A method, system, medium, and apparatus for monitoring oil and gas pipeline leaks

Abstract

The present application relates to the technical field of safety monitoring, and provides an oil and gas pipeline leakage monitoring method, system, medium and equipment, comprising: acquiring leakage vibration signals of each measuring point of the oil and gas pipeline, extracting time domain features, frequency domain features and space-time features, and an initial distance from the leakage point to the starting end of the optical fiber; performing cross-correlation analysis on the leakage vibration signals of adjacent measuring points to obtain a propagation time difference and a cross-correlation coefficient, and calculating a corrected distance from the leakage point to the starting end of the optical fiber; if the leakage point is in a spiral segment, calibrating the corrected distance to obtain a calibrated distance; calculating a difference between the initial distance from the leakage point to the starting end of the optical fiber and the corrected distance to obtain a preliminary positioning error value; based on the time domain features, the frequency domain features and the space-time features, in combination with the cross-correlation coefficient, the calibrated distance, an amplitude enhancement coefficient and the preliminary positioning error value, a medium identification model is used to obtain a leakage medium type. Both positioning efficiency and accuracy are taken into account, and the accuracy and robustness of the leakage medium type identification result are greatly improved.

Description

Technical Field

[0001] This invention belongs to the field of safety monitoring technology, and in particular relates to a method, system, medium and equipment for monitoring oil and gas pipeline leaks. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] As core infrastructure for energy transportation, oil and gas pipelines have long faced risks such as corrosion, geological subsidence, and damage from third-party construction, making them highly susceptible to leakage accidents. Leaks not only cause huge energy losses, but crude oil leaks can also pollute soil and water bodies, while natural gas leaks can lead to major safety accidents such as explosions. Therefore, timely detection of leaks, accurate location of leak points, and identification of leaking media are crucial to reducing disaster losses.

[0004] Existing oil and gas pipeline leak monitoring technologies have many limitations: manual inspection methods are inefficient, have blind spots, and are difficult to carry out in severe weather; negative pressure wave methods are not sensitive enough to minor leaks, and the location error is easily affected by pipeline operating conditions; acoustic detection methods are easily affected by environmental noise and have a high false alarm rate in multi-interference scenarios.

[0005] DAS (Distributed Fiber Optic Acoustic Sensing) technology, with its advantages of long-distance continuous monitoring, resistance to electromagnetic interference, and high sensitivity, has been gradually applied in the field of pipeline monitoring. However, the current technology still faces key bottlenecks:

[0006] First, demodulating a single vibration signal makes it difficult to distinguish between leakage vibration and interference vibrations from construction, vehicles, etc., which limits the positioning accuracy.

[0007] Secondly, it can only determine whether there is a leak, but cannot identify the type of leaked medium, which is not conducive to targeted emergency response.

[0008] Third, signal attenuation is severe in complex geological environments, which further reduces the reliability of positioning and identification. Summary of the Invention

[0009] To address the technical problems existing in the background art, the present invention provides a method, system, medium, and equipment for monitoring oil and gas pipeline leaks, which balances positioning efficiency and accuracy, and significantly improves the accuracy and robustness of the leak medium type identification results.

[0010] To achieve the above objectives, the present invention adopts the following technical solution:

[0011] The first aspect of the present invention provides a method for monitoring leaks in oil and gas pipelines, comprising:

[0012] Acquire leakage vibration signals from various measuring points in oil and gas pipelines, and extract time-domain, frequency-domain, and spatiotemporal features;

[0013] Based on the leakage vibration signal, the initial distance from the leak point to the fiber start-up end is calculated using the optical time-domain reflectometry principle. Cross-correlation analysis is performed on the leakage vibration signals of adjacent measuring points to obtain the propagation time difference and cross-correlation coefficient between adjacent measuring points. Based on the propagation time difference, the distance difference from the leak point to two adjacent measuring points is calculated. Combined with the spacing relationship between the two adjacent measuring points, the correction distance from the leak point to the fiber start-up end is calculated. If the leak point is not within the helical segment, the correction distance is used as the calibration distance. If the leak point is within the helical segment, the amplitude enhancement coefficient of the leakage vibration signal within the helical segment is calculated. Based on the amplitude enhancement coefficient, the correction distance is calibrated to obtain the calibration distance. The difference between the initial distance and the correction distance from the leak point to the fiber start-up end is calculated to obtain the preliminary positioning error value.

[0014] Based on time-domain, frequency-domain, and spatiotemporal characteristics, and combined with cross-correlation coefficients, calibration distance, amplitude enhancement coefficients, and preliminary location error values, the type of leaking medium is obtained through a medium identification model.

[0015] Furthermore, it also includes: calculating the intensity of the leakage vibration signal. Among them, signal strength ; Vibration characteristic coefficient for the type of leaking medium; RMS value ; The environmental noise reference value is N; N is the number of sampling points in the vibration signal segment. To calibrate the leakage vibration signal at the measuring point corresponding to the distance.

[0016] Furthermore, the steps for acquiring leakage vibration signals at each measuring point of the oil and gas pipeline include:

[0017] When the pulsed laser emitted by the distributed fiber optic acoustic wave sensing unit is transmitted in the sensing fiber, it generates backscattered Rayleigh light. The photodetector receives the scattered light signal and mixes it with the local oscillator reference light. The optical signal is converted into an electrical signal through coherent heterodyne demodulation.

[0018] After the electrical signal is amplified by a low-noise amplifier, environmental noise is filtered out using a filtering algorithm, and the amplitude of the electrical signal is normalized by normalization to obtain the leakage vibration signal.

[0019] Furthermore, the filtering algorithm includes the following steps:

[0020] Wavelet decomposition of the electrical signal yields one layer of low-frequency approximation components and multiple layers of high-frequency detail components.

[0021] For each layer of high-frequency detail components, a dynamic threshold formula based on the signal local variance is used to calculate the threshold for each layer;

[0022] Based on the threshold of each layer, a soft and hard threshold trade-off function is used to perform thresholding on the high-frequency detail components of each layer.

[0023] The high-frequency detail components and low-frequency approximate components after thresholding are subjected to inverse wavelet transform to reconstruct the leakage vibration signal.

[0024] Furthermore, the time-domain features include peak value, kurtosis, impulse factor, rise time, waveform factor, mean, variance, and peak factor;

[0025] The frequency domain features include modal frequencies, spectral entropy, resonant frequency offset, wavelet packet energy entropy, and Hilbert-Huang transform marginal spectral peaks.

[0026] The spatiotemporal characteristics include signal propagation speed, distance value corresponding to the phase change position, and time difference of signal arrival between adjacent measurement points.

[0027] Furthermore, the corrected distance from the leak point to the fiber optic start end... = + Where d is the distance between two adjacent measuring points, and the difference in distance from the leak point to the two adjacent measuring points. , To facilitate the time difference in transmission, For signal propagation speed, Let be the distance from measurement point i to the starting end of the optical fiber.

[0028] Furthermore, the calibration correction distance based on the amplitude enhancement coefficient is expressed as: ;in, For calibrating distance; The corrected distance from the leak point to the fiber start-up point; amplitude enhancement factor. , This represents the leakage vibration signal of the non-helical segment adjacent to the helical segment. This is a vibration signal indicating leakage within the helical section; This is the initial distance from the helical segment to the starting end of the optical fiber.

[0029] A second aspect of the present invention provides an oil and gas pipeline leakage monitoring system, comprising:

[0030] The signal acquisition module is configured to acquire leakage vibration signals from various measuring points in the oil and gas pipeline and extract time-domain, frequency-domain, and spatiotemporal features.

[0031] The positioning module is configured to: calculate the initial distance from the leak point to the fiber optic start-up end based on the leakage vibration signal and using the principle of optical time-domain reflection; perform cross-correlation analysis on the leakage vibration signals of adjacent measuring points to obtain the propagation time difference and cross-correlation coefficient between the leakage vibration signals of adjacent measuring points; calculate the distance difference from the leak point to two adjacent measuring points based on the propagation time difference, and calculate the correction distance from the leak point to the fiber optic start-up end based on the distance relationship between the two adjacent measuring points; if the leak point is not within the helical segment, use the correction distance as the calibration distance; if the leak point is within the helical segment, calculate the amplitude enhancement coefficient of the leakage vibration signal within the helical segment, calibrate the correction distance based on the amplitude enhancement coefficient, and obtain the calibration distance; calculate the difference between the initial distance and the correction distance from the leak point to the fiber optic start-up end to obtain the preliminary positioning error value.

[0032] The identification module is configured to: based on time domain features, frequency domain features, and spatiotemporal features, combined with cross-correlation coefficients, calibration distance, amplitude enhancement coefficient, and preliminary positioning error value, obtain the leakage medium type through a medium identification model.

[0033] A third aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the steps of the oil and gas pipeline leakage monitoring method described above.

[0034] A fourth aspect of the present invention provides a computer device including a computer-readable storage medium, a processor, and a computer program stored on the computer-readable storage medium and executable on the processor, wherein the processor executes the program to implement the steps of the oil and gas pipeline leakage monitoring method described above.

[0035] Compared with the prior art, the beneficial effects of the present invention are:

[0036] This invention employs a stepped distance calculation strategy that combines initial positioning via optical time-domain reflectometry (OTDR), corrected positioning via cross-correlation analysis, and precise calibration by structural segment. First, the initial distance to the leak point is obtained using the principle of optical time-domain reflectometry. Then, the propagation time difference is obtained through cross-correlation analysis of signals from adjacent measuring points, and the distance correction value is precisely calculated based on the spacing between the measuring points. Simultaneously, the invention specifically distinguishes between spiral and non-spiral sections of the pipeline. For spiral sections, an amplitude enhancement coefficient is introduced for secondary distance calibration, effectively compensating for positioning deviations caused by changes in the propagation characteristics of leakage vibration signals in the spiral section. For non-spiral sections, the corrected distance is directly used as the calibration distance, balancing positioning efficiency and accuracy.

[0037] This invention, by calculating the preliminary positioning error value between the initial distance and the corrected distance, can intuitively reflect the error level of the positioning process, providing a quantitative basis for positioning result verification and subsequent optimization, significantly reducing leak location error, and improving the accuracy and reliability of leak location in long-distance, complex-structure oil and gas pipelines.

[0038] This invention realizes photoelectric conversion of vibration signals based on coherent heterodyne demodulation technology of distributed optical fiber acoustic wave sensing. It combines wavelet decomposition, dynamic thresholding based on local variance, and soft and hard threshold compromise filtering algorithm to accurately denoise electrical signals. Compared with traditional filtering methods, it can adaptively match the noise characteristics of high-frequency detail components of each layer of the signal, filter out environmental noise, and retain the effective signal of leakage vibration to the greatest extent.

[0039] This invention breaks through the limitations of traditional leakage medium identification, which relies solely on a single signal feature. It deeply integrates the three-dimensional signal features in the time domain, frequency domain, and spatiotemporal domain with the core quantitative parameters of the positioning process (cross-correlation coefficient, calibration distance, amplitude enhancement coefficient, and preliminary positioning error value), and inputs them into the medium identification model to complete the judgment of the leakage medium type. This enriches the basis dimensions for medium identification and significantly improves the accuracy and robustness of the identification results.

[0040] This invention designs a quantitative calculation formula for leakage vibration signal intensity. The basic signal-to-noise ratio is calculated by using the effective value of the signal and the environmental noise baseline value. The final signal intensity is then obtained by combining the vibration characteristic coefficient of the leakage medium type. This enables a quantitative assessment of leakage intensity and provides accurate quantitative data support for maintenance personnel to judge the degree of leakage hazard and formulate differentiated handling strategies.

[0041] This invention is based on distributed optical fiber acoustic sensing technology to achieve signal acquisition. By utilizing the end-to-end distributed sensing characteristics of optical fiber, it can perform blind-spot-free monitoring of all measurement points of oil and gas pipelines. It is suitable for monitoring the needs of long-distance and large-span oil and gas pipelines, and can also be applied to oil and gas pipelines with different laying forms and different structural section combinations, such as direct burial, overhead, and seabed, which greatly improves the engineering practicality and scenario adaptability of the method. Attached Figure Description

[0042] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0043] Figure 1 This is a flowchart of an oil and gas pipeline leakage monitoring method according to Embodiment 1 of the present invention;

[0044] Figure 2 This is a diagram of the optical fiber duct laying according to Embodiment 1 of the present invention;

[0045] Figure 3 This is a schematic diagram of the structure of the improved CNN-GRU hybrid model according to Embodiment 1 of the present invention;

[0046] Figure 4 This is a comparison diagram of the frequency domain characteristics of leakage in different media according to Embodiment 1 of the present invention;

[0047] Figure 5This is a schematic diagram of the structure of a computer device according to Embodiment 4 of the present invention. Detailed Implementation

[0048] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings.

[0049] It should be noted that the following detailed description is illustrative and intended to provide further explanation of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0050] Example 1

[0051] This embodiment provides a method for monitoring leaks in oil and gas pipelines.

[0052] This embodiment provides a method for monitoring oil and gas pipeline leaks, which is applicable to real-time monitoring of leaks in long-distance crude oil, natural gas and multi-media mixed pipelines. It can achieve accurate location of leak points and rapid identification of leaking media, and is suitable for complex laying environments such as deserts, seabeds, and permafrost.

[0053] To address the issues of low positioning accuracy and inability to identify leaking media in existing DAS technology for oil and gas pipeline monitoring, this embodiment provides an oil and gas pipeline leak monitoring method. Through multi-dimensional signal feature extraction, fusion positioning algorithm and media identification model collaborative design, it achieves high-precision positioning of leak points and rapid differentiation of media such as crude oil and natural gas, thereby improving the intelligence level of oil and gas pipeline leak monitoring.

[0054] This embodiment provides a method for monitoring oil and gas pipeline leaks, which specifically includes the following steps:

[0055] Step 1: Deployment and initialization of the DAS sensing fiber optic array.

[0056] DAS sensing fiber optic array deployment: such as Figure 2 As shown, the DAS sensing optical fiber is laid in the same trench as the oil and gas pipeline. High pressure resistant and corrosion resistant quartz optical fiber is selected. Spiral winding sections of optical fiber are added at key locations prone to leakage, such as oil and gas pipeline interfaces, elbows, and valves, with 3-5 turns of winding to enhance the sensitivity of leakage signal acquisition. The distance between the DAS sensing optical fiber and the outer wall of the oil and gas pipeline is controlled at 0.2-0.5m to ensure effective transmission of vibration signals.

[0057] Parameter configuration: Start the DAS sensing unit, output pulsed laser from the narrow linewidth laser, set the pulse width to 20-60ns, and the repetition frequency to 5-15kHz; set the sampling frequency of the signal conditioning unit to 200-600kHz, initialize the parameters of the coherent heterodyne demodulation module to ensure accurate capture of the phase change of Rayleigh scattered light; at the same time, import the pipeline basic data, including pipeline material, pipe diameter, type of transported medium, laying path, etc., to provide benchmark parameters for subsequent signal analysis.

[0058] Step 2: Leakage vibration signal acquisition and preprocessing.

[0059] Signal Acquisition: When the pulsed laser emitted by the DAS sensing unit is transmitted in the DAS sensing fiber, it generates backscattered Rayleigh light. When a leak occurs in an oil and gas pipeline, the leaking medium impacts the inner wall of the oil and gas pipeline and the surrounding medium, causing vibration and deformation of the DAS sensing fiber, which leads to a change in the phase of the Rayleigh scattered light. The photodetector receives the scattered light signal and mixes it with the local oscillator reference light. The optical signal is converted into an electrical signal through coherent heterodyne demodulation and synchronously transmitted to the signal conditioning unit.

[0060] Signal preprocessing: The signal conditioning unit first amplifies the electrical signal by 30-50dB through a low-noise amplifier; then, an improved wavelet threshold filtering algorithm is used to filter out environmental noise, including interference signals such as soil settlement and vehicle movement; finally, the amplitude of the electrical signal is normalized to the [0,1] interval through normalization processing to eliminate the amplitude difference caused by signal attenuation and obtain a standardized leakage vibration signal.

[0061] Among them, the improved wavelet threshold filtering algorithm, targeting the complex interference characteristics of oil and gas pipeline scenarios, achieves noise reduction optimization through dynamic threshold adjustment and adaptive noise differentiation. The specific steps are as follows:

[0062] (1) Determination of wavelet basis and number of decomposition layers: The db8 wavelet basis (adapted to the non-stationary characteristics of leakage vibration signal) is selected. The number of decomposition layers is set to 5 according to the signal sampling frequency (200-600kHz). Wavelet decomposition is performed on the electrical signal to obtain 1 layer of low-frequency approximate component (A5) and 5 layers of high-frequency detail components (D1-D5). The low-frequency component A5 contains the core features of the leakage signal, and the high-frequency components D1-D5 are mainly environmental interference noise.

[0063] (2) Dynamic threshold calculation: For each layer of high-frequency detail components, the dynamic threshold formula based on the local variance of the signal is used to calculate the threshold of each layer.

[0064] The dynamic threshold formula based on the local variance of the signal is as follows:

[0065] ;

[0066] in, For the number of high-frequency detail component layers (1≤k≤5), intermediate parameters , For the high-frequency detail component of the k-th layer, median represents the median value. By using a dynamic threshold formula based on the local variance of the signal, the threshold is dynamically adjusted according to the noise intensity of each layer, avoiding feature loss or incomplete noise reduction caused by a fixed threshold.

[0067] (3) Improved threshold function processing: A soft and hard threshold trade-off function is adopted to perform threshold processing on the high-frequency detail components of each layer;

[0068] The expression for the soft and hard threshold tradeoff function is as follows:

[0069] ;

[0070] in, This represents the amplitude of the i-th sampling point of the high-frequency detail component in the k-th layer after processing with the soft-hard thresholding tradeoff function. Let be the amplitude of the i-th sampling point of the original k-th layer high-frequency detail component, and sign() be the sign function. This improved threshold function smooths the abrupt change points of soft and hard thresholds through the exponential term, which not only preserves the signal abrupt change characteristics (such as leakage, impact, and vibration) but also suppresses random noise, thus solving the signal distortion problem of the traditional hard threshold function.

[0071] (4) Signal reconstruction: The high-frequency detail components D'1-D'5 after thresholding and the low-frequency approximation component A5 are subjected to inverse wavelet transform to reconstruct the noise-reduced vibration signal.

[0072] The above steps can improve the signal-to-noise ratio of the leakage signal by 5-8 dB, effectively separating leakage vibration from environmental interference.

[0073] Step 3: Multi-dimensional feature extraction.

[0074] Temporal feature extraction: Temporal feature parameters are extracted from the preprocessed vibration signal, including peak value, kurtosis, impulse factor, rise time, waveform factor, mean, variance, peak factor, etc. For example, the peak value of the vibration signal generated by natural gas leakage is relatively flat, while the peak value of the crude oil leakage signal fluctuates more significantly, providing a basis for medium identification.

[0075] Frequency domain feature extraction: The preprocessed vibration signal is converted into a frequency domain signal using Fast Fourier Transform (FFT). Frequency domain features such as modal frequency, spectral entropy, resonant frequency offset, wavelet packet energy entropy (4 frequency bands), and Hilbert-Huang transform marginal spectrum peak values ​​(4 frequency bands) are extracted. For example, the center frequency of vibrations from crude oil leaks is concentrated in the 50-200Hz range, while the center frequency of vibrations from natural gas leaks is concentrated in the 200-500Hz range. This frequency domain difference provides core features for distinguishing the media. Figure 4 As shown.

[0076] Spatiotemporal feature extraction: Extracting signal propagation velocity based on the time difference of vibration signal transmission at different measuring points in the optical fiber. Phase abrupt change location The corresponding distance value and the time difference of arrival of signals from adjacent measuring points The spatiotemporal characteristics provide parameter support for locating the leak point.

[0077] Step 4: Leak point fusion and localization.

[0078] Preliminary positioning: Based on the principle of optical time domain reflectance (OTDR), according to the propagation time of Rayleigh scattered light, combined with the propagation speed of light in optical fiber (taken as 1.93 × 10⁻⁶), 8 The initial location of the signal abrupt change point is calculated using the formula: (m / s). ,in, This represents the initial distance from the leak point to the fiber optic start end, obtained from preliminary location. At the speed of light, The round-trip time of the leakage vibration signal (i.e., the complete time for the pulsed laser to travel from the starting end of the optical fiber, through the sensing optical fiber to the location of the leak, and then for the backscattered Rayleigh light generated at the leak point to return to the starting end of the optical fiber along the original path). The refractive index of the optical fiber is given.

[0079] Precise Correction: A cross-correlation positioning algorithm is introduced to perform cross-correlation analysis on the vibration signals of adjacent fiber optic measuring points. Combined with the enhanced signals collected from the fiber optic helical segments of key pipeline locations, a two-level calibration mechanism of "algorithm correction + signal enhancement" is constructed to ultimately achieve precise positioning with an accuracy of ≤0.5m. The specific implementation process is as follows:

[0080] (1) Input to the cross-correlation positioning algorithm: preprocessed vibration signals collected from two adjacent fiber optic measuring points (the distance between the measuring points is fixed at d=1m). , Where i < j, the signal sampling frequency is uniformly f s =500kHz.

[0081] (2) Output of the cross-correlation localization algorithm: the propagation time difference of the leakage vibration signal between the two measuring points. Cross-relationships ,in, The time difference is determined to be valid.

[0082] (3) The specific correction process for the initial positioning error.

[0083] Let the distances from two adjacent measuring points i and j of the leak point to the starting end of the optical fiber be... , The steps are as follows:

[0084] Based on cross-correlation time difference Calculate the distance difference between the leak point and two adjacent measuring points. ;

[0085] Combining the distance relationship between two adjacent measuring points By combining the equations, we can obtain the distance from the leak point to the measuring point. ;

[0086] The corrected distance from the leak point to the fiber optic start segment is calculated using a corrected formula. : = + .

[0087] The steps to obtain the distance from the leak point to the measuring point using simultaneous equations are as follows:

[0088] Given: measuring point The distance to the starting end of the optical fiber is Relationship between measuring points The difference in distance from the leak point to the two measuring points: ;Calculated from the cross-correlation time difference: ;

[0089] Geometric relationship derivation: ;

[0090] Solve the simultaneous equations: Substitute into the distance difference formula After sorting, we get: .

[0091] (4) Precise calibration process for fiber helical segment enhancement signal.

[0092] like Figure 2 As shown, the optical fiber in key parts of the pipeline (interfaces, elbows, valves) is laid using spiral winding (3-5 turns, winding length...). (It is known that) this laying method can amplify the leakage signal amplitude by 3-5 times. The calibration procedure is as follows:

[0093] To determine whether the initial location of the leak is within the coverage area of ​​the helical segment, specifically, this is done by analyzing the fiber optic topology. Is it in interval, , This is the starting / ending distance from the helical segment to the fiber optic start end;

[0094] If the leakage is located within the helical section, extract the leakage vibration signal collected from the helical section. Calculate its amplitude enhancement coefficient ; This refers to the signal adjacent to the non-helical segment (the signal collected by the ordinary fiber optic segment next to the helical segment).

[0095] Based on the reinforcement coefficient, the leak distance is calibrated using the following formula: ;

[0096] If the leak point is not within the spiral segment, directly use As calibration distance .

[0097] Step 5: Calculate the initial positioning error value and the corrected positioning accuracy value.

[0098] (1) Preliminary positioning error value (denoted as) Based on the initial positioning distance Distance from precise positioning The deviation quantification formula is: = ;in, This is the precise distance of the leak points after correction using the cross-correlation algorithm. This is the preliminary positioning distance calculated based on the OTDR principle.

[0099] (2) The corrected positioning accuracy value (denoted as) The problem is divided into two scenarios: "leakage point inside the spiral segment" and "leakage point outside the spiral segment." The quantification is based on the deviation between the final location result and the actual leak location, using the following formula:

[0100] The leak point is not within the spiral segment: = ;

[0101] The leak point is inside the spiral segment: = ;

[0102] in, Locate leaks - identify fusion baseline distance; The accurate distance after correction by the cross-correlation algorithm; This is the final positioning distance after calibration using the enhanced signal from the spiral segment.

[0103] in, The calculation formula is as follows:

[0104] ;

[0105] in, The signal propagation speed correction factor is calculated using the following formula:

[0106] ;

[0107] in, The signal propagation speed is the actual speed measured based on the time difference of arrival of signals from adjacent measuring points. The signal propagation speed is preset based on the material / diameter when deploying pipelines.

[0108] Step 6: Training the intelligent identification model for leaking media.

[0109] An improved lightweight CNN-GRU (convolutional neural network-gated recurrent unit) hybrid model was constructed as a medium recognition model. The preprocessed multi-dimensional feature vectors (including time domain, frequency domain, spatiotemporal features, and computational features) were divided into training set and test set (7:3 ratio). The training set was used for parameter optimization of the CNN-GRU hybrid model, and the test set was used for performance verification of the CNN-GRU hybrid model.

[0110] In the CNN-GRU hybrid model, the CNN layer extracts the spatial correlation of features, enhancing the discriminative power of frequency domain features; the GRU layer captures the temporal variation of signals, adapting to the vibration temporal differences in leakage under different media. For example... Figure 3 As shown, the CNN-GRU hybrid model is a series structure consisting of an input layer → a lightweight CNN module → an attention fusion layer → a GRU module → a fully connected layer → an output layer, specifically including:

[0111] Input layer: The input feature vector has a dimension of 28 (including time domain, frequency domain, spatiotemporal features, and computational features). The input format is [batchsize,28,1], where batchsize is the number of samples input into the CNN-GRU hybrid model for training or inference each time.

[0112] The lightweight CNN module (parameter reduction and efficiency improvement) includes multiple convolutional layers. The output feature map of the second convolutional layer (Conv2) has 64 channels, and the output feature map is fed into the attention fusion layer. Depthwise separable convolution is used to reduce parameters by 60% to alleviate overfitting. Specifically: Convolutional layer 1: 3×1 kernel size, 16 kernels, stride 1, activation function ELU (Exponential Linear Unit, replacing ReLU to alleviate gradient vanishing), using depthwise separable convolution (reducing parameter count by 60%); Pooling layer 1: Max pooling, 2×1 kernel, stride 2; Convolutional layer 2: 3×1 kernel size, 32 kernels, stride 1, activation function ELU, depthwise separable convolution; Pooling layer 2: Average pooling, 2×1 kernel, stride 2; Dropout layer: dropout rate 0.2 (suppressing overfitting);

[0113] Attention Fusion Layer: Introduces a channel attention mechanism (SE module) to assign weights to the feature maps extracted by CNN, thereby strengthening the feature channels that are strongly correlated with medium recognition;

[0114] GRU module: 2 hidden layers, 64 hidden units per layer, using a bidirectional GRU structure (Bi-GRU), processing feature sequences from both forward and reverse directions simultaneously to fully capture the temporal context of leakage vibration signals; a gated forgetting mechanism is introduced, with the initial weight of the forgetting gate set to 0.7; dropout layer: dropout rate of 0.2;

[0115] Fully connected layer: 16 hidden units, ELU activation function;

[0116] Output layer: Output unit 4 corresponds to four media: crude oil, natural gas, water, and chemical solvents. The recognition frequency of each media is output through the activation function Softmax.

[0117] The feature vector input to the CNN-GRU hybrid model contains four main categories of features, as follows:

[0118] Temporal characteristics (8 dimensions): peak value, kurtosis, impulse factor, rise time, waveform factor, mean, variance, peak factor;

[0119] Frequency domain features (12 dimensions): modal frequency, spectral entropy, resonant frequency offset, wavelet packet energy entropy (4 frequency bands), Hilbert-Huang transform marginal spectral peak (4 frequency bands).

[0120] Spatiotemporal characteristics: signal propagation speed Phase abrupt change location Corresponding distance value, average time difference between adjacent measuring points ;

[0121] The calculated characteristic obtained above: cross-correlation coefficient Calibration distance, helical segment signal enhancement coefficient k, preliminary positioning error value, and corrected positioning accuracy value.

[0122] The multi-dimensional feature vector to be identified is input into the trained CNN-GRU hybrid model. The CNN-GRU hybrid model compares the feature parameters with the preset feature templates of crude oil, natural gas and other media, and outputs the media identification result. If the confidence of the identification result is ≥95%, the media type is directly determined. If the confidence is between 80-95%, a secondary signal acquisition and verification is triggered. If the confidence is <80%, a suspected leak warning is issued and the area to be verified is marked.

[0123] The CNN-GRU hybrid model extracts spatial correlations of features through CNN layers and captures temporal correlations of features through GRU layers, achieving a media recognition accuracy of ≥96%.

[0124] Step 7: Result Output and Early Warning Data Processing.

[0125] The system transmits information such as the final leak location (calibration distance), medium type, and leak signal strength. It displays the three-dimensional path of the pipeline and the coordinates of the leak point through a visualization platform. Based on the medium type, it matches the emergency plan and sends early warning information to maintenance personnel through audible and visual alarms, SMS push notifications, and other means.

[0126] The leakage signal strength was obtained through standardized calculation and feature correlation, and the specific acquisition process is as follows:

[0127] (1) Extraction of fundamental signal amplitude: Select the preprocessed vibration signal from the fiber optic measuring point corresponding to the calibration distance. Calculate the peak amplitude of the vibration signal. With effective value .

[0128] Among them, effective value The calculation formula is:

[0129] ;

[0130] Where N is the number of sampling points for this vibration signal segment. Let be the timestamp of the i-th sampling point.

[0131] (2) Environmental noise benchmark calibration: Extract the preprocessed vibration signals from 5 normal measuring points adjacent to the leak point that are free from leakage interference, and calculate their mean effective value. As a benchmark value for environmental noise, the influence of background noise on intensity determination is eliminated.

[0132] (3) Signal strength normalization calculation: The leakage signal leakage strength is quantified using the signal-to-noise ratio (SNR), and the formula is:

[0133] ;

[0134] The SNR unit is dB, and the range of values ​​corresponds to the signal strength: when 10≤SNR<20, it is a weak leakage; when 20≤SNR<35, it is a medium leakage; when SNR≥35, it is a high leakage.

[0135] (4) Strength verification correction: combining the vibration characteristic coefficient of the leakage medium type The SNR is then corrected to obtain the final leakage signal strength. This ensures consistency in determining the intensity of leaks from different media. Specifically, crude oil (k) m =1.0, natural gas k m =1.2, water k m =0.8, chemical solvent k m =1.1, determined experimentally.

[0136] Experiment 1: Monitoring leaks in onshore natural gas pipelines.

[0137] For example, consider a long-distance natural gas pipeline made of X80 steel with a diameter of 1200mm, laid underground, traversing plains and hilly areas. The monitoring method described in this embodiment is deployed, and the specific implementation process is as follows:

[0138] The sensing optical fiber adopts the same trench laying method. A 0.8m diameter spiral wound section of optical fiber is laid at the pipe interface, with 4 turns of winding. The distance between the optical fiber and the outer wall of the pipe is 0.3m. The pulse width of the DAS sensing unit is set to 40ns, the repetition frequency is 10kHz, and the signal sampling frequency is 300kHz.

[0139] A small leak was simulated at a pipeline interface with a leakage rate of 0.1 m³ / h. The DAS sensing unit collected the leakage vibration signal, and after improved wavelet threshold filtering, the signal-to-noise ratio was improved by 5.2 dB.

[0140] The extracted signal has a time domain peak value of 0.75V, a kurtosis of 3.8, a frequency domain center frequency of 320Hz, and a spectral entropy of 0.62. Using the OTDR principle, the leak point was initially located at 28.5km from the monitoring starting point. After correction by the cross-correlation algorithm, the location result was 28.5003km, which deviated from the actual leak point by 0.3m.

[0141] The feature vector is input into the CNN-GRU model, and the output natural gas identification result is 98.2% with a confidence level. The system triggers a level 2 warning and pushes the warning information within 30 seconds.

[0142] Experiment 2: Monitoring of leaks in subsea crude oil pipelines.

[0143] For a certain subsea crude oil pipeline, buried at a depth of 1.5m, with surrounding ocean current interference, the monitoring method described in this embodiment is deployed.

[0144] The sensing fiber is armored and laid along the pipeline axis. A spiral-wound section of the fiber is added at the valve. The pulse width is set to 50ns, the repetition frequency to 12kHz, and the sampling frequency to 400kHz.

[0145] The simulated crude oil leakage rate was 0.5 m³ / h. After filtering, the vibration signal generated by the leakage was extracted with a time domain peak value of 1.2 V, a kurtosis of 5.1, a frequency domain center frequency of 120 Hz, and a spectral entropy of 0.58.

[0146] The leak location was determined to be 0.4m off by the fusion positioning algorithm. The CNN-GRU model output the crude oil identification result with a confidence level of 97.5%, effectively avoiding misjudgment caused by ocean current interference, and simultaneously pushing out emergency response guidelines for crude oil leaks.

[0147] This embodiment provides a method for monitoring leaks in oil and gas pipelines, which significantly improves the positioning accuracy: by combining the initial positioning based on the OTDR principle with the precise correction by the cross-correlation algorithm, and by the fiber optic reinforcement layout design in key parts, the positioning accuracy can reach within 0.5m, which is more than 80% lower than the error of the traditional DAS positioning method, and solves the problem of ambiguous leak location in complex environments.

[0148] This embodiment provides a method for monitoring oil and gas pipeline leaks, achieving efficient identification of leaking media: It innovatively constructs a CNN-GRU hybrid model, integrating multi-dimensional features in the time and frequency domains, which can quickly distinguish between leaking media such as crude oil and natural gas, with an identification accuracy of over 96%. This fills the technical gap of traditional DAS technology, which can only detect leaks but cannot identify the media, providing accurate basis for emergency response.

[0149] This embodiment provides a method for monitoring oil and gas pipeline leaks, which has strong anti-interference capabilities: it adopts an improved wavelet threshold filtering algorithm to filter out environmental interference signals, effectively distinguishes leakage vibration from interference signals from third-party construction, vehicles, etc., reduces the false alarm rate to below 2%, and is suitable for various complex laying environments.

[0150] This embodiment provides a method for monitoring oil and gas pipeline leaks, which is suitable for long-distance monitoring needs: relying on the advantages of long-distance continuous monitoring of DAS distributed optical fiber, no additional sensors are required, and integrated monitoring of a single pipeline section of more than 50km can be achieved, which greatly reduces the deployment cost of the monitoring system and is suitable for continuous operation in extreme environments around the clock.

[0151] Example 2

[0152] This embodiment provides an oil and gas pipeline leakage monitoring system, including:

[0153] The signal acquisition module is configured to acquire leakage vibration signals from various measuring points in the oil and gas pipeline and extract time-domain, frequency-domain, and spatiotemporal features.

[0154] The positioning module is configured to: calculate the initial distance from the leak point to the fiber optic start-up end based on the leakage vibration signal and using the principle of optical time-domain reflection; perform cross-correlation analysis on the leakage vibration signals of adjacent measuring points to obtain the propagation time difference and cross-correlation coefficient between the leakage vibration signals of adjacent measuring points; calculate the distance difference from the leak point to two adjacent measuring points based on the propagation time difference, and calculate the correction distance from the leak point to the fiber optic start-up end based on the distance relationship between the two adjacent measuring points; if the leak point is not within the helical segment, use the correction distance as the calibration distance; if the leak point is within the helical segment, calculate the amplitude enhancement coefficient of the leakage vibration signal within the helical segment, calibrate the correction distance based on the amplitude enhancement coefficient, and obtain the calibration distance; calculate the difference between the initial distance and the correction distance from the leak point to the fiber optic start-up end to obtain the preliminary positioning error value.

[0155] The identification module is configured to: based on time domain features, frequency domain features, and spatiotemporal features, combined with cross-correlation coefficients, calibration distance, amplitude enhancement coefficient, and preliminary positioning error value, obtain the leakage medium type through a medium identification model.

[0156] It should be noted that each module in this embodiment corresponds one-to-one with each step in Embodiment 1, and their specific implementation processes are the same, so they will not be repeated here.

[0157] Example 3

[0158] This embodiment provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the oil and gas pipeline leakage monitoring method described in Embodiment 1 above.

[0159] Example 4

[0160] This embodiment provides a computer device, such as... Figure 5 As shown, the system includes a computer-readable storage medium 1003, a processor 1001, a communication interface 1002, and a computer program stored on the computer-readable storage medium 1003 and executable on the processor 1001. The processor 1001, communication interface 1002, and computer-readable storage medium 1003 can be connected via a bus or other means. The communication interface 1002 is used to receive and transmit data. When the processor 1001 executes the program, it implements the steps of the oil and gas pipeline leakage monitoring method described in Embodiment 1 above.

[0161] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for monitoring leaks in oil and gas pipelines, characterized in that, include: Acquire leakage vibration signals from various measuring points in oil and gas pipelines, and extract time-domain, frequency-domain, and spatiotemporal features; Based on the leakage vibration signal, the initial distance from the leak point to the fiber start-up end is calculated using the optical time-domain reflectometry principle. Cross-correlation analysis is performed on the leakage vibration signals of adjacent measuring points to obtain the propagation time difference and cross-correlation coefficient between adjacent measuring points. Based on the propagation time difference, the distance difference from the leak point to two adjacent measuring points is calculated. Combined with the spacing relationship between the two adjacent measuring points, the correction distance from the leak point to the fiber start-up end is calculated. If the leak point is not within the helical segment, the correction distance is used as the calibration distance. If the leak point is within the helical segment, the amplitude enhancement coefficient of the leakage vibration signal within the helical segment is calculated. Based on the amplitude enhancement coefficient, the correction distance is calibrated to obtain the calibration distance. The difference between the initial distance and the correction distance from the leak point to the fiber start-up end is calculated to obtain the preliminary positioning error value. Based on time-domain features, frequency-domain features, and spatiotemporal features, combined with cross-correlation coefficients, calibration distance, amplitude enhancement coefficients, and preliminary location error values, the type of leaking medium is obtained through a medium identification model. The calibration correction distance based on the amplitude enhancement coefficient is expressed as: ;in, For calibrating distance; The corrected distance from the leak point to the fiber start-up point; amplitude enhancement factor. , This represents the leakage vibration signal of the non-helical segment adjacent to the helical segment. This is a vibration signal indicating leakage within the helical section; This is the initial distance from the helical segment to the starting end of the optical fiber.

2. The method for monitoring oil and gas pipeline leaks as described in claim 1, characterized in that, Also includes: Calculate the intensity of the leakage vibration signal. Among them, signal strength ; Vibration characteristic coefficient for the type of leaking medium; RMS value ; The environmental noise reference value is N; N is the number of sampling points in the vibration signal segment. To calibrate the leakage vibration signal at the measuring point corresponding to the distance.

3. The method for monitoring oil and gas pipeline leaks as described in claim 1, characterized in that, The steps for acquiring leakage vibration signals at each measuring point of the oil and gas pipeline include: When the pulsed laser emitted by the distributed fiber optic acoustic wave sensing unit is transmitted in the sensing fiber, it generates backscattered Rayleigh light. The photodetector receives the scattered light signal and mixes it with the local oscillator reference light. The optical signal is converted into an electrical signal through coherent heterodyne demodulation. After the electrical signal is amplified by a low-noise amplifier, environmental noise is filtered out using a filtering algorithm, and the amplitude of the electrical signal is normalized by normalization to obtain the leakage vibration signal.

4. The method for monitoring oil and gas pipeline leaks as described in claim 3, characterized in that, The filtering algorithm includes the following steps: Wavelet decomposition of the electrical signal yields one layer of low-frequency approximation components and multiple layers of high-frequency detail components. For each layer of high-frequency detail components, a dynamic threshold formula based on the signal local variance is used to calculate the threshold for each layer; Based on the threshold of each layer, a soft and hard threshold trade-off function is used to perform thresholding on the high-frequency detail components of each layer. The high-frequency detail components and low-frequency approximate components after thresholding are subjected to inverse wavelet transform to reconstruct the leakage vibration signal.

5. The method for monitoring oil and gas pipeline leaks as described in claim 1, characterized in that, The time-domain features include peak value, kurtosis, impulse factor, rise time, waveform factor, mean, variance, and peak factor. The frequency domain features include modal frequencies, spectral entropy, resonant frequency offset, wavelet packet energy entropy, and Hilbert-Huang transform marginal spectral peaks. The spatiotemporal characteristics include signal propagation speed, distance value corresponding to the phase change position, and time difference of signal arrival between adjacent measurement points.

6. The method for monitoring oil and gas pipeline leaks as described in claim 1, characterized in that, The correction distance from the leak point to the fiber start end = + Where d is the distance between two adjacent measuring points, and the difference in distance from the leak point to the two adjacent measuring points. , To facilitate the time difference in transmission, For signal propagation speed, Let be the distance from measurement point i to the starting end of the optical fiber.

7. A leak monitoring system for oil and gas pipelines, characterized in that, include: The signal acquisition module is configured to acquire leakage vibration signals from various measuring points in the oil and gas pipeline and extract time-domain, frequency-domain, and spatiotemporal features. The positioning module is configured to: calculate the initial distance from the leak point to the fiber optic start-up end based on the leakage vibration signal and using the principle of optical time-domain reflectometry; perform cross-correlation analysis on the leakage vibration signals of adjacent measuring points to obtain the propagation time difference and cross-correlation coefficient between the leakage vibration signals of adjacent measuring points; calculate the distance difference from the leak point to two adjacent measuring points based on the propagation time difference, and calculate the correction distance from the leak point to the fiber optic start-up end based on the distance relationship between the two adjacent measuring points; if the leak point is not within the helical segment, use the correction distance as the calibration distance; if the leak point is within the helical segment, calculate the amplitude enhancement coefficient of the leakage vibration signal within the helical segment, and calibrate the correction distance based on the amplitude enhancement coefficient to obtain the calibration distance; Calculate the difference between the initial distance and the corrected distance from the leak point to the fiber optic start end to obtain the preliminary positioning error value; The identification module is configured to: based on time domain features, frequency domain features, and spatiotemporal features, combined with cross-correlation coefficients, calibration distance, amplitude enhancement coefficient, and preliminary positioning error value, obtain the leakage medium type through a medium identification model; The calibration correction distance based on the amplitude enhancement coefficient is expressed as: ;in, For calibrating distance; The corrected distance from the leak point to the fiber start-up point; amplitude enhancement factor. , This represents the leakage vibration signal of the non-helical segment adjacent to the helical segment. This is a vibration signal indicating leakage within the helical section; This is the initial distance from the helical segment to the starting end of the optical fiber.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps in the oil and gas pipeline leakage monitoring method as described in any one of claims 1-6.

9. A computer device comprising a computer-readable storage medium, a processor, and a computer program stored on the computer-readable storage medium and executable on the processor, characterized in that, When the processor executes the program, it implements the steps in the oil and gas pipeline leakage monitoring method as described in any one of claims 1-6.

Images