Pipeline leakage crack morphology identification method, device and equipment and readable storage medium

By enhancing wavelet time-frequency map feature extraction and fusing multi-scale convolutional neural networks, the problem of misjudgment in pipeline leak identification under strong background noise was solved, and high-accuracy pipeline leak crack morphology identification was achieved.

CN119004068AActive Publication Date: 2024-11-22XI AN JIAOTONG UNIV

Patent Information

Application Number
CN202411024460.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-29
Publication Date
2024-11-22
Estimated Expiration
2044-07-29

AI Technical Summary

Technical Problem

Existing technologies struggle to accurately identify pipeline leaks under strong background noise, and insufficient multi-scale information fusion leads to a high misjudgment rate.

Method used

An enhanced wavelet time-frequency map feature extraction method is adopted, combined with an improved remapping function for fast local Laplacian filtering, and a multi-scale convolutional neural network (MSCA-DAA) is used for feature fusion, including a multi-scale convolutional module, a dimension-aware attention module, and a global self-attention layer, to improve the model's resolution.

Benefits of technology

It effectively reduced the false positive rate of pipeline leak identification and improved the identification accuracy under strong background noise, reaching an identification accuracy rate of 98.22%.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119004068A_ABST
    Figure CN119004068A_ABST
Patent Text Reader

Abstract

The invention discloses a pipeline leakage crack morphology identification method, device and equipment and a readable storage medium, and the method comprises the steps: carrying out the quick local Laplace filtering of a wavelet time-frequency diagram through an improved remapping function, and obtaining an enhanced wavelet time-frequency diagram; inputting the enhanced wavelet time-frequency graph into a pre-trained pipeline leakage crack morphology identification model to obtain a pipeline leakage crack morphology; wherein the pipeline leakage crack morphology identification model is obtained by training a convolutional neural network based on MSCA-DAA by using an enhanced wavelet time-frequency graph and a corresponding leakage crack morphology as training samples; the convolutional neural network based on the MSCA-DAA comprises a multi-scale convolution module, a dimension perception attention module and a global self-attention layer which are connected in sequence. According to the method, the fault frequency band of the wavelet time-frequency diagram can be highlighted, the noise interference information can be smoothed, and the resolution capability of a subsequent model is improved, so that misjudgment is reduced, and the accuracy and efficiency of the model are considered.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the field of pipeline leakage detection, and in particular relates to a pipeline leakage crack morphology recognition method, device, equipment and a readable storage medium. Background Art

[0002] Water supply pipelines are an important part of the water supply system. Welding is a common method for connecting pipelines. Although it has strict design, manufacturing and installation standards, due to the uneven chemical composition and unstable mechanical properties of the welded joints, leakage may occur during long-term service due to external interference, corrosion, pipe materials and construction quality, leading to accidents such as fire and explosion, causing significant economic losses, casualties and environmental pollution. Therefore, the problem of online monitoring of pressure pipeline leakage needs to be solved urgently.

[0003] Under the conditions of strong background noise and weak leakage rate, it is necessary to improve the characterization ability of the proposed features under the conditions of weak leakage and strong background noise, and explore the applicable methods of pipeline leakage detection under various conditions. The patent with publication number CN105864643A discloses a gas pipeline leakage positioning experimental device and method based on RBF neural network, which constructs feature vectors under different leakage pressures by obtaining the maximum peak value of the leakage signal, the energy of the (3, 0) node wavelet packet, and the power spectrum peak at the corresponding leakage characteristic frequency, thereby realizing the identification of pipeline leakage. The patent with publication number CN118293367A discloses a gas pipeline working condition discrimination method based on multi-scale deep convolution autoencoder, which uses a multi-scale deep convolution layer for feature extraction, and uses the combination of multiple convolution cores at various sizes to perform convolution and operation on the features, making the feature extraction more flexible and comprehensive. There are the following problems with both: 1. Simple time-frequency domain signals are difficult to reflect the essential characteristics of leakage signals under strong background noise, which can easily lead to misjudgment; 2. The existing network model only focuses on the extraction of multi-scale information, but ignores the problems caused by the subsequent fusion of information of different scales in the network. Summary of the invention

[0004] In view of the problems existing in the prior art, the present invention provides a pipeline leakage crack morphology recognition method, device, equipment and readable storage medium, and proposes a feature extraction method for enhanced wavelet time-frequency diagram, which can highlight the fault frequency band of the wavelet time-frequency diagram, smooth the noise interference information, and improve the resolution ability of the subsequent model, thereby reducing misjudgment; finally, in order to take into account the accuracy and efficiency of the model, a multi-scale convolutional neural network with dimension-aware attention (MSCA-DAA) is designed.

[0005] In order to solve the above technical problems, the present invention is implemented by the following technical solutions:

[0006] According to a first aspect of the present invention, a pipeline leakage crack morphology identification method is provided, comprising:

[0007] Acquire the acoustic emission signal generated when there is leakage at the crack of the pipeline;

[0008] Downsampling the acoustic emission signal to obtain a downsampled acoustic emission signal;

[0009] Extracting a wavelet time-frequency diagram of the downsampled acoustic emission signal to obtain a wavelet time-frequency diagram of the acoustic emission signal;

[0010] Using an improved remapping function, fast local Laplace filtering is performed on the wavelet time-frequency diagram to obtain an enhanced wavelet time-frequency diagram;

[0011] The enhanced wavelet time-frequency graph is input into a pre-trained pipeline leakage crack morphology recognition model to obtain the pipeline leakage crack morphology; wherein the pipeline leakage crack morphology recognition model is obtained by training a convolutional neural network based on MSCA-DAA using the enhanced wavelet time-frequency graph and its corresponding leakage crack morphology as training samples, and the convolutional neural network based on MSCA-DAA includes a multi-scale convolution module, a dimension-aware attention module and a global self-attention layer connected in sequence, and the dimension-aware attention module is used to realize the average pooling operation along the vertical dimension and the horizontal dimension, the average pooling operation along the channel dimension and the horizontal dimension, and the average pooling operation along the channel dimension and the vertical dimension, and finally the information of the channel dimension, the vertical dimension and the horizontal dimension is weightedly fused.

[0012] In a possible implementation manner of the first aspect, the improved remapping function is:

[0013] r(i)=i-(ig)(-m factor G σ (ig) α )

[0014] Where r(i) is the improved remapping function; G σ (ig) is the Gaussian function when the intensity threshold is σ, where g is the coefficient of the Gaussian pyramid, i is the coefficient of the input image at pixel (x, y), and G σ (ig) = exp(-(ig) 2 / 2σ 2 );m factor is the amplification factor and α is the control coefficient.

[0015] In a possible implementation manner of the first aspect, the use of an improved remapping function to perform fast local Laplace filtering on the wavelet time-frequency graph is specifically as follows:

[0016] a. Use point-by-point nonlinear function to obtain improved Gaussian pyramid coefficient g = G l [I](x,y), where I is the input image, l is the number of layers in the Gaussian pyramid, G l It is the Gaussian function corresponding to the Gaussian pyramid layer number l;

[0017] b. Obtain the intermediate image r(I g ), where for different g, different intermediate images are obtained;

[0018] c. Obtain the result image L l [O](x,y), where L is the Laplace operation and O is the output image.

[0019] In a possible implementation manner of the first aspect, the multi-scale convolution module includes convolution kernels of three scales: 3×3, 5×5, and 7×7, and performs a concatenation operation on the same input after completing three convolutions of different scales.

[0020] In a possible implementation of the first aspect, the multi-scale convolution module is used to extract features from the enhanced wavelet time-frequency map, the dimension-aware attention module is used to perform weighted fusion on the extracted features to obtain fused features, and the global self-attention layer is used to perform global self-attention calculation on the fused features.

[0021] In a possible implementation manner of the first aspect, extracting a wavelet time-frequency graph of the downsampled acoustic emission signal is specifically:

[0022] The downsampled acoustic emission signal is subjected to continuous wavelet transform using a wavelet basis function to obtain a wavelet time-frequency diagram of the acoustic emission signal.

[0023] According to a second aspect of the present invention, there is provided a pipeline leakage crack morphology identification device, comprising:

[0024] An acquisition module is used to acquire the acoustic emission signal generated when there is leakage at the crack of the pipeline;

[0025] A downsampling module is used to downsample the acoustic emission signal to obtain a downsampled acoustic emission signal;

[0026] An extraction module, used for extracting a wavelet time-frequency diagram of the downsampled acoustic emission signal to obtain a wavelet time-frequency diagram of the acoustic emission signal;

[0027] A filtering module, used for performing fast local Laplace filtering on the wavelet time-frequency diagram using an improved remapping function to obtain an enhanced wavelet time-frequency diagram;

[0028] An identification module is used to input the enhanced wavelet time-frequency diagram into a pre-trained pipeline leakage crack morphology recognition model to obtain the pipeline leakage crack morphology; wherein the pipeline leakage crack morphology recognition model is obtained by training a convolutional neural network based on MSCA-DAA using the enhanced wavelet time-frequency diagram and its corresponding leakage crack morphology as training samples, and the convolutional neural network based on MSCA-DAA includes a multi-scale convolution module, a dimension-aware attention module and a global self-attention layer connected in sequence, and the dimension-aware attention module is used to implement average pooling operations along vertical and horizontal dimensions, average pooling operations along channel dimensions and horizontal dimensions, and average pooling operations along channel dimensions and vertical dimensions, and finally weightedly fuse the information of channel dimensions, vertical dimensions and horizontal dimensions.

[0029] According to a third aspect of the present invention, there is provided a device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the pipeline leakage crack morphology identification method when executing the computer program.

[0030] According to a fourth aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the pipeline leakage crack morphology identification method are implemented.

[0031] Compared with the prior art, the present invention has at least the following beneficial effects:

[0032] The present invention provides a pipeline leakage crack morphology recognition method, which extracts wavelet time-frequency diagrams from downsampled acoustic emission signals, and uses an improved remapping function to perform fast local Laplace filtering on the wavelet time-frequency diagrams. The fast local Laplace filtering retains edges and enhances contour information details to obtain enhanced wavelet time-frequency diagrams, thereby providing a more reliable input for a pipeline leakage crack morphology recognition model. At the same time, the fast local Laplace filtering effectively enhances the contrast of the wavelet time-frequency diagram, making the information of the fault frequency band more prominent, thereby reducing the misjudgment of the pipeline leakage crack morphology recognition model. Regarding the pipeline leakage crack morphology recognition model, firstly, multi-scale convolution can extract feature information at different scales, paving the way for the subsequent global self-attention layer interaction; secondly, in order to fuse information of different dimensions, a dimension-aware attention module is designed, and information at different dimensions is adjusted after each multi-scale convolution, which can fuse information of three different dimensions and ensure that the output size is consistent with the input size, successfully solving the problem caused by scale inconsistency; finally, in order to give full play to the advantages of self-attention, after dimensional perception, a global self-attention layer is added to realize the interaction of global information. The pipeline leakage crack morphology recognition model can greatly improve the accuracy of pipeline leakage crack morphology recognition.

[0033] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the technical solutions in the specific implementation modes of the present invention, the drawings required for use in the description of the specific implementation modes will be briefly introduced below. Obviously, the drawings described below are some implementation modes of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0035] Figure 1 The present invention is a flow chart of a pipeline leakage crack morphology identification method.

[0036] Figure 2 Schematic diagram of the acoustic emission signal acquisition system in the embodiment.

[0037] Figure 3 Figure 2 is a pipeline diagram of cracks with different morphologies in the embodiments.

[0038] Figure 4a It is the wavelet time-frequency diagram of the zigzag crack pipeline leakage in the embodiment, Figure 4b It is the enhanced wavelet time-frequency diagram of the zigzag crack pipeline leakage in the embodiment.

[0039] Figure 5This is a structural diagram of a convolutional neural network based on MSCA-DAA in the embodiment.

[0040] In the figure: 1-acoustic emission sensor; 2-preamplifier; 3-data acquisition card; 4-host computer; 5-weld crack pipe with different crack morphologies. DETAILED DESCRIPTION

[0041] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0042] like Figure 1 As shown, an embodiment of the present invention provides a pipeline leakage crack morphology recognition method, which specifically includes the following steps:

[0043] S1. Obtain the acoustic emission signal generated when the pipeline leaks at the crack.

[0044] Specifically, a professional acoustic emission detection system is configured to capture the acoustic emission signal when the pipeline cracks leak. The system includes a high-sensitivity acoustic emission sensor, a data acquisition unit (DAQ), and a signal preprocessing module. The acoustic emission sensor needs to be installed at the pipeline to be detected to accurately capture the weak acoustic emission signal generated when the leak occurs.

[0045] Start the acoustic emission detection system to continuously monitor the pipeline area suspected of having cracks. When the system detects an abnormal acoustic emission signal (i.e., the signal generated when the pipeline cracks and leaks), it immediately triggers the data acquisition mechanism to record and save the complete acoustic emission signal data.

[0046] S2. Downsampling the acoustic emission signal to obtain a downsampled acoustic emission signal.

[0047] Specifically, the original acoustic emission signal is downsampled to reduce the amount of data and improve the efficiency of subsequent processing. The specific method of downsampling can be selected according to the signal characteristics and system requirements, such as downsampling according to Shannon's theorem to obtain an undistorted acoustic emission signal, that is, to obtain a downsampled acoustic emission signal.

[0048] S3, extracting a wavelet time-frequency diagram of the downsampled acoustic emission signal to obtain a wavelet time-frequency diagram of the acoustic emission signal.

[0049] Specifically, the downsampled acoustic emission signal is subjected to continuous wavelet transform (CWT) to obtain the energy distribution of the signal at different time and frequency scales. The appropriate wavelet basis function is selected to ensure that the key features of the signal can be fully captured. The result of the wavelet transform is converted into a time-frequency graph, that is, a three-dimensional image of time-frequency-energy density. The time-frequency graph can intuitively display the frequency components and energy intensity of the signal changing over time, providing a basis for subsequent feature extraction and recognition.

[0050] S4. Using an improved remapping function, fast local Laplace filtering is performed on the wavelet time-frequency diagram to obtain an enhanced wavelet time-frequency diagram.

[0051] Specifically, the generated wavelet time-frequency image is processed by fast local Laplace filtering using the improved remapping function, aiming to enhance the edge and texture information in the image, improve the image contrast and clarity, and facilitate subsequent crack morphology recognition. The best image enhancement effect can be achieved by adjusting the filtering parameters.

[0052] The improved remapping function is specifically:

[0053] r(i)=i-(ig)(-m factor G σ (ig) α )

[0054] Where r(i) is the improved remapping function; G σ (ig) is the Gaussian function when the intensity threshold is σ, where g is the coefficient of the Gaussian pyramid, i is the coefficient of the input image at pixel (x, y), and G σ (ig) = exp(-(ig) 2 / 2σ 2 );m factor is the amplification factor and α is the control coefficient.

[0055] S5, inputting the enhanced wavelet time-frequency graph into a pre-trained pipeline leakage crack morphology recognition model to obtain the pipeline leakage crack morphology. The pipeline leakage crack morphology recognition model is obtained by training a convolutional neural network based on MSCA-DAA using the enhanced wavelet time-frequency graph and its corresponding leakage crack morphology as training samples.

[0056] Combination Figure 5As shown, the convolutional neural network based on MSCA-DAA includes a multi-scale convolution module, a dimension-aware attention module and a global self-attention layer connected in sequence. The dimension-aware attention module is used to implement average pooling operations along vertical and horizontal dimensions (VH Avg pool), average pooling operations along channel and horizontal dimensions (CH Avgpool), and average pooling operations along channel and vertical dimensions (CV Avg pool), and finally solves the problem of semantic inconsistency caused by convolution operations of different scales by weighted fusion of information of channel dimension, vertical dimension and horizontal dimension.

[0057] like Figure 5 As shown, the multi-scale convolution module includes convolution kernels of three scales: 3×3, 5×5 and 7×7. After completing three convolutions of different scales on the same input, a concatenation operation is performed on it. The multi-scale convolution module is used to extract features from the enhanced wavelet time-frequency graph. The dimension-aware attention module is used to perform weighted fusion on the extracted features to obtain fused features. The global self-attention layer is used to perform global self-attention calculation on the fused features, specifically: for a given pixel point First extract a ij The local area ab∈X with a spatial range of k is centered k (i,j), called a memory block, the single-head attention calculation for pixel output is as follows:

[0058]

[0059] Among them, q ij =W Q x ij , k ab =W K x ab , v ab =W V x ab It is the linear transformation of the pixel at position ij and the neighboring pixel; softmax ab represents the softmax calculated for all logits in the ij neighborhood; They are all learning transformations, and their essence is similar to the variable weights of convolution kernels.

[0060] Specifically, the pre-training of the pipeline leakage crack morphology recognition model is as follows: a series of enhanced wavelet time-frequency graphs and their corresponding pipeline leakage crack morphology data are prepared in advance as training samples. Then, a convolutional neural network model based on MSCA-DAA (Multi-Scale Channel Attention with Dense Aggregation and AdaptiveActivation) is constructed. The model improves the feature representation ability and recognition accuracy through multi-scale feature extraction, channel attention mechanism, dense connection and adaptive activation function strategy; the convolutional neural network based on MSCA-DAA is trained using training samples until the predetermined performance indicators are achieved.

[0061] After the model training is completed, the enhanced wavelet time-frequency image is input into the trained pipeline leakage crack morphology recognition model. The model automatically extracts the key features in the image and classifies and recognizes them, and finally outputs the morphology description of the pipeline leakage crack. The morphology description can include specific parameters such as the length, width, depth, and direction of the crack, as well as information such as the geometric shape and distribution characteristics of the crack.

[0062] In one possible implementation, the improved remapping function is used to perform fast local Laplace filtering on the wavelet time-frequency diagram, specifically:

[0063] a. Use point-by-point nonlinear function to obtain improved Gaussian pyramid coefficient g = G l [I](x,y), where I is the input image, l is the number of layers in the Gaussian pyramid, G l It is the Gaussian function corresponding to the Gaussian pyramid layer number l;

[0064] b. Obtain the intermediate image r(I g ), where for different g, different intermediate images are obtained;

[0065] c. Obtain the result image L l [O](x,y), where L is the Laplace operation and O is the output image.

[0066] In one achievable method, the wavelet time-frequency diagram of the downsampled acoustic emission signal is extracted, specifically: considering the non-stationary and multi-scale characteristics of the pipeline leakage acoustic emission signal, the Daubechies (dbN) series wavelet is used as the basis function. Specifically, the db4 wavelet can be selected because it has a good balance between the time domain and the frequency domain, which can not only capture the detailed information of the signal, but also maintain a certain smoothness, which is helpful for subsequent feature extraction and recognition. Using the selected db4 wavelet basis, the downsampled acoustic emission signal is subjected to a continuous wavelet transform, and a set of coefficients at different scales (frequency) and translations (time) are obtained through the continuous wavelet transform. These coefficients reflect the energy distribution of the signal at different times and frequencies. By visualizing these coefficients, the wavelet time-frequency diagram of the acoustic emission signal can be obtained. The generated wavelet time-frequency diagram can clearly show the time-frequency characteristics of the pipeline leakage acoustic emission signal, including the time when the leakage event occurs, the frequency distribution range, and the energy intensity, etc. These characteristic information provides an important basis for feature extraction and crack morphology recognition.

[0067] Exemplarily, the wavelet basis function formula is expressed as:

[0068]

[0069] Where a is the scaling factor, τ is the translation factor, and ψ a,τ (t) is the basis wavelet function with parameters a and τ, ψ a,τ (t) is a continuous basis wavelet function.

[0070] The downsampled acoustic emission signal is subjected to continuous wavelet transform, which can be expressed as:

[0071]

[0072] Where f(t) is the downsampled acoustic emission signal, WT f (a,τ) are the coefficients of the wavelet transform.

[0073] Example of implementation:

[0074] like Figure 2 The figure shows a schematic diagram of acoustic emission signal acquisition for pipeline weld crack leakage in this embodiment, which mainly includes an acoustic emission sensor 1, a preamplifier 2, a data acquisition card 3, a host computer 4 and a weld crack pipeline 5 with different crack morphologies, wherein the acoustic emission sensor 1, the preamplifier 2, the data acquisition card 3, and the host computer 4 are sequentially connected to form an acoustic emission signal acquisition system, and the acoustic emission signal acquisition system collects acoustic emission signals in real time during the pipeline leakage process.

[0075] In this embodiment, the acoustic emission sensor 1 is 100 mm away from the leakage center. The weld crack pipe 5 with different crack morphologies used has a diameter of 60 mm and a wall thickness of 4 mm. It is formed by butt welding two sections of 50 cm galvanized steel pipes. Flange connections are adopted at both ends of the pipe. A coupling agent is used to ensure that the acoustic emission sensor 1 fits tightly with the weld crack pipe 5 with different crack morphologies, and tape is used to ensure that the acoustic emission sensor 1 does not move during the leakage of the weld crack pipe 5 with different crack morphologies. The sampling rate is 1 MHz.

[0076] In this embodiment, three pipes with different crack morphologies are used, namely, axial crack, flexural crack and circumferential crack. Figure 3 shown.

[0077] Step 1: Use Figure 2 The provided pipeline weld crack leakage acoustic emission monitoring system collects pipeline leakage acoustic emission signals under different crack morphologies.

[0078] Step 2: Taking the flexural crack as an example, the collected acoustic emission signal is downsampled and the downsampled acoustic emission signal is subjected to wavelet transform to obtain the following Figure 4a The wavelet time-frequency diagram is shown.

[0079] Step 3: Use the improved remapping function to perform fast local Laplace filtering on the obtained wavelet time-frequency diagram to obtain the enhanced wavelet time-frequency diagram, such as Figure 4b shown.

[0080] Step 4: Repeat steps 1 to 3 to establish pipeline leakage enhanced wavelet time-frequency map data sets under three different crack morphologies.

[0081] Step 5: Input the enhanced wavelet time-frequency image data set obtained in step 4 into the pipeline leakage crack morphology recognition model. Figure 5 The convolutional neural network based on MSCA-DAA shown in the figure is trained to complete the classification and recognition of input features, with the highest accuracy rate of 98.22%.

[0082] From the above experiments, the present invention can make full use of the information of the acoustic emission signal to extract the essential characteristics of the pipeline leakage acoustic emission signal. The information of the fault frequency band has been significantly enhanced, and the information of the non-fault frequency band has been significantly suppressed and smoothed. Combined with the convolutional neural network based on MSCA-DAA, an accurate recognition model is constructed to achieve accurate recognition of the pipeline leakage acoustic emission signal under different working conditions, thereby providing an accurate and reliable method for pipeline leakage crack morphology identification. Specifically, the present invention makes full use of the acoustic information of the acoustic emission signal, uses an improved remapping function, performs fast local Laplace filtering on the wavelet time-frequency graph, and uses enhanced wavelet time-frequency graphs to realize feature extraction and identification of pipeline leakage with different morphological cracks, with an accuracy rate of up to 98.22%. The method proposed by the present invention is simple, the physical meaning of the features is clear, and it has high reliability and robustness, providing an effective method for online positioning and identification of pipeline leakage crack morphology.

[0083] In one embodiment of the present invention, a pipeline leakage crack morphology recognition device is provided, comprising:

[0084] The acquisition module is used to acquire the acoustic emission signal generated when the pipeline crack leaks.

[0085] The downsampling module is used to perform downsampling processing on the acoustic emission signal to obtain a downsampled acoustic emission signal.

[0086] The extraction module is used to extract the wavelet time-frequency diagram of the downsampled acoustic emission signal to obtain the wavelet time-frequency diagram of the acoustic emission signal.

[0087] The filtering module is used to perform fast local Laplace filtering on the wavelet time-frequency diagram using an improved remapping function to obtain an enhanced wavelet time-frequency diagram.

[0088] An identification module is used to input the enhanced wavelet time-frequency diagram into a pre-trained pipeline leakage crack morphology recognition model to obtain the pipeline leakage crack morphology; wherein the pipeline leakage crack morphology recognition model is obtained by training a convolutional neural network based on MSCA-DAA using the enhanced wavelet time-frequency diagram and its corresponding leakage crack morphology as training samples, and the convolutional neural network based on MSCA-DAA includes a multi-scale convolution module, a dimension-aware attention module and a global self-attention layer connected in sequence, and the dimension-aware attention module is used to implement average pooling operations along vertical and horizontal dimensions, average pooling operations along channel dimensions and horizontal dimensions, and average pooling operations along channel dimensions and vertical dimensions, and finally weightedly fuse the information of channel dimensions, vertical dimensions and horizontal dimensions.

[0089] All relevant contents of each step involved in the aforementioned embodiment of a pipeline leakage crack morphology identification method can be referred to the functional description of the functional module corresponding to a pipeline leakage crack morphology identification device in the embodiment of the present invention, and will not be repeated here. The division of modules in the embodiment of the present invention is schematic, which is only a logical function division. There may be other division methods in actual implementation. In addition, each functional module in each embodiment of the present invention can be integrated into a processor, or it can exist physically separately, or two or more modules can be integrated into one module. The above-mentioned integrated module can be implemented in the form of hardware or in the form of software functional modules.

[0090] In another embodiment of the present invention, a computer device is provided, the computer device includes a processor and a memory, the memory is used to store a computer program, the computer program includes program instructions, and the processor is used to execute the program instructions stored in the computer storage medium. The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. It is the computing core and control core of the terminal, which is suitable for implementing one or more instructions, and is specifically suitable for loading and executing one or more instructions in the computer storage medium to implement the corresponding method flow or corresponding function; the processor described in the embodiment of the present invention can be used for the operation of a pipeline leakage crack morphology recognition method.

[0091] In another embodiment of the present invention, the present invention also provides a storage medium, specifically a computer-readable storage medium (Memory), which is a memory device in a computer device for storing programs and data. It is understandable that the computer-readable storage medium here can include both built-in storage media in the computer device and, of course, extended storage media supported by the computer device. The computer-readable storage medium provides a storage space, which stores the operating system of the terminal. In addition, one or more instructions suitable for being loaded and executed by the processor are also stored in the storage space, and these instructions can be one or more computer programs (including program codes). It should be noted that the computer-readable storage medium here can be a high-speed RAM memory or a non-volatile memory, such as at least one disk memory. The processor can load and execute one or more instructions stored in the computer-readable storage medium to implement the corresponding steps of the above-mentioned embodiment of a pipeline leakage crack morphology recognition method.

[0092] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0093] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0094] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0095] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0096] In the present invention, the terms "one embodiment", "some embodiments", "examples", "specific examples", or "some examples" etc. mean that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the described specific features, structures, materials or characteristics may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0097] Finally, it should be noted that the above-described embodiments are only specific implementations of the present invention, which are used to illustrate the technical solutions of the present invention, rather than to limit them. The protection scope of the present invention is not limited thereto. Although the present invention is described in detail with reference to the above-described embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the above-described embodiments within the technical scope disclosed by the present invention, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A pipeline leakage crack morphology identification method, characterized in that: include: Acquire the acoustic emission signal generated when there is leakage at the crack of the pipeline; Downsampling the acoustic emission signal to obtain a downsampled acoustic emission signal; Extracting a wavelet time-frequency diagram of the downsampled acoustic emission signal to obtain a wavelet time-frequency diagram of the acoustic emission signal; Using an improved remapping function, fast local Laplace filtering is performed on the wavelet time-frequency diagram to obtain an enhanced wavelet time-frequency diagram; The enhanced wavelet time-frequency graph is input into a pre-trained pipeline leakage crack morphology recognition model to obtain the pipeline leakage crack morphology; wherein the pipeline leakage crack morphology recognition model is obtained by training a convolutional neural network based on MSCA-DAA using the enhanced wavelet time-frequency graph and its corresponding leakage crack morphology as training samples, and the convolutional neural network based on MSCA-DAA includes a multi-scale convolution module, a dimension-aware attention module and a global self-attention layer connected in sequence, and the dimension-aware attention module is used to realize the average pooling operation along the vertical dimension and the horizontal dimension, the average pooling operation along the channel dimension and the horizontal dimension, and the average pooling operation along the channel dimension and the vertical dimension, and finally the information of the channel dimension, the vertical dimension and the horizontal dimension is weightedly fused.

2. A pipeline leakage crack morphology identification method according to claim 1, characterized in that: The improved remapping function is: r(i)=i-(i-g)(-m factor G σ (i-g) α ) Where r(i) is the improved remapping function; G σ (ig) is the Gaussian function when the intensity threshold is σ, where g is the coefficient of the Gaussian pyramid, i is the coefficient of the input image at pixel (x, y), and G σ (ig) = exp(-(ig) 2 / 2σ 2 );m factor is the amplification factor and α is the control coefficient.

3. A pipeline leakage crack morphology identification method according to claim 2, characterized in that: The improved remapping function is used to perform fast local Laplace filtering on the wavelet time-frequency diagram, specifically: a. Use point-by-point nonlinear function to obtain improved Gaussian pyramid coefficient g = G l [I](x,y), where I is the input image, l is the number of layers in the Gaussian pyramid, G l It is the Gaussian function corresponding to the Gaussian pyramid layer number l; b. Obtain the intermediate image r(I g ), where for different g, different intermediate images are obtained; c. Obtain the result image L l [O](x,y), where L is the Laplace operation and O is the output image.

4. A pipeline leakage crack morphology identification method according to claim 1, characterized in that: The multi-scale convolution module includes convolution kernels of three scales: 3×3, 5×5, and 7×7. After completing three convolutions of different scales on the same input, a concatenation operation is performed on it.

5. A pipeline leakage crack morphology identification method according to claim 1, characterized in that: The multi-scale convolution module is used to extract features from the enhanced wavelet time-frequency graph, the dimension-aware attention module is used to perform weighted fusion on the extracted features to obtain fused features, and the global self-attention layer is used to perform global self-attention calculation on the fused features.

6. A pipeline leakage crack morphology identification method according to claim 1, characterized in that: The wavelet time-frequency diagram extraction of the downsampled acoustic emission signal is specifically performed as follows: The downsampled acoustic emission signal is subjected to continuous wavelet transform using a wavelet basis function to obtain a wavelet time-frequency diagram of the acoustic emission signal.

7. A pipeline leakage crack morphology identification device, characterized in that: include: An acquisition module is used to acquire the acoustic emission signal generated when there is leakage at the crack of the pipeline; A downsampling module is used to downsample the acoustic emission signal to obtain a downsampled acoustic emission signal; An extraction module, used for extracting a wavelet time-frequency diagram of the downsampled acoustic emission signal to obtain a wavelet time-frequency diagram of the acoustic emission signal; A filtering module, used for performing fast local Laplace filtering on the wavelet time-frequency diagram using an improved remapping function to obtain an enhanced wavelet time-frequency diagram; An identification module is used to input the enhanced wavelet time-frequency diagram into a pre-trained pipeline leakage crack morphology recognition model to obtain the pipeline leakage crack morphology; wherein the pipeline leakage crack morphology recognition model is obtained by training a convolutional neural network based on MSCA-DAA using the enhanced wavelet time-frequency diagram and its corresponding leakage crack morphology as training samples, and the convolutional neural network based on MSCA-DAA includes a multi-scale convolution module, a dimension-aware attention module and a global self-attention layer connected in sequence, and the dimension-aware attention module is used to implement average pooling operations along vertical and horizontal dimensions, average pooling operations along channel dimensions and horizontal dimensions, and average pooling operations along channel dimensions and vertical dimensions, and finally weightedly fuse the information of channel dimensions, vertical dimensions and horizontal dimensions.

8. A device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the pipeline leakage crack morphology identification method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of a pipeline leakage crack morphology identification method as claimed in any one of claims 1 to 6 are implemented.

Citation Information

Patent Citations

  • Precise positioning method and system for pipeline crack, terminal and storage medium thereof

    CN115436470A

  • Tone mapping method and system based on lookup table and Laplacian filtering

    CN116843585A

  • Online positioning identification method, device and equipment for pipeline leakage crack morphology

    CN116857571A

  • Explanatable deep learning framework for pipeline leakage monitoring

    CN118277830A

  • Systems and methods for improved pipeline leak detection

    US20230332976A1

Cited By

  • Pipeline identification method and device, air conditioner, storage medium and computer program product

    CN119715806A

  • Pipeline identification method and device, air conditioner, storage medium and computer program product

    CN119715806B