clustering is performed by using a DBSCAN algorithm, and finally a leakage position is determined by a weighted center method. According to the method, the sensitivity, the accuracy and the spatial resolution of leakage detection are remarkably improved, the false alarm rate is reduced, and the pipeline operation safety is enhanced.

Description

Technical Field

[0001] The present invention belongs to the technical field of natural gas pipeline leakage detection, and relates to a method for monitoring pipeline leakage using bilateral optical fiber sensing technology. Background Art

[0002] Currently, the leakage detection of long-distance natural gas pipelines mainly uses unilateral optical fiber distributed detection technology. The system of this method mainly consists of three parts: an optical fiber sensor laid unilaterally along the pipeline, an optical time domain reflectometer (OTDR) for sending optical pulses and receiving backscattered optical signals, and a signal processing unit for analyzing the received optical signals and judging whether leakage occurs.

[0003] The working principle of this technology is to send optical pulses to the optical fiber through the OTDR, and then analyze the scattered optical signals returned from the optical fiber. When a pipeline leaks, the temperature and strain near the leakage point will change, resulting in abnormal scattered optical signals. The system detects leakage by continuously monitoring the changes in these signals. Once an abnormal signal is detected, an alarm will be triggered and an attempt will be made to locate the leakage point.

[0004] However, this unilateral optical fiber detection technology has some limitations. First, since the optical fiber is only laid on one side of the pipeline, small leaks far from the optical fiber side may be missed, affecting the detection accuracy. Second, unilateral detection may lead to a large positioning error of the leakage point, reducing the accuracy of positioning. Finally, the unilateral optical fiber is easily affected by external factors such as surface temperature changes, increasing the false alarm rate of the system. Summary of the Invention

[0005] To solve the problems existing in the background art, the present invention proposes a method for detecting natural gas pipeline leakage using bilateral optical fibers.

[0006] To achieve the above object, the technical solution adopted by the present invention is as follows: A method for detecting natural gas pipeline leakage using bilateral optical fibers, comprising: Arranging a single-mode optical fiber on each side of the natural gas pipeline, namely Fiber A and Fiber B respectively, and recording the backscattered light intensity distribution of each optical fiber; Performing a preprocessing operation on the obtained optical fiber backscattered light intensity data to obtain optical fiber data; After the preprocessing is completed, performing feature extraction on the optical fiber data to extract optical fiber feature data, and analyzing the optical fiber feature data to perform leakage detection based on the optical fiber feature data.

[0007] Further, the two single-mode optical fibers are arranged on both sides of the pipeline and are parallel, and both optical fibers are fixed to the pipeline surface in a closely attached manner.

[0008] Further, the specific method for recording the backward scattering light intensity distribution of each optical fiber is as follows: Use OTDR to perform a complete scan every 5 minutes, and the original data of each scan is timestamped and distance as indices to record the backward scattering light intensity distribution along the optical fiber and store it as a two-dimensional array .

[0009] Further, the preprocessing operations on the obtained optical fiber backward scattering light intensity data include three steps: signal smoothing, baseline correction, and differential analysis.

[0010] The signal smoothing applies the Savitzky-Golay filter to smooth the signal and reduce measurement noise, and performs filtering operations on the data of each scan The specific formula is: ; where is the filtered data , is the corresponding function of the Savitzky-Golay filter, is the window length of the filter, is the order of the polynomial used for fitting, and the window length of the filter and the order of the polynomial used for fitting are parameters adjusted according to the actual signal characteristics; The baseline correction uses adaptive polynomial fitting to correct the baseline and remove long-term drift; For each position , select the data points of the past 24 hours to fit order polynomial , calculate the fitting residual, and the specific formula is: ; where is a order polynomial function used to fit the long-term trend or baseline of the data; When stop incrementing , and finally use the determined polynomial for baseline correction. The specific formula is: ; where is the corrected baseline; The differential analysis is completed by calculating the signal difference between adjacent time points. The specific formula is: ; where is the signal difference between adjacent time points.

[0011] Furthermore, the specific method for extracting features from the fiber optic data is as follows: The feature extraction stage includes four steps: local anomaly detection, temporal correlation analysis, spatial gradient analysis, and bilateral fiber comparison; In local anomaly detection, a sliding window is used to calculate local statistical features, and for each position , the local mean is calculated within a certain distance range and the local standard deviation , and an anomaly index is defined. The specific formula is: ; where is the local anomaly fiber feature data within the sliding window; Temporal correlation analysis calculates the temporal correlation of the anomaly indices for six consecutive measurements. The specific formula is: ; where is the anomaly fiber feature data in the time series; Spatial gradient analysis calculates the spatial gradient of the signal intensity to detect sharp changes. The specific formula is: ; where is the anomaly fiber feature data in the spatial sequence; Bilateral fiber comparison enhances the detection reliability by calculating the correlation coefficient of the anomaly indices at the same positions on both sides of the fiber. The specific formula is: ; where is the anomaly fiber feature data under bilateral fiber comparison.

[0012] Furthermore, the specific method for analyzing the fiber feature data and performing leakage detection based on the fiber feature data is as follows: Define a comprehensive anomaly index based on the anomaly fiber feature data. The specific formula is: ; where the subscripts A and B represent Fiber A and Fiber B respectively, are the weight coefficients that need to be optimized by machine learning methods; Use the constant false alarm rate (CFAR) algorithm to dynamically adjust the detection threshold. The specific formula is: ; where is an adjustable coefficient, with a value between 2.5 and 3.5. If , then this point is marked as a potential leakage point; Use the DBSCAN algorithm to perform spatial clustering on the detected abnormal points, and set the clustering parameters to a search radius of 20 meters and a minimum number of points of 3; For each cluster, check whether it is continuous in time, whether it appears in the data of both Fiber A and Fiber B at the same time, and whether the average comprehensive anomaly index exceeds 3 times the global average. If all conditions are met, it is confirmed as a leakage event.

[0013] Furthermore, the analysis of the optical fiber characteristic data and the leakage detection based on the optical fiber characteristic data further include: Use the weighted center method to determine the final leakage location. The specific formula is: ; where is the position of the th point in the cluster, is the distance from the OTDR device to the leakage point, and the OTDR calculates this distance by measuring the propagation time of the optical pulse in the optical fiber.

[0014] Compared with the prior art, the present invention has the following beneficial effects: Through the bilateral optical fiber arrangement and advanced signal processing algorithms, the present invention significantly improves the sensitivity and accuracy of natural gas pipeline leakage detection.

[0015] The present invention can quickly identify and locate small leaks in natural gas pipelines, significantly reduce the false alarm rate, and at the same time improve the spatial resolution of detection. This method has strong adaptability and high reliability, and can effectively cope with complex environmental interference.

[0016] By integrating machine learning technology, the system of the present invention has the ability of self-optimization and continuously improves the detection performance.

[0017] The present invention significantly enhances the safety of pipeline operation and reduces the maintenance cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 is a schematic diagram of the operation of the method of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0019] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0020] As Figure 1 shown, this embodiment describes an innovative bilateral optical fiber natural gas pipeline leakage detection system, which utilizes advanced optical fiber sensing technology and optical time domain reflectometry analysis to achieve all-round and real-time monitoring of long-distance natural gas pipelines. The core components of the system include bilaterally arranged optical fibers, an optical time domain reflectometer (OTDR), signal data preprocessing, feature extraction, and leakage detection. These components work together to form an efficient and reliable leakage detection network.

[0021] In the actual deployment of the system, a single-mode optical fiber is arranged on each side of the pipeline, and the two single-mode optical fibers are installed in parallel, named Fiber A and Fiber B respectively. These two optical fibers are closely attached to the surface of the pipeline. During the installation of the optical fibers, special attention is paid to protecting the optical fibers from physical damage in the external environment, while ensuring that they can sensitively capture the temperature and strain changes of the pipeline.

[0022] The present invention adopts a bilateral optical fiber arrangement method, and these two optical fibers are both fixed to the surface of the pipeline in a closely attached manner to ensure that the optical fibers can effectively sense the temperature and strain changes of the pipeline. The ends of the optical fibers are connected to an optical time domain reflectometer (OTDR) device to form a closed-loop measurement system. This arrangement method can provide all-round pipeline monitoring and greatly improve the sensitivity and reliability of leakage detection.

[0023] The optical time domain reflectometer (OTDR) is the core detection device of this system. The OTDR is placed at one end of the pipeline and is connected to Fiber A and Fiber B through optical fiber connectors. During operation, the OTDR sends an optical pulse to the two optical fibers simultaneously every 5 minutes, and the pulse width is 10 nanoseconds. When these optical pulses propagate in the optical fibers, backward scattered light is generated due to Rayleigh scattering and Fresnel reflection. The OTDR precisely measures the time and intensity characteristics of these scattered lights to obtain the optical fiber state information along the entire pipeline and stores it as a two-dimensional array .

[0024] During the data acquisition process, the system performs a complete scan every 5 minutes and records the backward scattered light intensity distribution of each optical fiber. For each optical fiber, the OTDR provides the backward scattered light intensity distribution along the optical fiber, where represents the distance from the OTDR device. The original data of each scan is indexed by the timestamp and the distance and stored as a two-dimensional array . This high-frequency data acquisition can timely capture the minute changes in the pipeline state and provide high-quality original data for subsequent leakage detection.

[0025] The data preprocessing unit is responsible for receiving and analyzing the raw data from the OTDR. This unit is equipped with high-performance computers and runs self-developed data processing algorithms. The algorithm first preprocesses the raw data, including removing noise using the Savitzky-Golay filter, applying adaptive polynomial fitting for baseline correction, and calculating the signal differences between adjacent time points.

[0026] The data preprocessing stage includes three main steps: signal smoothing, baseline correction, and differential analysis.

[0027] First, the Savitzky-Golay filter is applied for signal smoothing to reduce measurement noise. For the data of each scan the filtering operation is performed: ; where is the filtered data , is the corresponding function of the Savitzky-Golay filter, is the window length of the filter, is the order of the polynomial used for fitting. The window length of the filter and the order of the polynomial used for fitting are parameters adjusted according to the actual signal characteristics, usually .

[0028] Then, adaptive polynomial fitting is used for baseline correction to remove long-term drift. For each position , the data points of the past 24 hours are selected to fit order polynomial , and the fitting residual is calculated: ; where is a order polynomial function used to fit the long-term trend or baseline of the data.

[0029] When increases . Finally, the determined polynomial is used for baseline correction: ; where is the corrected baseline.

[0030] Finally, differential analysis is performed to calculate the signal differences between adjacent time points: ; where is the signal difference between adjacent time points.

[0031] These preprocessing steps can effectively improve the signal quality and lay a foundation for subsequent feature extraction and leakage detection.

[0032] The preprocessed data enters the feature extraction stage. Our algorithm analyzes the data from multiple dimensions, including calculating local anomaly metrics, time correlation analysis, spatial gradient analysis, and comparing the signal characteristics of both sides of the optical fiber. This multi-dimensional analysis method enables the system to effectively distinguish real leakage signals from environmental noise and greatly reduces the false alarm rate.

[0033] In local anomaly detection, a sliding window is used to calculate local statistical features. For each position , the local mean and local standard deviation are calculated within a certain distance range, and the anomaly metric is defined as: ; where is the local anomaly optical fiber feature data within the sliding window.

[0034] Time correlation analysis calculates the time correlation of the anomaly metrics corresponding to six consecutive measurements, which is 30 minutes: ; where is the anomaly optical fiber feature data under the time series.

[0035] Spatial gradient analysis calculates the spatial gradient of the signal intensity to detect sharp changes: ; where is the anomaly optical fiber feature data under the spatial sequence.

[0036] Finally, bilateral optical fiber comparison enhances the detection reliability by calculating the correlation coefficient of the anomaly metrics at the same positions on both sides of the optical fiber: ; where is the anomaly optical fiber feature data under bilateral optical fiber comparison.

[0037] These features together constitute a comprehensive leakage detection index system, which can effectively capture various possible leakage signals.

[0038] The leakage detection unit is the core of this system. It comprehensively considers all the extracted features and uses an adaptive threshold method to identify potential leakage points. Once an anomaly is detected, the algorithm will conduct further confirmation, requiring the anomaly signal to appear simultaneously in both optical fibers and last for a certain period of time (usually more than 15 minutes). This strict confirmation mechanism ensures the reliability of the detection results. For the confirmed leakage events, the system uses the weighted center method to accurately calculate the leakage location.

[0039] The leakage detection algorithm first defines a comprehensive anomaly index, which combines all the previously extracted features: ; where the subscripts A and B represent Fiber A and Fiber B respectively, are the weight coefficients that need to be optimized by machine learning methods. Then, the constant false alarm rate (CFAR) algorithm is used to dynamically adjust the detection threshold: ; where is an adjustable coefficient, with a value between 2.5 and 3.5. If , then this point is marked as a potential leakage point. Next, the DBSCAN (Density-Based Spatial Clustering of Applications with Noise) algorithm is used to perform spatial clustering on the detected anomaly points, with the clustering parameters set to a search radius of 20 meters and a minimum number of points of 3. For each cluster, check whether it is continuous in time for at least 15 minutes, whether it appears in the data of both Fiber A and Fiber B simultaneously, and whether the average comprehensive anomaly index exceeds 3 times the global average. If all conditions are met, it is confirmed as a leakage event.

[0040] Finally, the weighted center method is used to determine the final leakage location: ; where is the position of the th point in the cluster. Actually represents the distance from the OTDR device to the leakage point. The OTDR calculates this distance by measuring the propagation time of the optical pulse in the optical fiber. This calculation principle enables us to accurately locate the specific position of the leakage point on the pipeline.

[0041] Combined with the accurate pipeline wiring diagram, this method enables us to accurately mark the leakage point location on the pipeline diagram in a very short time. This greatly improves the efficiency of pipeline maintenance and safety management, enabling the operation personnel to quickly respond to and handle potential safety hazards.

[0042] Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for detecting leakage in a natural gas pipeline using double-sided optical fibers, characterized in that: Included are: A single-mode optical fiber, Fiber A and Fiber B, is arranged on both sides of the natural gas pipeline, and the backscattered light intensity distribution of each optical fiber is recorded; Performing preprocessing operations on the acquired optical fiber backscattered light intensity data to obtain optical fiber data; After the preprocessing is completed, the optical fiber data is subjected to feature extraction to obtain optical fiber feature data, the optical fiber feature data is analyzed, and leakage detection is performed based on the optical fiber feature data.

2. The method for detecting natural gas pipeline leakage using double-sided optical fiber according to claim 1, characterized in that: The two single-mode optical fibers are located on both sides of the pipeline and arranged in parallel, and the two optical fibers are fixed to the surface of the pipeline in a tightly attached manner.

3. The method for detecting natural gas pipeline leakage using double-sided optical fiber according to claim 1, characterized in that: The specific method for recording the backscattered light intensity distribution of each optical fiber is: Use OTDR to perform a full scan every 5 minutes, with the original data scanned in time stamped and distance As an index, record the backscattered light intensity distribution along the optical fiber and store it as a two-dimensional array .

4. A method for detecting leakage of natural gas pipelines according to claim 1, characterized in that The preprocessing operation of the acquired optical fiber backscattered light intensity data includes three steps: signal smoothing, baseline correction and differential analysis; The signal smoothing is applied to the Savitzky-Golay filter to smooth the signal, reduce measurement noise, and the data scanned each time Perform filtering operation, the specific formula is: ； in, The filtered data , is the function corresponding to the Savitzky-Golay filter, is the filter window length, The window length of the filter is for the order of the polynomial used for fitting and the order of the polynomial used for fitting It is a parameter adjusted according to the actual signal characteristics; The baseline correction uses adaptive polynomial fitting to correct the baseline and remove long-term drift; For each position , select the data points of the past 24 hours for fitting Polynomial , calculate the fitting residual, the specific formula is: ； in, is a Order polynomial function, used to fit long-term trends or baselines of data; when Stop incrementing when , finally use the determined polynomial for baseline correction, the specific formula is: ； in, is the corrected baseline; The differential analysis is completed by calculating the signal difference between adjacent time points. The specific formula is: ； in, is the signal difference between adjacent time points.

5. The method for detecting natural gas pipeline leakage using double-sided optical fibers according to claim 1, characterized in that: The specific method for extracting features from optical fiber data is as follows: The feature extraction stage includes four steps: local anomaly detection, time correlation analysis, spatial gradient analysis, and bilateral fiber comparison; In local anomaly detection, a sliding window is used to calculate local statistical features. , calculate the local mean within a certain distance range and the local standard deviation , define the abnormal index, the specific formula is: ； in It is the local abnormal optical fiber characteristic data in the sliding window; Time correlation analysis calculates the time correlation of abnormal indicators measured for 6 consecutive times. The specific formula is: ； in It is the abnormal optical fiber characteristic data in time series; Spatial gradient analysis calculates the spatial gradient of signal intensity to detect sharp changes. The specific formula is: ； in It is the abnormal optical fiber characteristic data in the spatial sequence; The double-sided optical fiber comparison enhances the detection reliability by calculating the correlation coefficient of the abnormal indicators of the optical fibers on both sides at the same position. The specific formula is: ； in It is the abnormal optical fiber characteristic data under the comparison of the optical fibers on both sides.

6. A method for detecting natural gas pipeline leakage using double-sided optical fibers according to claim 1 or 5, characterized in that: The specific method of analyzing the optical fiber characteristic data and performing leakage detection based on the optical fiber characteristic data is as follows: A comprehensive abnormality index is defined based on the abnormal optical fiber characteristic data. The specific formula is: ； The subscripts A and B represent Fiber A and Fiber B respectively. is the weight coefficient that needs to be optimized through machine learning methods; Use the constant false alarm rate algorithm to dynamically adjust the detection threshold. The specific formula is: ； in is an adjustable coefficient, ranging from 2.5 to 3.

5. , then the point is marked as a potential leakage point; The DBSCAN algorithm is used to perform spatial clustering on the detected outliers, and the clustering parameters are set to a search radius of 20 meters and a minimum number of points of 3; For each cluster, check whether it is continuous in time, whether it appears in the data of Fiber A and Fiber B at the same time, and whether the average comprehensive anomaly index exceeds 3 times the global average. If all conditions are met, it is confirmed as a leakage event.

7. A method for detecting natural gas pipeline leakage using double-sided optical fibers according to claim 6, characterized in that: The analyzing the optical fiber characteristic data and performing leakage detection based on the optical fiber characteristic data further includes: The weighted center method is used to determine the final leak location. The specific formula is: ； in It is the first The location of the point, The distance from the OTDR device to the leak point is calculated by measuring the propagation time of the light pulse in the optical fiber.

Images