CN114926707A - Pipeline defect identification method, processor and pipeline defect identification device - Google Patents
Pipeline defect identification method, processor and pipeline defect identification device Download PDFInfo
- Publication number
- CN114926707A CN114926707A CN202210562099.9A CN202210562099A CN114926707A CN 114926707 A CN114926707 A CN 114926707A CN 202210562099 A CN202210562099 A CN 202210562099A CN 114926707 A CN114926707 A CN 114926707A
- Authority
- CN
- China
- Prior art keywords
- pipeline
- image
- defect
- pipeline defect
- defect identification
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 230000007547 defect Effects 0.000 title claims abstract description 251
- 238000000034 method Methods 0.000 title claims abstract description 50
- 230000004907 flux Effects 0.000 claims abstract description 32
- 230000002950 deficient Effects 0.000 claims abstract description 19
- 238000005520 cutting process Methods 0.000 claims abstract description 5
- 238000012549 training Methods 0.000 claims description 66
- 238000012360 testing method Methods 0.000 claims description 36
- 238000013528 artificial neural network Methods 0.000 claims description 21
- 230000008569 process Effects 0.000 claims description 14
- 238000012545 processing Methods 0.000 claims description 13
- 238000003466 welding Methods 0.000 claims description 11
- 238000013095 identification testing Methods 0.000 claims description 9
- 230000008030 elimination Effects 0.000 claims description 4
- 238000003379 elimination reaction Methods 0.000 claims description 4
- 230000006870 function Effects 0.000 description 10
- 238000001514 detection method Methods 0.000 description 9
- 238000003860 storage Methods 0.000 description 8
- 238000004590 computer program Methods 0.000 description 7
- 238000010586 diagram Methods 0.000 description 7
- 239000002609 medium Substances 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 4
- 238000000605 extraction Methods 0.000 description 3
- 238000004519 manufacturing process Methods 0.000 description 3
- 210000002569 neuron Anatomy 0.000 description 3
- 230000004913 activation Effects 0.000 description 2
- 238000013459 approach Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 230000001066 destructive effect Effects 0.000 description 2
- 238000004880 explosion Methods 0.000 description 2
- 239000007789 gas Substances 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 230000007774 longterm Effects 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- VNWKTOKETHGBQD-UHFFFAOYSA-N methane Chemical compound C VNWKTOKETHGBQD-UHFFFAOYSA-N 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000009960 carding Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000002485 combustion reaction Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000007906 compression Methods 0.000 description 1
- 230000006835 compression Effects 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 230000007797 corrosion Effects 0.000 description 1
- 238000005260 corrosion Methods 0.000 description 1
- 238000013144 data compression Methods 0.000 description 1
- 230000008034 disappearance Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000003628 erosive effect Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 239000002184 metal Substances 0.000 description 1
- 239000007769 metal material Substances 0.000 description 1
- 244000005700 microbiome Species 0.000 description 1
- 239000003345 natural gas Substances 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 239000003209 petroleum derivative Substances 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 229920006395 saturated elastomer Polymers 0.000 description 1
- 239000002689 soil Substances 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 239000006163 transport media Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F17—STORING OR DISTRIBUTING GASES OR LIQUIDS
- F17D—PIPE-LINE SYSTEMS; PIPE-LINES
- F17D5/00—Protection or supervision of installations
- F17D5/02—Preventing, monitoring, or locating loss
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N27/00—Investigating or analysing materials by the use of electric, electrochemical, or magnetic means
- G01N27/72—Investigating or analysing materials by the use of electric, electrochemical, or magnetic means by investigating magnetic variables
- G01N27/82—Investigating or analysing materials by the use of electric, electrochemical, or magnetic means by investigating magnetic variables for investigating the presence of flaws
- G01N27/83—Investigating or analysing materials by the use of electric, electrochemical, or magnetic means by investigating magnetic variables for investigating the presence of flaws by investigating stray magnetic fields
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T11/00—2D [Two Dimensional] image generation
- G06T11/20—Drawing from basic elements, e.g. lines or circles
- G06T11/206—Drawing of charts or graphs
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Databases & Information Systems (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Biophysics (AREA)
- Chemical & Material Sciences (AREA)
- Multimedia (AREA)
- Medical Informatics (AREA)
- Molecular Biology (AREA)
- Biomedical Technology (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Biochemistry (AREA)
- Analytical Chemistry (AREA)
- Electrochemistry (AREA)
- Chemical Kinetics & Catalysis (AREA)
- Mechanical Engineering (AREA)
- Investigating Or Analyzing Materials By The Use Of Magnetic Means (AREA)
Abstract
The embodiment of the invention provides a pipeline defect identification method, a processor and a pipeline defect identification device, wherein the pipeline defect identification method comprises the following steps: acquiring magnetic flux leakage signal data of a pipeline defective pipe section; drawing a magnetic flux leakage curve image according to the magnetic flux leakage signal data; cutting the magnetic leakage curve image to obtain a magnetic leakage curve image with a standard size; determining a pipeline defect identification image according to the magnetic leakage curve image with the standard size; inputting the pipeline defect identification image into a pipeline defect identification model to extract characteristic information of the pipeline defect identification image; the defect type of the pipeline defective pipe section is identified according to the characteristic information, the method is high in identification efficiency and accuracy when used for identifying the defects of the pipeline defective pipe section, and the method is simple and easy to implement.
Description
Technical Field
The invention relates to the technical field of pipeline defect identification, in particular to a pipeline defect identification method, a processor and a pipeline defect identification device.
Background
Because the oil gas pipeline is mostly buried underground with a complex soil environment and is often threatened by safety such as environmental erosion, fatigue damage, third party damage, natural disasters, misoperation and the like, under the corrosion action of transport media and microorganisms and long-term service of the pipeline, the pipeline is very easy to have defects of various degrees, the problems often cause various problems on the pipeline to reduce the service life of the pipeline, and petroleum or natural gas leakage, even combustion or explosion can be caused in severe cases. Although there are various causes that contribute to the formation of various defects on the pipe, the surface defects of the pipe mainly include surface peeling, perforations, pits, cracks, surface large-area peeling, and the like.
In order to prevent pipeline accidents and maintain safe operation of pipelines, pipeline defect information needs to be obtained in time. The non-destructive detection technology of the pipeline is developed by domestic and foreign scholars so as to determine the position information of the geometric defect of the pipeline, but in order to further obtain the information of the pipeline defect, the magnetic leakage detection signal needs to be further analyzed, and when the traditional pipeline defect identification model is used for classifying and identifying the detection image, the non-destructive detection model is not well suitable for a large amount of magnetic leakage detection data measured for years, and the problems of insufficient sample data of individual defect types and discontinuous numerical span exist.
Disclosure of Invention
The embodiment of the invention aims to provide a pipeline defect identification method, a processor and a pipeline defect identification device, which can improve the efficiency of pipeline defect identification and are simple and easy to implement.
In order to achieve the above object, a first aspect of the present invention provides a pipe defect identification method, including:
acquiring magnetic flux leakage signal data of a defective pipe section of the pipeline;
drawing a magnetic flux leakage curve image according to the magnetic flux leakage signal data;
cutting the magnetic leakage curve image to obtain a magnetic leakage curve image with a standard size;
determining a pipeline defect identification image according to the magnetic leakage curve image with the standard size;
inputting the pipeline defect identification image into a pipeline defect identification model to extract characteristic information of the pipeline defect identification image;
and identifying the defect type of the defective pipe section of the pipeline according to the characteristic information.
In an embodiment of the invention, the pipeline defect identification model is constructed by the following steps:
determining the structure, the size and the welding seam type of the pipeline;
eliminating the magnetic flux leakage curve image with the standard size according to the structure, the size and the welding seam type;
determining the quantity corresponding to each defect type in the magnetic leakage curve image after the elimination processing so as to construct a pipeline defect image sample set;
constructing a pipeline defect recognition training network;
and constructing a pipeline defect identification model according to the defect image sample set and the pipeline defect identification training network.
In an embodiment of the present invention, the pipeline defect image sample set includes a pipeline defect identification image and a defect-free pipeline image.
In the embodiment of the invention, the eliminating processing of the magnetic leakage curve image with the standard size according to the structure, the size and the welding seam type comprises the following steps:
determining a three-way valve and a flange in the pipeline according to the structure and the size;
determining a circumferential weld and a spiral weld in the pipeline according to the type and the size of the weld;
and eliminating the magnetic leakage curve image with the three-way valve, the flange, the circumferential weld and the spiral weld in the magnetic leakage curve image with the standard size.
In an embodiment of the present invention, constructing a pipeline defect recognition training network includes:
acquiring a convolution kernel weight value of a residual error neural network;
determining a pipeline defect image training set according to the pipeline defect image sample set;
and inputting the convolution kernel weight value and the pipeline defect image training set into a residual error neural network to construct a pipeline defect identification training network.
In an embodiment of the present invention, obtaining the convolution kernel weight value of the residual neural network includes:
acquiring a training network of a sparse self-encoder;
inputting the pipeline defect image training set into a training network of a sparse self-encoder for training so as to construct a test network of the sparse self-encoder;
determining a pipeline defect image test set according to the pipeline defect image sample set;
and inputting the pipeline defect image test set into a test network of the sparse self-encoder to test so as to output a convolution kernel weight value.
In the embodiment of the invention, the constructing of the pipeline defect identification model according to the defect image sample set and the pipeline defect identification training network comprises the following steps:
inputting the pipeline defect image training set into a pipeline defect identification training network for training so as to construct a pipeline defect identification test network;
and inputting the pipeline defect image test set into a pipeline defect identification test network for testing so as to construct a pipeline defect identification model.
A second aspect of the invention provides a processor configured to perform the above-described method for identifying pipe defects.
In a third aspect, the present invention provides a pipeline defect identifying device, which includes the processor.
In an embodiment of the present invention, the pipe defect identifying apparatus further includes:
and the magnetic sensor is used for detecting a magnetic leakage signal of the defective pipe section of the pipeline.
Through above-mentioned technical scheme, draw the magnetic leakage curve image according to the magnetic leakage signal data of pipeline defect pipeline section, cut out the processing in order to confirm pipeline defect identification image to the magnetic leakage curve image, with pipeline defect identification image input to pipeline defect identification model in to the defect type of discernment pipeline defect pipeline section, can promote the defect identification efficiency and the rate of accuracy of pipeline defect pipeline section, the practicality is stronger.
Additional features and advantages of embodiments of the invention will be set forth in the detailed description which follows.
Drawings
The accompanying drawings, which are included to provide a further understanding of the embodiments of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the embodiments of the invention without limiting the embodiments of the invention. In the drawings:
FIG. 1 is a schematic flow chart of a pipeline defect identification method in an embodiment of the invention;
FIG. 2 is a schematic view of a magnetic leakage curve image of a circumferential weld according to an embodiment of the present invention;
FIG. 3 is a schematic view of a magnetic leakage curve image of a spiral weld seam according to an embodiment of the present invention;
FIG. 4 is a magnetic leakage curve image diagram of a three-way valve in an embodiment of the invention;
FIG. 5 is a magnetic flux leakage curve image of a flange according to an embodiment of the present invention;
FIG. 6 is a magnetic flux leakage curve image of a defective segment of a pipeline in an embodiment of the present invention;
FIG. 7 is a schematic illustration of a defect-free pipeline image in an embodiment of the present invention.
Detailed Description
The following detailed description of embodiments of the invention refers to the accompanying drawings. It should be understood that the detailed description and specific examples, while indicating embodiments of the invention, are given by way of illustration and explanation only, not limitation.
An embodiment of the present invention provides a method for identifying a pipeline defect, as shown in fig. 1, the method for identifying a pipeline defect includes the following steps:
step S101: and acquiring magnetic flux leakage signal data of the defective pipe section of the pipeline.
Specifically, the pipeline defect identification method in this embodiment is applicable to a pipeline defect identification device, and a pipeline specifically refers to an oil and gas pipeline, and the pipeline defect identification device includes a processor and a magnetic sensor which are connected by signals, the magnetic sensor is used for detecting a magnetic leakage signal of a defective pipe section of the pipeline, and in magnetic leakage detection, after the metal pipeline is magnetized, the magnetic conductivities of the defect-free partial pipelines are the same; in the defective part, because the magnetic resistance at the defect part is different from the metal material, a leakage magnetic field is further formed, so that an operator can detect the defective pipe section of the pipeline by adopting the magnetic sensor, the leakage magnetic signal data of the defective pipe section of the pipeline can be obtained, and the magnetic sensor transmits the data to the processor after the detection is finished.
Step S102: and drawing a magnetic leakage curve image according to the magnetic leakage signal data.
Specifically, after acquiring magnetic flux leakage signal data of a pipeline defect pipe section, the processor performs curve drawing on the data in a point drawing mode to draw a magnetic flux leakage curve image.
Step S103: and (4) cutting the magnetic leakage curve image to obtain a magnetic leakage curve image with a standard size.
Specifically, after a leakage flux curve image is drawn, preprocessing is performed on the leakage flux curve image, for example, the longitudinal interval of the leakage flux curve image is increased, and the size of a single leakage flux curve image and the quantity of leakage flux signals contained in the single leakage flux curve image, which are convenient to display and analyze, are determined; then, performing batch cutting processing on the magnetic leakage curve image so as to classify the types of the defects of the pipelines, wherein the ninety-percent magnetic leakage curve image with the standard size after being cut contains a complete defect outline as far as possible, and the magnetic leakage curve image with the standard size of more than sixty percent only contains one type of defects; for the case that the leakage flux curve image with a few standard sizes contains a plurality of defect types, the other defects except a certain defect type can be replaced by backgrounds, and one leakage flux curve image with the standard size containing a plurality of defect types is converted into a plurality of leakage flux curve images with standard sizes containing one defect type, or the leakage flux curve images are deleted.
Further, for a pipeline containing a spiral weld joint, a leakage flux signal contained in a single leakage flux curve image with standard size should be capable of completely presenting a spiral weld joint so as to perform subsequent observation, feature extraction and data integration and carding; for the pipeline without the spiral welding seam, the quantity of the presented magnetic leakage signals and the quantity of the pipeline which is supposed to be laid in the spiral welding seam mode are ensured, and the single magnetic leakage curve image is cut into the magnetic leakage curve image with the standard size of the fixed pixel value in batch according to the size of the pipeline defect
Step S104: determining a pipeline defect identification image according to the magnetic leakage curve image with the standard size;
step S105: and inputting the pipeline defect identification image into a pipeline defect identification model to extract the characteristic information of the pipeline defect identification image.
Specifically, in one embodiment of the present invention, the pipeline defect identification model is constructed by:
step S201: the structure, dimensions and weld type of the pipe are determined.
An operator can acquire the relevant data of the pipeline body such as the structure, the size, the welding seam type and the like of the pipeline by observing and measuring the pipeline, and then the data can be input into the processor.
Step S202: and eliminating the magnetic flux leakage curve image with the standard size according to the structure, the size and the welding seam type.
Further, step S202: the step of eliminating the magnetic leakage curve image with the standard size according to the structure, the size and the welding seam type further comprises a step S301-a step S303, wherein:
step S301: determining a three-way valve and a flange in the pipeline according to the structure;
step S302: determining a circumferential weld and a spiral weld in the pipeline according to the type of the weld;
step S303: and eliminating the magnetic leakage curve images with the three-way valve, the flange, the circumferential weld and the spiral weld in the magnetic leakage curve images with the standard sizes according to the sizes.
Specifically, the processor further determines the specific defect type to be classified on the pipeline after determining the structure, the size and the welding seam type of the pipeline, and can divide a plurality of defects into one defect type when the defects are too concentrated; because the pipeline contains a girth weld, a spiral weld, a three-way valve and a flange, and the magnetic leakage curve images (shown in figures 2-5) of the pipe fittings can interfere with the identification of the pipeline defects, namely the accuracy of the subsequent identification of the pipeline defects can be influenced, the magnetic leakage curve images containing the pipe fittings are required to be removed by combining the relevant data of the pipeline body such as the structure, the size, the type of the weld and the like of the pipeline, the size of the pipeline can be used for helping to determine the positions of the three-way valve, the flange, the circular weld and the spiral weld, the structures of the three-way valve, the flange and the pipeline are associated, and the three-way valve and the flange in the pipeline are required to be determined according to the structures; in the same way, the types of the circumferential weld, the spiral weld and the pipeline are related, the circumferential weld and the spiral weld in the pipeline need to be determined according to the types of the welds, then the corresponding positions in the three-way valve, the flange, the circumferential weld and the spiral weld are determined according to the sizes of the pipelines, then the magnetic flux leakage curve image with the standard size corresponding to the position is determined, and then the magnetic flux leakage curve image with the standard size at the position is removed.
Step S203: and determining the defect types contained in the magnetic leakage curve image after the elimination processing and the number corresponding to each defect type to construct a pipeline defect image sample set.
Specifically, in this embodiment, the pipeline defect image sample set is mainly established according to magnetic leakage signal data measured by an actual pipeline, and there are two approaches for obtaining the magnetic leakage signal data of an actual pipeline defect pipe section: one method is an experimental method for artificially manufacturing the pipeline defect, and then a magnetic sensor is utilized to obtain a magnetic leakage signal of a pipeline section with the pipeline defect; and the other method is to adopt the in-pipeline detection technology to obtain the magnetic flux leakage detection signal data of the defective pipeline section of the pipeline. And if the number of the magnetic leakage curve images corresponding to a certain defect type of the actual pipeline is very small, the number of the magnetic leakage curve images can be supplemented by adopting an artificial manufacturing method so as to improve the universality of the pipeline defect identification method, wherein the number range corresponding to each defect type in the magnetic leakage curve images after the elimination is 1000-10000.
In an embodiment of the present invention, the pipeline defect image sample set includes a pipeline defect identification image and a defect-free pipeline image.
Specifically, a large number of small-size magnetic leakage curve images (namely, magnetic leakage curve images with standard sizes) are cut on the basis of a large-size magnetic leakage curve image, and a defect-free small-size magnetic leakage curve image appears in a defect image sample set, so that a defect-free classification type is supplemented, the pipeline defect image sample set is composed of a pipeline defect identification image (shown in fig. 6) and a defect-free pipeline image (shown in fig. 7), the fault tolerance capability can be increased for early-stage data processing, meanwhile, the accuracy of a pipeline defect identification model can be evaluated, and the practicability of the pipeline defect identification model is widened. In this embodiment, after the pipeline defect image sample set is constructed, a defect type label is made for each leakage flux curve image with a standard size in the pipeline defect image sample set, and the defect type label has a defect type corresponding to the leakage flux curve image with the standard size.
Step S204: and constructing a pipeline defect identification training network.
Further, step S204: the method for constructing the pipeline defect identification training network further comprises the steps S401-S403, wherein:
step S401: and acquiring a convolution kernel weight value of the residual error neural network.
Further, in an embodiment of the present invention, step S401: obtaining the convolution kernel weight value of the residual neural network further comprises steps S501 to S504, wherein:
step S501: acquiring a training network of a sparse self-encoder;
specifically, the sparse autoencoder is an unsupervised machine learning mode and uses a back propagation algorithm, can quickly extract features of data, continuously adjusts parameters of the autoencoder by calculating errors between output of the autoencoder and original input, and finally trains a model. The sparse self-encoder is composed of an encoder, a decoder and a loss function, wherein the encoder maps high-dimensional data into low-dimensional data, and the mapping relation is obtained through correlation among the data; the decoder maps the low-dimensional data into high-dimensional data with the same shape by increasing the data volume; the loss function is used to measure the loss of information due to data compression.
The encoding process is shown in the following formula:
y=f(Wx+b) (1)
where x is the input, y is the new feature, W is the encoded weight value, b is the encoded bias value, and f (Wx + b) is the nonlinear activation function.
The decoding process is shown by the following formula:
wherein W 'is the weight value of the code, b' is the bias value of the code, f (W 'y + b') is the target function,
The input is reconstructed with the new feature y.
The input and output are made to approach infinity with a loss function that minimizes the negative log-likelihood, the calculation formula is as follows:
L=-logP(x|x') (3)
where P (x | x') is the loss function and L is the negative logarithm of the loss value.
Sparse representation is to represent a signal by atoms as small as possible, and a more concise representation mode makes it easier to obtain information contained in the signal and facilitates processing steps such as compression, encoding and the like of the signal. Sparse constraints are that neurons are selectively inhibited, and that neuron outputs can be used as the degree of neuron activation. And adding the sparse limitation into the self-coding network to form a sparse self-coder.
In this embodiment, the training network of the sparse self-encoder is pre-stored in the processor, and is called out when needed.
Step S502: and inputting the pipeline defect image training set into a training network of the sparse self-encoder for training so as to construct a test network of the sparse self-encoder.
Collecting a magnetic leakage curve image with a standard size corresponding to each defect type in a pipeline defect image sample set according to the following steps of: and 3, dividing the images according to the proportion, and taking the set of the magnetic flux leakage curve images with a large number and standard sizes as a pipeline defect image training set. After the pipeline defect image training set is obtained, the pipeline defect image training set is input into a training network of a sparse self-encoder to be trained, so that an initial value of a convolution kernel in a convolution neural network is initialized. In the training process, if the image entropy reaches a preset image entropy threshold range, stopping iteration and determining a convolution kernel weight initial value; if the image entropy does not reach the preset image entropy threshold range, continuously performing iteration, wherein the calculation formula of the image entropy E is as follows:
wherein E is the image entropy; w i,j ,W i ' ,j Respectively are weight matrixes corresponding to any two convolution kernels.
Step S503: and determining a pipeline defect image test set according to the pipeline defect image sample set.
And dividing the magnetic leakage curve images with standard sizes corresponding to each defect type in the pipeline defect image sample set according to a proportion, and collecting the magnetic leakage curve images with small number and standard sizes to form a pipeline defect image test set.
Step S504: and inputting the pipeline defect image test set into a test network of the sparse self-encoder to test so as to output a convolution kernel weight value.
In the testing process, if the image entropy calculated by the testing network of the sparse self-encoder is always within the range of the preset image entropy threshold value, determining the initial value of the convolution kernel weight output by the testing network of the sparse self-encoder as the weight value of the convolution kernel; if the image entropy calculated by the test network of the sparse self-encoder is not always within the preset image entropy threshold range in the test process, it indicates that the test network of the sparse self-encoder is not qualified, and the step S502 should be returned to, and the training network of the sparse self-encoder is continuously trained until the obtained image entropy in the test network of the sparse self-encoder is always within the preset image entropy threshold range. In the embodiment, the sparse autoencoder is used for obtaining the convolution kernel weight value, the sparse limitation is used for reducing the useless information and redundant information amount of the convolution kernel weight value, the blind random setting is avoided, the overall operation efficiency of the model is increased, the improvement of the feature extraction capability of the network model is facilitated, and the model training speed is accelerated.
Step S402: determining a pipeline defect image training set according to the pipeline defect image sample set;
step S403: and inputting the convolution kernel weight value and the pipeline defect image training set into a residual error neural network to construct a pipeline defect identification training network.
The conventional convolutional neural network has the problem of gradient disappearance or gradient explosion along with the continuous deepening of the network depth, so that the classification and identification effects of the network are influenced, compared with the conventional shallow neural network, the residual neural network takes the residual between output and input as a learning target, converts a problem into a plurality of residual problems, reduces the optimization difficulty, and is connected with a plurality of congruent mapping layers behind the conventional shallow neural network when the conventional shallow neural network is saturated to solve the problem of unstable gradient. And the residual error neural network can increase the depth of the network model, is more beneficial to processing a large number of magnetic leakage curve images with standard sizes containing defects for many years, enhances the capability of extracting image characteristic information, and improves the accuracy and the practicability of the pipeline defect identification model.
Specifically, a residual error neural network is pre-stored in the processor, and the convolution kernel weight value obtained in the above process is put into the residual error neural network, so that the pipeline defect identification training network can be obtained. Further, if the total number of convolution kernels of the residual neural network is a, the number of convolution kernel weight values that need to be obtained by adopting the above process is also a.
Furthermore, the residual neural network can adopt the commonly used classical level number, such as ResNet50, the setting of the parameter value is tested by long-term practice, and the residual neural network has good reliability and strong practical applicability; and replacing the last FC layer without the residual error neural network with a full connection layer and an output layer, and classifying all extracted features to obtain n +1 classes, wherein n is the number of defect types contained in the pipeline defect image sample set.
Step S205: and constructing a pipeline defect identification model according to the defect image sample set and the pipeline defect identification training network.
Further, step S205: the method for constructing the pipeline defect identification model according to the defect image sample set and the pipeline defect identification training network further comprises the steps S601-S602, wherein:
step S601: inputting the pipeline defect image training set into a pipeline defect recognition training network for training so as to construct a pipeline defect recognition test network;
inputting a convolution kernel weight values and a pipeline defect image training set into a pipeline defect identification training network for training, acquiring characteristic information (such as color characteristic information, texture characteristic information, shape characteristic information and the like) of a magnetic leakage curve image with a standard size when the pipeline defect identification training network identifies the magnetic leakage curve image with the standard size in the pipeline defect image training set, identifying the defect type of the magnetic leakage curve image with the standard size according to the characteristic information, and comparing the identified defect type with the defect type on a label of the magnetic leakage curve image with the standard size, so that whether the pipeline defect identification training network identifies the defect type of the magnetic leakage curve image with the standard size accurately can be determined.
In the training process, if the accuracy of the pipeline defect recognition training network on the pipeline defect recognition of the pipeline defect image training set is in a preset accuracy range, stopping training, and obtaining a pipeline defect recognition training network which is a pipeline defect recognition testing network; if the accuracy of the pipeline defect recognition training network on the pipeline defect recognition of the pipeline defect image training set is not in the preset accuracy range, continuously training until the accuracy is in the preset accuracy range, wherein the preset accuracy range is 90% -100%.
Step S602: and inputting the pipeline defect image test set into a pipeline defect identification test network for testing so as to construct a pipeline defect identification model.
In the testing process, if the accuracy of the pipeline defect identification testing network for identifying the pipeline defects of the pipeline defect image testing set is always within the preset accuracy range, determining a pipeline defect identification model as the pipeline defect identification testing network; if the accuracy of the pipeline defect identification test network for identifying the pipeline defects of the pipeline defect image test set is not always within the preset accuracy range, the method returns to the step S601, trains the pipeline defect identification training network again and obtains the pipeline defect identification test network until the accuracy of the pipeline defect identification test network for identifying the pipeline defects of the pipeline defect image test set is always within the preset accuracy range.
In step S105, the pipe defect recognition model performs a feature information extraction operation on the input pipe defect recognition image to be recognized, so as to extract feature information (such as color feature information, texture feature information, shape feature information, and the like) of the pipe defect recognition image.
Step S106: and identifying the defect type of the defective pipe section of the pipeline according to the characteristic information.
The pipeline defect identification model can identify the defect type of the pipeline defect section corresponding to the pipeline defect identification image according to the characteristic information of the pipeline defect identification image.
Another embodiment of the present invention provides a processor configured to perform the pipe defect identifying method of the above embodiment.
Another embodiment of the present invention provides a pipeline defect identifying apparatus, which includes the processor of the above embodiment.
In an embodiment of the present invention, the pipe defect identifying apparatus further comprises a magnetic sensor in signal communication with the processor and configured to detect a leakage magnetic signal of a defective pipe section of the pipe.
According to the pipeline defect identification method, the processor and the pipeline defect identification device, provided by the embodiment of the invention, the magnetic leakage curve image is drawn according to the magnetic leakage signal data of the pipeline defect section, the magnetic leakage curve image is cut to determine the pipeline defect identification image, and the pipeline defect identification image is input into the pipeline defect identification model to identify the defect type of the pipeline defect section, so that the defect identification efficiency and accuracy of the pipeline defect section can be improved, and the practicability is strong.
As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and so forth) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
In a typical configuration, a computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.
The memory may include forms of volatile memory in a computer readable medium, Random Access Memory (RAM) and/or non-volatile memory, such as Read Only Memory (ROM) or flash memory (flash RAM). The memory is an example of a computer-readable medium.
Computer-readable media, including both non-transitory and non-transitory, removable and non-removable media, may implement information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), Read Only Memory (ROM), Electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), Digital Versatile Disks (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transmission medium, which can be used to store information that can be accessed by a computing device. As defined herein, a computer readable medium does not include a transitory computer readable medium such as a modulated data signal and a carrier wave.
It should also be noted that the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in the process, method, article, or apparatus that comprises the element.
The above are merely examples of the present application and are not intended to limit the present application. Various modifications and changes may occur to those skilled in the art to which the present application pertains. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.
Claims (10)
1. A pipeline defect identification method is characterized by comprising the following steps:
acquiring magnetic flux leakage signal data of a pipeline defective pipe section;
drawing a magnetic leakage curve image according to the magnetic leakage signal data;
cutting the magnetic leakage curve image to obtain a magnetic leakage curve image with a standard size;
determining a pipeline defect identification image according to the magnetic leakage curve image with the standard size;
inputting the pipeline defect identification image into a pipeline defect identification model to extract the characteristic information of the pipeline defect identification image;
and identifying the defect type of the defective pipeline section of the pipeline according to the characteristic information.
2. The method of identifying pipe defects according to claim 1, wherein the pipe defect identification model is constructed by:
determining the structure, the size and the welding seam type of the pipeline;
eliminating the magnetic leakage curve image with the standard size according to the structure, the size and the welding seam type;
determining the defect types contained in the magnetic leakage curve image after the elimination processing and the number corresponding to each defect type to construct a pipeline defect image sample set;
constructing a pipeline defect recognition training network;
and constructing a pipeline defect identification model according to the defect image sample set and the pipeline defect identification training network.
3. The pipe defect identification method of claim 2, wherein the pipe defect image sample set comprises the pipe defect identification image and a non-defective pipe image.
4. The pipeline defect identification method according to claim 2, wherein the eliminating process of the leakage magnetic curve image with the standard size according to the structure, the size and the weld type comprises the following steps:
determining a three-way valve and a flange in the pipeline according to the structure and the size;
determining a girth weld and a spiral weld in the pipeline according to the weld type and the size;
and eliminating the magnetic leakage curve images of the three-way valve, the flange, the circumferential weld and the spiral weld in the magnetic leakage curve image with the standard size.
5. The method for identifying the pipeline defect according to claim 2, wherein the constructing the training network for identifying the pipeline defect comprises:
acquiring a convolution kernel weight value of a residual error neural network;
determining a pipeline defect image training set according to the pipeline defect image sample set;
inputting the convolution kernel weight values and the pipeline defect image training set into the residual error neural network to construct the pipeline defect recognition training network.
6. The pipeline defect identification method according to claim 5, wherein the obtaining of the convolution kernel weight value of the residual neural network comprises:
acquiring a training network of a sparse self-encoder;
inputting the pipeline defect image training set into a training network of the sparse self-encoder for training so as to construct a test network of the sparse self-encoder;
determining a pipeline defect image test set according to the pipeline defect image sample set;
and inputting the pipeline defect image test set into a test network of the sparse self-encoder for testing so as to output the convolution kernel weight value.
7. The pipeline defect identification method according to claim 6, wherein the constructing a pipeline defect identification model according to the defect image sample set and the pipeline defect identification training network comprises:
inputting the pipeline defect image training set into the pipeline defect recognition training network for training so as to construct a pipeline defect recognition test network;
inputting the pipeline defect image test set into the pipeline defect identification test network for testing so as to construct the pipeline defect identification model.
8. A processor configured to perform the method of any one of claims 1 to 7.
9. A pipeline defect identification device, characterized in that the pipeline defect identification device comprises a processor according to claim 8.
10. The pipe defect identifying apparatus of claim 9, further comprising:
and the magnetic sensor is used for detecting a magnetic leakage signal of the defective pipe section of the pipeline.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210562099.9A CN114926707A (en) | 2022-05-23 | 2022-05-23 | Pipeline defect identification method, processor and pipeline defect identification device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210562099.9A CN114926707A (en) | 2022-05-23 | 2022-05-23 | Pipeline defect identification method, processor and pipeline defect identification device |
Publications (1)
Publication Number | Publication Date |
---|---|
CN114926707A true CN114926707A (en) | 2022-08-19 |
Family
ID=82810050
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210562099.9A Pending CN114926707A (en) | 2022-05-23 | 2022-05-23 | Pipeline defect identification method, processor and pipeline defect identification device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114926707A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116307300A (en) * | 2023-05-23 | 2023-06-23 | 四川都得利管业有限公司 | Production process optimization method and system for plastic pipeline |
KR102607277B1 (en) * | 2023-06-07 | 2023-11-29 | 한전케이피에스 주식회사 | System for detecting fault and method thereof |
CN117934959A (en) * | 2024-02-02 | 2024-04-26 | 中国矿业大学 | Classification recognition algorithm for pipeline girth weld defect signal image |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110728654A (en) * | 2019-09-06 | 2020-01-24 | 台州学院 | Automatic pipeline detection and classification method based on deep residual error neural network |
CN112381165A (en) * | 2020-11-20 | 2021-02-19 | 河南爱比特科技有限公司 | Intelligent pipeline defect detection method based on RSP model |
CN113063843A (en) * | 2021-02-22 | 2021-07-02 | 广州杰赛科技股份有限公司 | Pipeline defect identification method and device and storage medium |
-
2022
- 2022-05-23 CN CN202210562099.9A patent/CN114926707A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110728654A (en) * | 2019-09-06 | 2020-01-24 | 台州学院 | Automatic pipeline detection and classification method based on deep residual error neural network |
CN112381165A (en) * | 2020-11-20 | 2021-02-19 | 河南爱比特科技有限公司 | Intelligent pipeline defect detection method based on RSP model |
CN113063843A (en) * | 2021-02-22 | 2021-07-02 | 广州杰赛科技股份有限公司 | Pipeline defect identification method and device and storage medium |
Non-Patent Citations (2)
Title |
---|
SHIYANG ZHOU ET AL.: "Classification of surface defects on steel sheet using convolutional neural networks", 《MATERIALS AND TECHNOLOGY》 * |
王竹筠: "基于深度学习的管道漏磁信号智能识别方法研究", 《中国优秀硕士学位论文全文数据库 工程科技I辑》 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116307300A (en) * | 2023-05-23 | 2023-06-23 | 四川都得利管业有限公司 | Production process optimization method and system for plastic pipeline |
CN116307300B (en) * | 2023-05-23 | 2023-08-08 | 四川都得利管业有限公司 | Production process optimization method and system for plastic pipeline |
KR102607277B1 (en) * | 2023-06-07 | 2023-11-29 | 한전케이피에스 주식회사 | System for detecting fault and method thereof |
CN117934959A (en) * | 2024-02-02 | 2024-04-26 | 中国矿业大学 | Classification recognition algorithm for pipeline girth weld defect signal image |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN114926707A (en) | Pipeline defect identification method, processor and pipeline defect identification device | |
US20210319265A1 (en) | Method for segmentation of underground drainage pipeline defects based on full convolutional neural network | |
Kumar et al. | Deep learning–based automated detection of sewer defects in CCTV videos | |
WO2020133639A1 (en) | Intelligent analysis system for magnetic flux leakage detection data in pipeline | |
KR102239377B1 (en) | System and method for safety inspection on bridge facilities based on image and xai | |
CN112883538B (en) | Corrosion prediction system and method for buried crude oil pipeline | |
CN115935241B (en) | Multi-parameter mutually-fused pipe cleaner real-time positioning method and device | |
CN114299359B (en) | Method, equipment and storage medium for detecting power transmission line faults | |
CN116953196B (en) | Defect detection and safety state assessment method for steel tee joint | |
CN115631132A (en) | Network training method, defect detection method, device, storage medium and equipment | |
Hou et al. | A deep learning technique based flood propagation experiment | |
CN115047064A (en) | Pipeline defect quantification method, processor and pipeline defect quantification device | |
CN110852318A (en) | Drainage pipeline defect accurate positioning method and system | |
CN106404893A (en) | Automatic axial magnetic flux leakage defect signal detection method | |
Carvalho et al. | Pattern recognition techniques applied to the detection and classification of welding defects by magnetic testing | |
CN115063337A (en) | Intelligent maintenance decision-making method and device for buried pipeline | |
Bhowmik | Digital twin for offshore pipeline corrosion monitoring: A deep learning approach | |
CN116228770B (en) | Method and system for identifying and monitoring pipeline leakage | |
CN117237592A (en) | Intelligent identification method for weld joint images at same opening | |
CN115526860A (en) | Battery piece defect detection method and device, computer equipment and readable storage medium | |
Li et al. | Automatic crack detection method for loaded coal in vibration failure process | |
Yemelyanova et al. | Application of machine learning for recognizing surface welding defects in video sequences | |
Jia et al. | Defect severity assessment model for sewer pipeline based on automated pipe calibration | |
CN110082424B (en) | Multi-scale abnormal region recommendation system and method for rapid pipeline magnetic flux leakage data | |
Gao et al. | Quality assessment algorithm of X-ray images in overall girth welds based on deep neural network |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20220819 |