High-pressure pipeline leakage detection method and device based on voiceprint map

Abstract

The invention discloses a high-pressure pipeline leakage detection method and device based on a voiceprint map, and relates to the technical field of pipeline detection. According to the method, sound signals are collected through a distributed microphone array, multi-dimensional features such as wavelet packet frequency band energy and Mel-frequency cepstral coefficients are extracted after variational mode decomposition is combined with wavelet threshold denoising preprocessing, and dimensionality reduction is performed through principal component analysis; the time delay between the sensors is calculated based on a generalized cross-correlation-phase transformation algorithm, a leakage point is positioned through a particle swarm optimization algorithm, the leakage degree can be evaluated by combining a bidirectional long-short-term memory network of an attention mechanism, and the D-S evidence theory is utilized to fuse multi-sensor result early warning. The system realizes high-sensitivity detection and accurate positioning, is high in anti-interference capability, and is suitable for safety monitoring of high-pressure pipelines in the industries of petroleum, natural gas and the like.

Description

Technical Field

[0001] This invention belongs to the field of pipeline inspection technology, specifically relating to a method and device for detecting leaks in high-pressure pipelines based on acoustic signatures. Background Technology

[0002] High-pressure pipelines are widely used in industries such as oil, natural gas, and chemicals, and their safe operation is of paramount importance. Pipeline leaks not only cause resource waste and economic losses but can also trigger serious accidents such as environmental pollution, fires, and explosions. Therefore, timely and accurate detection of the location and extent of leaks in high-pressure pipelines is of great significance.

[0003] Currently, commonly used methods for detecting leaks in high-pressure pipelines include pressure testing, flow rate testing, and ultrasonic testing. Pressure testing and flow rate testing have low sensitivity to minute leaks and are difficult to pinpoint the exact location of the leak; ultrasonic testing is easily affected by environmental noise and performs poorly under complex operating conditions.

[0004] Voiceprint mapping technology, through the analysis and processing of sound signals, can extract specific voiceprint features and has been successfully applied in fields such as speech recognition and fault diagnosis. Applying voiceprint mapping technology to high-pressure pipeline leak detection is expected to improve the sensitivity and accuracy of detection.

[0005] In view of this, it is of great significance to propose a method and device for detecting high-pressure pipeline leaks based on acoustic signatures. Summary of the Invention

[0006] The present invention aims to provide a method and device for detecting high-pressure pipeline leaks based on acoustic signatures. Through multi-dimensional feature extraction and intelligent analysis, it achieves high-sensitivity detection and accurate location of high-pressure pipeline leaks, thereby solving the aforementioned technical defects.

[0007] In a first aspect, the present invention proposes a method for detecting leaks in high-pressure pipelines based on acoustic signatures, the method comprising the following steps:

[0008] S1. Acoustic signal acquisition: Deploy a distributed microphone array along the high-pressure pipeline to acquire acoustic signals during pipeline operation;

[0009] S2. Signal preprocessing: The acquired acoustic signal is preprocessed using the variational mode decomposition (VMD) algorithm combined with the wavelet threshold noise reduction algorithm.

[0010] S3. Feature Extraction: Wavelet packet decomposition is performed on the preprocessed signal to extract the energy features of each frequency band, and a multi-dimensional feature vector including time-frequency domain feature parameters such as Mel frequency cepstral coefficients (MFCC), spectral entropy, and zero-crossing rate is calculated.

[0011] S4. Feature Enhancement and Selection: Construct a feature correlation matrix and achieve feature dimensionality reduction through principal component analysis (PCA);

[0012] S5. Leakage Location: The signal delay between different sensors is calculated based on the generalized cross-correlation-phase transform (GCC-PHAT) algorithm, and the spatial coordinates of the leak point are solved using the particle swarm optimization (PSO) algorithm.

[0013] Preferred, including:

[0014] S6. Leakage Level Assessment: The extracted feature vectors are input into a Bidirectional Long Short-Term Memory (ABLSTM) network model based on an attention mechanism to assess the leakage level and output the leakage severity level.

[0015] Further preferred options include:

[0016] S7. Intelligent Diagnosis and Early Warning: It adopts the DS evidence theory to fuse the diagnostic results of multiple sensors and issues alarm signals of corresponding levels according to preset thresholds.

[0017] Preferably, in the signal preprocessing step S2, the variational mode decomposition (VMD) algorithm decomposes the signal into multiple intrinsic mode functions (IMFs) by minimizing the constrained variational problem. The specific parameters are set as follows: the number of decomposed modes K ranges from 5 to 8, and the penalty factor α ranges from 2000 to 3000.

[0018] Preferably, in the feature extraction step of S3, the calculation of the Mel frequency cepstral coefficients (MFCC) specifically uses 26 to 40 triangular filter banks to extract 12th to 13th order MFCC features.

[0019] Preferably, in the leak location step of S5, the inertia weight w of the particle swarm optimization (PSO) algorithm is set to 0.6 to 0.8, and the individual learning factor c1 and the social learning factor c2 are both set to 1.4 to 1.6.

[0020] Preferably, in the leakage assessment step of S6, the bidirectional long short-term memory network (ABLSTM) model based on the attention mechanism has a bidirectional LSTM layer containing 128 to 256 neurons, and the attention mechanism adopts an energy-based weight calculation method.

[0021] Secondly, embodiments of the present invention provide a high-pressure pipeline leakage detection device based on acoustic signature mapping, comprising:

[0022] The acoustic signal acquisition module is used to deploy a distributed microphone array along the high-pressure pipeline to acquire acoustic signals during pipeline operation.

[0023] The signal preprocessing module is used to preprocess the acquired acoustic signal by combining the variational mode decomposition (VMD) algorithm with the wavelet threshold noise reduction algorithm.

[0024] The feature extraction module is used to perform wavelet packet decomposition on the preprocessed signal, extract the energy features of each frequency band, and calculate a multi-dimensional feature vector including time-frequency domain feature parameters such as Mel frequency cepstral coefficients (MFCC), spectral entropy, and zero-crossing rate.

[0025] The feature enhancement and selection module is used to construct a feature correlation matrix and achieve feature dimensionality reduction through principal component analysis (PCA).

[0026] The leak location module is used to calculate the signal delay between different sensors based on the generalized cross-correlation-phase transform (GCC-PHAT) algorithm, and to solve for the spatial coordinates of the leak point using the particle swarm optimization (PSO) algorithm.

[0027] Preferably, it includes a leakage severity assessment module, which is used to input the extracted feature vector into a bidirectional long short-term memory network (ABLSTM) model based on an attention mechanism to assess the leakage severity and output the leakage severity level.

[0028] Preferably, it also includes: an intelligent diagnosis and early warning module, which uses DS evidence theory to fuse multi-sensor diagnostic results and issue alarm signals of corresponding levels according to preset thresholds.

[0029] Compared with the prior art, the beneficial results of the present invention are as follows:

[0030] This invention achieves a comprehensive breakthrough in detection performance, anti-interference capability, and intelligence level by constructing a multi-dimensional feature fusion recognition system based on acoustic signature. The minimum leakage detection rate is 0.01%, the positioning error is ≤±0.5m, the false alarm rate is <3% in a strong noise environment with a signal-to-noise ratio ≤5dB, the ABLSTM model achieves a classification accuracy of 99.2% for leakage degree, and supports wide temperature environment adaptation from -40℃ to 85℃. It provides a highly sensitive and reliable integrated solution for high-pressure pipeline leakage detection. Attached Figure Description

[0031] The accompanying drawings are included to provide a further understanding of the embodiments and are incorporated in and constitute a part of this specification. The drawings illustrate embodiments and, together with the description, serve to explain the principles of the invention. Other embodiments and many anticipated advantages of the embodiments will be readily recognized as they become better understood through reference to the following detailed description. Elements in the drawings are not necessarily to scale. The same reference numerals refer to corresponding similar parts.

[0032] Figure 1This is a schematic flowchart of a high-pressure pipeline leakage detection method based on acoustic signature mapping, according to an embodiment of the present invention.

[0033] Figure 2 This is a schematic diagram of the architecture of a high-pressure pipeline leakage detection device based on acoustic signature mapping, according to an embodiment of the present invention. Detailed Implementation

[0034] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it. Furthermore, it should be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings.

[0035] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0036] In a first aspect, embodiments of the present invention disclose a high-pressure pipeline leakage detection method based on acoustic signature mapping, such as... Figure 1 As shown, the method includes the following steps:

[0037] S1. Acoustic signal acquisition: Deploy a distributed microphone array along the high-pressure pipeline to acquire acoustic signals during pipeline operation;

[0038] Specifically, in one particular embodiment, a MEMS microphone array is constructed with a frequency response range of 10Hz-20kHz and a sensitivity of 120mV / Pa. The sensor spacing L is calculated as follows: L = v·Δt / 2, where the speed of sound v = 343m / s, and Δt is taken as the maximum measurable time delay of 0.02s, resulting in L = 3.43m, which is rounded to 3.5m in actual engineering.

[0039] Installation structure design: A three-way adjustable clamp is adopted to ensure that the sensor is at a 45° angle with the pipeline axis, thereby enhancing the ability to capture leakage sound waves.

[0040] S2. Signal preprocessing: Variational mode decomposition combined with wavelet threshold noise reduction algorithm is used to preprocess the acquired acoustic signal;

[0041] Specifically, the variational mode decomposition (VMD) algorithm decomposes a signal into multiple intrinsic mode functions (IMFs) by minimizing a constrained variational problem. The specific parameters are set as follows: the number of decomposed modes K ranges from 5 to 8, and the penalty factor α ranges from 2000 to 3000.

[0042] Variational mode decomposition (VMD) decomposes a signal into multiple intrinsic mode functions (IMFs) with different center frequencies by iteratively searching for the optimal solution of the variational model.

[0043] In this embodiment, the key parameter decomposition mode number K = 6, the penalty factor α = 2500, and the quadratic constraint Lagrange multiplier update parameter τ = 0.5.

[0044] The specific algorithm implementation code is as follows:

[0045]

[0046]

[0047] Furthermore, in wavelet thresholding, the wavelet basis is chosen as the Daubechies 4th order wavelet (db4); the threshold function is an improved semi-soft threshold function.

[0048] The specific implementation of the algorithm is as follows:

[0049]

[0050] S3. Feature Extraction: Wavelet packet decomposition is performed on the preprocessed signal to extract the energy features of each frequency band, and a multi-dimensional feature vector including time-frequency domain feature parameters such as Mel frequency cepstral coefficients (MFCC), spectral entropy, and zero-crossing rate is calculated.

[0051] Specifically, the calculation of Mel frequency cepstral coefficients (MFCC) uses 26 to 40 triangular filter banks to extract 12th to 13th order MFCC features.

[0052] Furthermore, the wavelet packet frequency band energy features are decomposed into 5 levels, resulting in 32 frequency bands; the energy is calculated as the sum of squares of the signal within each frequency band.

[0053] The implementation code is as follows:

[0054]

[0055] The number of triangular filters used for MFCC feature calculation is 32; the extraction order is: 13th order MFCC coefficients + 13th order first-order difference + 13th order second-order difference.

[0056] The specific implementation of the algorithm is as follows:

[0057]

[0058] S4. Feature Enhancement and Selection: Construct a feature correlation matrix and achieve feature dimensionality reduction through principal component analysis (PCA);

[0059] Specifically, the feature correlation matrix in this step is as follows: calculate the Pearson correlation coefficient; perform dimensionality reduction through principal component analysis (PCA): retain principal components with a cumulative variance contribution rate ≥ 95%.

[0060] The specific implementation code is as follows:

[0061]

[0062] S5. Leakage Location: The signal delay between different sensors is calculated based on the generalized cross-correlation-phase transform (GCC-PHAT) algorithm, and the spatial coordinates of the leak point are solved using the particle swarm optimization (PSO) algorithm.

[0063] Specifically, generalized cross-correlation (GCC) is a classic method for estimating time delay by analyzing the cross-correlation characteristics of signals acquired by two sensors, while phase transformation (PHAT) is a weighted optimization of GCC that can effectively improve the accuracy of time delay estimation in complex noise environments and is widely used in the field of sound source localization.

[0064] GCC-PHAT delay estimation uses a frequency weighting function: PHAT weighting; peak detection: parabolic interpolation method to improve accuracy.

[0065] The specific implementation code for this algorithm is as follows:

[0066]

[0067]

[0068] Furthermore, the inertia weight w of the particle swarm optimization (PSO) algorithm is set to 0.6–0.8, and the individual learning factor c1 and the social learning factor c2 are both set to 1.4–1.6.

[0069] In this embodiment, the particle swarm size is 50 particles; the inertia weight w = 0.7, the individual learning factor c1 = 1.5, and the social learning factor c2 = 1.5. The specific implementation code is as follows:

[0070]

[0071]

[0072]

[0073] In other preferred embodiments, the method further includes:

[0074] S6. Leakage Level Assessment: The extracted feature vectors are input into a Bidirectional Long Short-Term Memory (ABLSTM) network model based on an attention mechanism to assess the leakage level and output the leakage severity level.

[0075] Specifically, the attention-based bidirectional long short-term memory network (ABLSTM) model contains 128 to 256 neurons in the bidirectional LSTM layer, and the attention mechanism uses an energy-based weight calculation method.

[0076] In this embodiment, the ABLSTM model structure includes:

[0077] Input layer: 40-dimensional feature vectors;

[0078] Bidirectional LSTM layer: 256 neurons;

[0079] Attention layer: Calculates sequence weights;

[0080] Output layer: 5 levels of leakage severity classification (no leakage, minor leakage, moderate leakage, severe leakage, emergency leakage).

[0081]

[0082] Further training and optimization of the model:

[0083] Training data: 10,000 samples (including different leakage levels and operating conditions);

[0084] Data augmentation: Add Gaussian noise, temporal stretching / compression;

[0085] Training parameters: batch size 64, training epochs 100, early stopping strategy (patience=10).

[0086] The specific implementation code is as follows:

[0087]

[0088] S7. Intelligent Diagnosis and Early Warning: It adopts the DS evidence theory to fuse the diagnostic results of multiple sensors and issues alarm signals of corresponding levels according to preset thresholds.

[0089] Furthermore, the Dempster-Shafer (DS) evidence theory is an uncertainty reasoning method that achieves more reliable decisions than a single sensor by fusing information from multiple independent evidence sources. In high-pressure pipeline leak detection, this theory can effectively integrate the diagnostic results of distributed sensor arrays, reducing false alarm rates and improving detection reliability.

[0090] In practical engineering applications, by combining the leakage assessment of the ABLSTM model with the location results of the PSO algorithm, the DS evidence theory can achieve a leakage level correct classification rate of over 99%, significantly improving the level of pipeline safety monitoring.

[0091] This invention achieves high-precision detection and location of high-pressure pipeline leaks. In practical engineering applications, the minimum detectable leakage rate is 0.01%, the location error is ≤±0.5m, and the false alarm rate is <3% in complex industrial environments, which is significantly better than traditional detection methods.

[0092] Further reference Figure 2As an implementation of the methods shown in the above figures, this application provides an embodiment of a high-pressure pipeline leak detection device based on acoustic signature mapping. This device embodiment is similar to... Figure 1 Corresponding to the method embodiments shown, this device can be specifically applied to various electronic devices.

[0093] Secondly, embodiments of the present invention also disclose a high-pressure pipeline leakage detection device based on acoustic signature mapping, such as... Figure 2 As shown, the device includes: an acoustic signal acquisition module 21, a signal preprocessing module 22, a feature extraction module 23, a feature enhancement and selection module 24, and a leak location module 25.

[0094] In one specific embodiment, the acoustic signal acquisition module 21 is used to deploy a distributed microphone array along the high-pressure pipeline to acquire acoustic signals during pipeline operation; the signal preprocessing module 22 is used to preprocess the acquired acoustic signals using variational mode decomposition combined with wavelet threshold noise reduction algorithm; the feature extraction module 23 is used to perform wavelet packet decomposition on the preprocessed signal, extract the energy features of each frequency band, and calculate a multidimensional feature vector including time-frequency domain feature parameters such as Mel frequency cepstral coefficients (MFCC), spectral entropy, and zero-crossing rate.

[0095] The feature enhancement and selection module 24 is used to construct a feature correlation matrix and achieve feature dimensionality reduction through principal component analysis (PCA); the leak location module 25 is used to calculate the signal delay between different sensors based on the generalized cross-correlation-phase transform (GCC-PHAT) algorithm and solve for the spatial coordinates of the leak point using the particle swarm optimization (PSO) algorithm.

[0096] Preferably, it also includes: a leakage assessment module 26, which is used to input the extracted feature vector into a bidirectional long short-term memory network (ABLSTM) model based on an attention mechanism to assess the leakage level and output the leakage severity level.

[0097] Preferably, it also includes: an intelligent diagnosis and early warning module 27, which is used to fuse multi-sensor diagnostic results using DS evidence theory and issue alarm signals of corresponding levels according to preset thresholds.

[0098] The functions and methods of the above modules correspond to each other, and will not be repeated here.

[0099] This invention achieves a comprehensive breakthrough in detection performance, anti-interference capability, and intelligence level by constructing a multi-dimensional feature fusion recognition system based on acoustic signature. The minimum leakage detection rate is 0.01%, the positioning error is ≤±0.5m, the false alarm rate is <3% in a strong noise environment with a signal-to-noise ratio ≤5dB, the ABLSTM model achieves a classification accuracy of 99.2% for leakage degree, and supports adaptive wide temperature environment from -40℃ to 85℃. It provides a highly sensitive and reliable integrated solution for high-pressure pipeline leakage detection.

[0100] The above description is merely a preferred embodiment of the present invention and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention is not limited to the specific combination of the above-described technical features, but also includes other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the inventive concept. For example, technical solutions formed by substituting the above-described features with (but not limited to) technical features with similar functions disclosed in this invention.

Claims

1. A method for detecting leaks in high-pressure pipelines based on acoustic signature mapping, characterized in that, The method includes the following steps: S1. Acoustic signal acquisition: Deploy a distributed microphone array along the high-pressure pipeline to acquire acoustic signals during pipeline operation; S2. Signal preprocessing: The acquired acoustic signal is preprocessed using the variational mode decomposition (VMD) algorithm combined with the wavelet threshold noise reduction algorithm. S3. Feature Extraction: Perform wavelet packet decomposition on the preprocessed signal to extract the energy features of each frequency band, and calculate a multi-dimensional feature vector including time-frequency domain feature parameters such as Mel frequency cepstral coefficients (MFCC), spectral entropy, and zero-crossing rate. S4. Feature Enhancement and Selection: Construct a feature correlation matrix and achieve feature dimensionality reduction through principal component analysis (PCA). S5. Leakage Location: The signal delay between different sensors is calculated based on the generalized cross-correlation-phase transform (GCC-PHAT) algorithm, and the spatial coordinates of the leak point are solved using the particle swarm optimization (PSO) algorithm.

2. The high-pressure pipeline leakage detection method based on acoustic signature mapping according to claim 1, characterized in that, include: S6. Leakage severity assessment: The extracted feature vectors are input into the attention-based bidirectional long short-term memory network (ABLSTM) model to assess the leakage severity and output the leakage severity level.

3. The high-pressure pipeline leakage detection method based on acoustic signature mapping according to claim 2, characterized in that, Also includes: S7. Intelligent Diagnosis and Early Warning: It adopts the DS evidence theory to fuse the diagnostic results of multiple sensors and issues alarm signals of corresponding levels according to preset thresholds.

4. The high-pressure pipeline leakage detection method based on acoustic signature mapping according to claim 1, characterized in that, In the signal preprocessing step of S2, the variational mode decomposition (VMD) algorithm decomposes the signal into multiple intrinsic mode functions (IMFs) by minimizing the constrained variational problem. The specific parameters are set as follows: the number of decomposed modes K ranges from 5 to 8, and the penalty factor α ranges from 2000 to 3000.

5. The high-pressure pipeline leakage detection method based on acoustic signature mapping according to claim 1, characterized in that, In the feature extraction step of S3, the calculation of the Mel frequency cepstral coefficients (MFCCs) specifically uses 26 to 40 triangular filter banks to extract MFCC features of order 12 to 13.

6. The high-pressure pipeline leakage detection method based on acoustic signature mapping according to claim 1, characterized in that, In the leakage localization step of S5, the inertial weight w of the Particle Swarm Optimization (PSO) algorithm is set to 0.6–0.8, and the individual learning factor c1 and the social learning factor c2 are both set to 1.4–1.

6.

7. The high-pressure pipeline leakage detection method based on acoustic signature mapping according to claim 2, characterized in that, In the leakage assessment step of S6, the bidirectional long short-term memory network (ABLSTM) model based on the attention mechanism, the bidirectional LSTM layer contains 128 to 256 neurons, and the attention mechanism adopts an energy-based weight calculation method.

8. A high-pressure pipeline leak detection device based on acoustic signature mapping, characterized in that, include: The acoustic signal acquisition module is used to deploy a distributed microphone array along the high-pressure pipeline to acquire acoustic signals during pipeline operation. The signal preprocessing module is used to preprocess the acquired acoustic signal by combining the variational mode decomposition (VMD) algorithm with the wavelet threshold noise reduction algorithm. The feature extraction module is used to perform wavelet packet decomposition on the preprocessed signal, extract the energy features of each frequency band, and calculate a multi-dimensional feature vector including time-frequency domain feature parameters such as Mel frequency cepstral coefficients (MFCC), spectral entropy, and zero-crossing rate. The feature enhancement and selection module is used to construct the feature correlation matrix and achieve feature dimensionality reduction through principal component analysis (PCA). The leak location module is used to calculate the signal delay between different sensors based on the generalized cross-correlation-phase transform (GCC-PHAT) algorithm, and to solve for the spatial coordinates of the leak point using the particle swarm optimization (PSO) algorithm.

9. The high-pressure pipeline leakage detection device based on acoustic signature as described in claim 8, characterized in that, include: The leakage severity assessment module is used to input the extracted feature vectors into the attention-based bidirectional long short-term memory network (ABLSTM) model to assess the leakage severity and output the leakage severity level.

10. The high-pressure pipeline leakage detection device based on acoustic signature as described in claim 9, characterized in that, Also includes: The intelligent diagnosis and early warning module is used to fuse multi-sensor diagnostic results using DS evidence theory and issue alarm signals of corresponding levels according to preset thresholds.

Images