Small-pipe-diameter oil and gas pipe defect quantification method and system based on multi-mode magnetic flux leakage signals

Abstract

The invention relates to the technical field of nondestructive testing, in particular to a small-pipe-diameter oil and gas pipe defect quantification method and system based on a multi-mode magnetic flux leakage signal, and the method comprises the steps: building a small-pipe-diameter oil and gas pipeline model through finite element simulation software, solving the small-pipe-diameter oil and gas pipeline model, and obtaining a simulation model solving result; according to a simulation model solving result, three-axis defect magnetic flux leakage signals are extracted; based on the three-axis defect magnetic flux leakage signal, extracting the characteristic quantity of the three-axis defect magnetic flux leakage signal, and establishing a defect characteristic sample library; an ELMAN defect quantification model is constructed, and the ELMAN defect quantification model is optimized by using a genetic algorithm; and quantifying the defects of the small-pipe-diameter oil and gas pipeline through the optimized ELMAN defect quantification model to obtain the defect size of the small-pipe-diameter oil and gas pipeline. By means of the method, the applicability problem of small-pipe-diameter oil and gas pipeline defect quantification is effectively solved, the extracted characteristics of the three-axis defect magnetic flux leakage signals reflect defect information more accurately, and efficient and accurate quantification of the defects is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of nondestructive testing, and in particular to a method and system for quantifying defects of small-diameter oil and gas pipes based on multi-modal magnetic leakage signals. Background Art

[0002] Small-diameter oil and gas pipelines are important infrastructure and the lifeline of cities. The pipelines transport flammable and explosive dangerous goods such as oil, natural gas, and liquefied gas. They are characterized by large transportation volume and high safety requirements. In order to effectively avoid the occurrence of urban oil and gas pipeline accidents and reduce accidents caused by non-human factors, pipeline in-pipeline detection technology has become an important pre-control method for pipeline detection. However, since most oil and gas pipelines are buried deep underground and the working conditions are very harsh, defects such as corrosion pits and cracks will be caused, which will seriously affect the safety and service life of the pipelines, causing serious accidents, life and economic losses. Therefore, regular safety assessments are required. The leakage magnetic field detection quantification technology collects the leakage magnetic field signals near the pipeline defects, establishes the relationship between the three-axis leakage magnetic field signal characteristics and the defect size, and then realizes quantification. It has been widely used in pipeline safety assessment. Traditional quantification methods are mainly used for medium and large diameter long-distance oil pipelines and tank bottom plates. The applicability to small-diameter oil and gas pipelines needs to be studied. Therefore, it is necessary to establish a high-precision and intelligent size quantification method to achieve defect quantification of small-diameter oil and gas pipelines. Summary of the invention

[0003] The present invention provides a method and system for quantifying defects of small-diameter oil and gas pipes based on multi-modal magnetic leakage signals, which can effectively solve the problems in the background technology.

[0004] In order to achieve the above object, the technical solution adopted by the present invention is:

[0005] A method for quantifying defects of small-diameter oil and gas pipes based on multi-modal magnetic leakage signals, the method comprising:

[0006] Establishing a small-diameter oil and gas pipeline model by using finite element simulation software, solving the small-diameter oil and gas pipeline model, and obtaining a simulation model solution result;

[0007] Extracting triaxial defect magnetic leakage signals according to the solution results of the simulation model;

[0008] Based on the three-axis defect magnetic leakage signal, extract the characteristic quantity of the three-axis defect magnetic leakage signal and establish a defect characteristic sample library;

[0009] Constructing an ELMAN defect quantification model, and optimizing the ELMAN defect quantification model using a genetic algorithm;

[0010] The defects of the small-diameter oil and gas pipeline are quantified by using the optimized ELMAN defect quantification model to obtain the defect size of the small-diameter oil and gas pipeline.

[0011] Furthermore, the establishment of a small-diameter oil and gas pipeline model using finite element simulation software includes: using the finite element analysis software to establish a small-diameter oil and gas pipeline model, a magnetization unit model and four Hall sensors inside the small-diameter oil and gas pipeline, wherein the four Hall sensors are arranged circumferentially in sequence.

[0012] Furthermore, the extracted three-axis defect leakage magnetic signal includes an axial component of the leakage magnetic signal, a radial component of the leakage magnetic signal and a circumferential component of the leakage magnetic signal, and the extracted characteristic quantities of the three-axis defect leakage magnetic signal include the half-width of the leakage magnetic signal, the waveform area of ​​the leakage magnetic signal, the waveform energy of the leakage magnetic signal, the peak value of the leakage magnetic signal, the valley value of the leakage magnetic signal and the peak spacing of the leakage magnetic signal.

[0013] Furthermore, the establishing of the defect feature sample library includes:

[0014] The defect size in the small-diameter oil and gas pipeline model is changed, and the characteristic quantities of the triaxial defect magnetic leakage signals of different defect sizes are respectively extracted to establish a defect feature sample library.

[0015] Furthermore, the construction of the ELMAN defect quantification model includes:

[0016] Dividing the original data in the defect feature sample library into a training set and a test set, and performing normalization processing on the original data;

[0017] Determining the number of indicators of the input layer, hidden layer and output layer of the ELMAN defect quantification model;

[0018] Configuring the network parameters of the ELMAN defect quantization model, and randomly initializing the weights of the input layer and the hidden layer, and the threshold of the hidden layer;

[0019] The training set is used to train the ELMAN defect quantification model, and the test set is used to evaluate the ELMAN defect quantification model.

[0020] Furthermore, the calculation formula for determining the number of indicators of the hidden layer is:

[0021]

[0022] Among them, h is the number of hidden layers, m is the number of input layers, n is the number of output layers, and a is an integer between 0 and 10.

[0023] Furthermore, the optimization of the ELMAN defect quantification model by using a genetic algorithm includes:

[0024] Encoding the weights of the ELMAN defect quantization model and the initial values ​​of the thresholds to obtain an initial population, and setting fitness functions of individuals in the initial population;

[0025] Selecting parent individuals from the individuals, performing crossover and mutation on the parent individuals, and obtaining offspring individuals;

[0026] Calculating the fitness function of the offspring individuals, and merging the offspring individuals with the parent individuals to form a new offspring population;

[0027] The number of iterations of the genetic algorithm is used as the iteration termination condition to determine whether the result of the genetic algorithm meets the iteration termination condition. If not, continue to execute step B3. If so, obtain the optimal weight and threshold.

[0028] Furthermore, setting the fitness function of the individuals in the initial population includes:

[0029] Each individual in the initial population is trained by the ELMAN algorithm to obtain a mean square error of each individual on the test set, and the inverse of the mean square error is used as the fitness function of the individual in the genetic algorithm.

[0030] A small-diameter oil and gas pipe defect quantification system based on multi-modal magnetic leakage signals, the system comprising:

[0031] A simulation model solving module is used to establish a small-diameter oil and gas pipeline model through finite element simulation software, solve the small-diameter oil and gas pipeline model, and obtain a simulation model solving result;

[0032] A magnetic flux leakage signal extraction module is used to extract the magnetic flux leakage signal of the three-axis defect according to the solution result of the simulation model;

[0033] A sample library establishment module, based on the three-axis defect magnetic leakage signal, extracts the characteristic quantity of the three-axis defect magnetic leakage signal and establishes a defect feature sample library;

[0034] A quantization model building module, which builds an ELMAN defect quantization model and optimizes the ELMAN defect quantization model using a genetic algorithm;

[0035] The defect size acquisition module quantifies the defects of the small-diameter oil and gas pipeline by using the optimized ELMAN defect quantification model to obtain the defect size of the small-diameter oil and gas pipeline.

[0036] Furthermore, the quantitative model building module includes:

[0037] A data processing unit, which divides the original data in the defect feature sample library into a training set and a test set, and performs normalization processing on the original data;

[0038] A structure determination unit, which determines the number of indicators of the input layer, hidden layer and output layer of the ELMAN defect quantification model;

[0039] A parameter configuration unit, configured with respect to the network parameters of the ELMAN defect quantization model, and randomly initialized the weights of the input layer and the hidden layer, and the threshold of the hidden layer;

[0040] A model training unit is used to train the ELMAN defect quantification model using the training set, and to evaluate the ELMAN defect quantification model using the test set.

[0041] The technical solution of the present invention can achieve the following technical effects:

[0042] The present invention proposes a GA-ELMAN defect quantification model, which effectively solves the applicability problem of defect quantification of small-diameter oil and gas pipelines. The characteristics of the extracted three-axis defect leakage magnetic signal more accurately reflect the defect information and realize efficient and accurate quantification of defects.

[0043] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0045] Figure 1 It is a flow chart of a method for quantifying defects of small-diameter oil and gas pipes based on multi-modal magnetic flux leakage signals;

[0046] Figure 2 It is a schematic diagram of the characteristic definition of the axial component of the defect leakage magnetic field signal curve;

[0047] Figure 3 A schematic diagram showing the characteristic definition of the radial component of the defect leakage magnetic field signal curve;

[0048] Figure 4Schematic diagram of the structure of the small-diameter oil and gas pipe defect quantification system based on multi-modal magnetic leakage signals. DETAILED DESCRIPTION

[0049] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0050] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more related listed items.

[0051] Embodiment 1:

[0052] like Figure 1 As shown, a small diameter oil and gas pipe defect quantification method based on multi-modal magnetic leakage signals includes:

[0053] S1: Establish a small-diameter oil and gas pipeline model through finite element simulation software, solve the small-diameter oil and gas pipeline model, and obtain the simulation model solution result;

[0054] S2: Extract the triaxial defect magnetic leakage signal according to the simulation model solution;

[0055] S3: Based on the three-axis defect magnetic leakage signal, the characteristic quantity of the three-axis defect magnetic leakage signal is extracted and a defect feature sample library is established;

[0056] Specifically, by using finite element analysis software such as Ansoft Maxwell to establish a simulation model of a small-diameter oil and gas pipeline, simulation can be performed based on the geometric shape, material properties and loading conditions of the pipeline to obtain simulation results. Through these simulation results, the three-axis defect leakage magnetic signal can be extracted. The leakage magnetic signal is a common signal in non-destructive testing and is used to detect defects in materials. The three-axis leakage magnetic signal includes signals in the radial, axial and circumferential directions, which can be used to comprehensively evaluate the spatial distribution of defects. Based on the three-axis leakage magnetic signal, a series of feature quantities can be extracted. These features can effectively describe the morphology of the leakage magnetic signal and have a certain correlation with the geometric size of the defect. Therefore, a data set containing different defect sizes and corresponding leakage magnetic signal features can be constructed to provide data support for subsequent machine learning or model training.

[0057] S4: Construct the ELMAN defect quantification model and optimize it using genetic algorithm;

[0058] The ELMAN defect quantification model is a model for defect quantification based on the ELMAN neural network. The ELMAN network is a neural network with a recursive structure, which is suitable for processing time series data or signal features. When the model is built, the input is the features extracted from the magnetic flux leakage signal, and the output is the predicted defect size. After the model is built, the ELMAN model is optimized using a genetic algorithm. The genetic algorithm can effectively optimize the weights and thresholds of the neural network, find the optimal model parameters, and improve the accuracy and generalization ability of the model in defect size prediction.

[0059] S5: The defects of small-diameter oil and gas pipelines are quantified through the optimized ELMAN defect quantification model to obtain the defect size of small-diameter oil and gas pipelines.

[0060] The present invention takes small-diameter oil and gas pipelines as the research object, establishes a finite element simulation dynamic model, obtains a feature sample library of three-axis magnetic leakage signals, and uses the ELMAN neural network optimized by genetic algorithm for intelligent analysis to achieve accurate quantification of small-diameter oil and gas pipeline defects.

[0061] Through the present invention, a GA-ELMAN defect quantification model is proposed, which effectively solves the applicability problem of defect quantification of small-diameter oil and gas pipelines. The characteristics of the extracted three-axis defect leakage magnetic signal more accurately reflect the defect information, and realize efficient and accurate quantification of defects.

[0062] On the basis of the above-mentioned embodiment, a small-diameter oil and gas pipeline model is established by finite element simulation software, including: using finite element analysis software to establish a small-diameter oil and gas pipeline model, a magnetization unit model and four Hall sensors inside the small-diameter oil and gas pipeline, and the four Hall sensors are arranged circumferentially in sequence.

[0063] As a preferred embodiment of this invention, the construction of a three-dimensional leakage magnetic finite element simulation dynamic model can be based on the size and material of the actual small-diameter oil and gas pipeline, and the pipeline can be three-dimensionally modeled using Maxwell software. For the magnetization unit model, the modeling of the magnet, armature and pole shoe is completed through Boolean operations, merging and mirroring. For the four Hall sensors, in order to simplify the model and achieve the purpose of obtaining magnetic field signals at the same time, the Hall sensors can be replaced by detection points, and 4 detection points are set between the magnetization unit and the inner wall of the pipeline. The 4 detection points are arranged circumferentially, and the extraction path is set to the detection point. Since the actual sensor box has a certain thickness, the lift-off value is set to 2mm to obtain the three-axis defect leakage magnetic signal. In order to simulate the movement process of the magnetization unit in the pipeline and reduce the calculation amount of the simulation model, a move motion command can be added to the pipeline, and the move can be parameterized and modeled so that the pipeline model has a relative movement with respect to the magnetization unit model, and the finite element simulation model is solved.

[0064] Furthermore, the extracted three-axis defect leakage magnetic signal includes the axial component of the leakage magnetic signal, the radial component of the leakage magnetic signal and the circumferential component of the leakage magnetic signal, and the extracted characteristic quantities of the three-axis defect leakage magnetic signal include the half-width of the leakage magnetic signal, the waveform area of ​​the leakage magnetic signal, the waveform energy of the leakage magnetic signal, the peak value of the leakage magnetic signal, the valley value of the leakage magnetic signal and the peak spacing of the leakage magnetic signal.

[0065] Specifically, the extracted axial component of the leakage magnetic signal is along the length of the pipeline, the radial component of the leakage magnetic signal is along the direction perpendicular to the inner wall of the pipeline, and the circumferential component of the leakage magnetic signal is along the circumference of the pipeline; the extracted signal half-width is the peak width at 1 / 2 of the signal height. For the axial component, the distance from the peak to the baseline is taken as the height of the signal, and the signal interception length corresponding to half the height is the half-width of the axial component; for the radial component, the distance between the left and right peaks and the baseline is taken as the signal length to obtain the half-width, and the average of the two half-widths is the half-width of the radial component. For example, the half-width F1 of the axial component is as follows: Figure 2 As shown in the figure, the horizontal axis represents the axial distance, and the vertical axis represents the magnetic induction intensity. The extraction method is: the distance from the peak to the baseline is taken as the height of the signal, and the signal intercept length corresponding to half the height is the half-height width F1 of the axial component; the half-height width F2 of the radial component is as follows: Figure 3 As shown in the figure, the extraction method is: take the distance between the left and right peaks and the baseline as the signal length to obtain the left peak half-height width F2-ɑ and the right peak half-height width F2-β respectively, and take the average value of F2-ɑ and F2-β to obtain the half-height width of the radial component; the extracted signal peak is the maximum point of the signal, the valley is the minimum point of the signal, and the signal peak spacing is the absolute value of the difference between the corresponding horizontal coordinates of the radial component peak and valley; the calculation formula for the extracted signal waveform area is:

[0066]

[0067] The calculation formula of the extracted signal waveform energy is:

[0068]

[0069] In the formula, S is the signal waveform area, E is the signal waveform energy, n is the starting point of the selected signal waveform, N is the total number of data points, and the value of t is n.

[0070] Furthermore, a defect feature sample library is established, including:

[0071] By changing the defect size in the small-diameter oil and gas pipeline model, the characteristic quantities of the triaxial defect magnetic leakage signals of different defect sizes are extracted respectively, and a defect feature sample library is established.

[0072] Specifically, by changing the defect size, including the defect length L, defect depth H and defect width W, the characteristics of the leakage magnetic signals of defects of different sizes can be extracted respectively by changing the defect size, and a feature sample library can be constructed. For example, by changing the defect length L, defect depth H and defect width W parameters, the defect length can vary in the range of 15-35mm, the defect width can vary in the range of 15-35mm, and the defect depth can vary in the range of 20%t-60%t.

[0073] As a preferred embodiment of this invention, the ELMAN defect quantification model is constructed, including:

[0074] A1: Divide the original data in the defect feature sample library into a training set and a test set, and normalize the original data;

[0075] Specifically, the original data containing the leakage magnetic signal characteristics and the corresponding defect size are read from the defect feature sample library, and the data set is divided into a training set and a test set according to a certain ratio, for example, 70% training set and 30% test set. In order to accelerate the training process of the neural network and avoid calculation instability caused by differences in different feature dimensions, the data in the training set and the test set can be normalized. The calculation formula for data normalization is:

[0076]

[0077] In the formula, x′ is the normalized data, x is the original data, and x min is the minimum value, x max is the maximum value.

[0078] A2: Determine the number of indicators in the input layer, hidden layer and output layer of the ELMAN defect quantification model;

[0079] Based on the above embodiments, the number of nodes in each layer of the network can be determined according to the nature of the problem and the characteristics of the input data. The number of nodes in the input layer is the number of features, the number of nodes in the output layer is the number of prediction targets, and the number of nodes in the hidden layer usually needs to be adjusted according to the complexity of the model and the actual problem.

[0080] A3: Configure the network parameters of the ELMAN defect quantization model, and randomly initialize the weights of the input layer and hidden layer, as well as the threshold of the hidden layer;

[0081] In order to provide the necessary initial settings for model training, the network parameters of the ELMAN defect quantization model can be configured, including learning rate, activation function, optimization algorithm, etc. At the same time, in order to ensure that the network can be effectively trained, the initialization of weights and thresholds is usually random to break the symmetry so that the network can learn different features.

[0082] A4: Use the training set to train the ELMAN defect quantification model, and use the test set to evaluate the ELMAN defect quantification model.

[0083] After the network model is built, the model can be trained using training data, and the weights and biases are updated through the back-propagation algorithm. The training process minimizes the loss function through optimization algorithms, such as gradient descent, Adam, etc., and gradually adjusts the weights and thresholds so that the model can better fit the training data. After training, the test set is used to evaluate the generalization ability of the model on unseen data. The performance of the test set can help confirm whether the model has truly learned effective features and will not be biased against new data.

[0084] Based on the above embodiment, the calculation formula for determining the number of indicators of the hidden layer is:

[0085]

[0086] Among them, h is the number of hidden layers, m is the number of input layers, n is the number of output layers, and a is an integer between 0 and 10.

[0087] Specifically, the input layer of this model includes signal characteristic parameters such as the half-height width of the leakage magnetic signal, signal waveform area and signal waveform energy, signal peak and valley values, and signal peak spacing, and the output layer includes the length, depth, and width dimensions of the defect.

[0088] In order to improve the performance of the ELMAN defect quantification model, the genetic algorithm is used to optimize the ELMAN defect quantification model, including:

[0089] B1: Encode the initial values ​​of the weights and thresholds of the ELMAN defect quantification model to obtain the initial population, and set the fitness function of the individuals in the initial population;

[0090] In a genetic algorithm, the initial population is a set of multiple candidate solutions. Each individual represents a possible solution to a neural network. The size of the population usually depends on the complexity of the problem and the computing resources. The initial population can be completed by encoding the initial values ​​of weights and thresholds using a genetic algorithm.

[0091] B2: Select parent individuals from individuals, perform crossover and mutation on the parent individuals, and obtain offspring individuals;

[0092] Specifically, the parent individuals are individuals obtained from the current population through selection operations, such as roulette selection, tournament selection, etc. The selection operation determines the probability of the individual being inherited to the next generation based on the principle of individual fitness. Individuals with high fitness have a greater chance of being selected; the offspring individuals are generated by the parent individuals through crossover and mutation operations, inheriting some genes of the parent individuals, and introducing new gene characteristics through mutation. When generating offspring individuals, the crossover operation can use the single-point crossover operator, which is a commonly used crossover method in genetic algorithms. It can generate new individuals by randomly selecting a point in the parent generation and exchanging the genes of the parent individuals, that is, weights and thresholds, which helps to explore a wider range of solutions in the search space; the mutation operation uses the single-point mutation operator. The mutation operation introduces new solutions to a certain extent by randomly changing a part of the genes of the individual, which helps to avoid the algorithm from falling into the local optimal solution. The single-point mutation operator achieves mutation by randomly changing the value of a gene in the individual.

[0093] B3: Calculate the fitness function of the offspring individuals and merge the offspring individuals with the parent individuals to form a new offspring population;

[0094] Specifically, after calculating the fitness of the offspring individuals, it is necessary to merge the individuals of the parent and offspring generations, and retain the individuals with higher fitness to form a new offspring population, ready to enter the next generation of iteration. This step usually adopts the "elite strategy", that is, retaining the individuals with the best fitness without making changes, which ensures that the optimal solution can enter the next generation and avoids individuals with poor fitness occupying the population.

[0095] B4: Use the number of iterations of the genetic algorithm as the iteration termination condition to determine whether the genetic algorithm result meets the iteration termination condition. If not, continue to execute step B3. If it is satisfied, the optimal weight and threshold are obtained.

[0096] The number of iterations is one of the common termination conditions of genetic algorithms. You can set a maximum number of iterations or stop training until the fitness function reaches a preset value. If the optimal solution has converged to a stable result within the specified number of iterations, the genetic algorithm can stop. The optimal weights and thresholds found by the genetic algorithm can be used as the initialization parameters of the ELMAN neural network. After that, these initialization parameters can be used to further train the network. The genetic algorithm provides a good starting point for the ELMAN neural network, and subsequent training can further improve the model performance.

[0097] As a preferred embodiment of this embodiment, setting the fitness function of individuals in the initial population includes:

[0098] Each individual in the initial population is trained by the ELMAN algorithm to obtain the mean square error of each individual on the test set, and the inverse of the mean square error is used as the fitness function of the individual in the genetic algorithm.

[0099] As the preferred embodiment of this invention, the fitness function is a standard for measuring the performance of an individual. Since the genetic algorithm usually hopes to select the best performing individual, and the smaller the error, the better the individual performance, the inverse of the mean square error can be used as the fitness function. A smaller error value will correspond to a larger fitness value, ensuring that individuals with high fitness have a greater probability of entering the next generation. The calculation formula of the fitness function is:

[0100]

[0101] In the formula, is the predicted value, U={u1,u2,u3,…,u n} is the true value, and n is the number of samples.

[0102] Embodiment 2:

[0103] like Figure 4 As shown, a small diameter oil and gas pipe defect quantification system based on multi-modal magnetic leakage signals includes:

[0104] The simulation model solving module establishes a small-diameter oil and gas pipeline model through finite element simulation software, solves the small-diameter oil and gas pipeline model, and obtains the simulation model solution result;

[0105] The leakage magnetic flux signal extraction module extracts the leakage magnetic flux signals of three-axis defects according to the solution results of the simulation model;

[0106] The sample library establishment module extracts the characteristic quantity of the three-axis defect magnetic leakage signal based on the three-axis defect magnetic leakage signal and establishes a defect characteristic sample library;

[0107] Quantitative model building module, builds the ELMAN defect quantification model, and uses genetic algorithm to optimize the ELMAN defect quantification model;

[0108] The defect size acquisition module quantifies the defects of small-diameter oil and gas pipelines through the optimized ELMAN defect quantification model to obtain the defect size of small-diameter oil and gas pipelines.

[0109] The above adjustment system in the present invention can effectively implement a small-diameter oil and gas pipe defect quantification method based on multi-modal magnetic leakage signals, and the technical effects that can be achieved are as described in the above embodiments, which will not be repeated here.

[0110] Furthermore, the quantitative model building module includes:

[0111] The data processing unit divides the original data in the defect feature sample library into a training set and a test set, and performs normalization processing on the original data;

[0112] A structure determination unit, which determines the number of indicators of the input layer, hidden layer and output layer of the ELMAN defect quantification model;

[0113] A parameter configuration unit, which configures the network parameters of the ELMAN defect quantization model and randomly initializes the weights of the input layer and the hidden layer, as well as the threshold of the hidden layer;

[0114] The model training unit uses the training set to train the ELMAN defect quantification model and uses the test set to evaluate the ELMAN defect quantification model.

[0115] Similarly, the above-mentioned optimization schemes for the system can also respectively achieve the optimization effects corresponding to the method in Example 1, which will not be repeated here.

[0116] Although the present application has been described in conjunction with specific features and embodiments thereof, it is apparent that various modifications and combinations may be made thereto without departing from the spirit and scope of the present application. Accordingly, this specification and the accompanying drawings are merely exemplary illustrations of the present application as defined herein, and are deemed to have covered any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, a person skilled in the art may make various modifications and variations to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application is intended to include these modifications and variations.

Claims

1. A method for quantifying defects in small-diameter oil and gas pipes based on multi-modal magnetic flux leakage signals, characterized in that: The method comprises: Establishing a small-diameter oil and gas pipeline model by using finite element simulation software, solving the small-diameter oil and gas pipeline model, and obtaining a simulation model solution result; Extracting triaxial defect magnetic leakage signals according to the solution results of the simulation model; Based on the three-axis defect magnetic leakage signal, extract the characteristic quantity of the three-axis defect magnetic leakage signal and establish a defect characteristic sample library; Constructing an ELMAN defect quantification model, and optimizing the ELMAN defect quantification model using a genetic algorithm; The defects of the small-diameter oil and gas pipeline are quantified by using the optimized ELMAN defect quantification model to obtain the defect size of the small-diameter oil and gas pipeline.

2. The method for quantifying defects of small-diameter oil and gas pipes based on multi-modal magnetic flux leakage signals according to claim 1 is characterized in that: The method of establishing a small-diameter oil and gas pipeline model by using finite element simulation software includes: using the finite element analysis software to establish a small-diameter oil and gas pipeline model, a magnetization unit model and four Hall sensors inside the small-diameter oil and gas pipeline, wherein the four Hall sensors are arranged circumferentially in sequence.

3. The method for quantifying small diameter oil and gas pipe defects based on multi-modal magnetic flux leakage signals according to claim 1 is characterized in that: The extracted three-axis defect leakage magnetic signal includes an axial component of the leakage magnetic signal, a radial component of the leakage magnetic signal and a circumferential component of the leakage magnetic signal. The extracted characteristic quantities of the three-axis defect leakage magnetic signal include the half-width of the leakage magnetic signal, the waveform area of ​​the leakage magnetic signal, the waveform energy of the leakage magnetic signal, the peak value of the leakage magnetic signal, the valley value of the leakage magnetic signal and the peak spacing of the leakage magnetic signal.

4. The method for quantifying defects of small-diameter oil and gas pipes based on multi-modal magnetic flux leakage signals according to claim 3 is characterized in that: The step of establishing a defect feature sample library comprises: The defect size in the small-diameter oil and gas pipeline model is changed, and the characteristic quantities of the triaxial defect magnetic leakage signals of different defect sizes are respectively extracted to establish a defect feature sample library.

5. The method for quantifying defects of small-diameter oil and gas pipes based on multi-modal magnetic flux leakage signals according to claim 1, characterized in that: The constructing of the ELMAN defect quantification model comprises: Dividing the original data in the defect feature sample library into a training set and a test set, and performing normalization processing on the original data; Determining the number of indicators of the input layer, hidden layer and output layer of the ELMAN defect quantification model; Configuring the network parameters of the ELMAN defect quantization model, and randomly initializing the weights of the input layer and the hidden layer, and the threshold of the hidden layer; The training set is used to train the ELMAN defect quantification model, and the test set is used to evaluate the ELMAN defect quantification model.

6. The method for quantifying defects of small-diameter oil and gas pipes based on multi-modal magnetic flux leakage signals according to claim 5 is characterized in that: The calculation formula for determining the number of indicators of the hidden layer is: Among them, h is the number of hidden layers, m is the number of input layers, n is the number of output layers, and a is an integer between 0 and 10.

7. The method for quantifying defects of small-diameter oil and gas pipes based on multi-modal magnetic flux leakage signals according to claim 1, characterized in that: The method of optimizing the ELMAN defect quantification model by using a genetic algorithm includes: Encoding the weights of the ELMAN defect quantization model and the initial values ​​of the thresholds to obtain an initial population, and setting fitness functions of individuals in the initial population; Selecting parent individuals from the individuals, performing crossover and mutation on the parent individuals, and obtaining offspring individuals; Calculating the fitness function of the offspring individuals, and merging the offspring individuals with the parent individuals to form a new offspring population; The number of iterations of the genetic algorithm is used as the iteration termination condition to determine whether the result of the genetic algorithm meets the iteration termination condition. If not, continue to execute step B3. If so, obtain the optimal weight and threshold.

8. The method for quantifying defects of small-diameter oil and gas pipes based on multi-modal magnetic flux leakage signals according to claim 7 is characterized in that: Setting the fitness function of individuals in the initial population includes: Each individual in the initial population is trained by the ELMAN algorithm to obtain a mean square error of each individual on the test set, and the inverse of the mean square error is used as the fitness function of the individual in the genetic algorithm.

9. A small diameter oil and gas pipe defect quantification system based on multi-modal magnetic flux leakage signals, characterized in that: The system comprises: A simulation model solving module is used to establish a small-diameter oil and gas pipeline model through finite element simulation software, solve the small-diameter oil and gas pipeline model, and obtain a simulation model solving result; A magnetic flux leakage signal extraction module is used to extract the magnetic flux leakage signal of the three-axis defect according to the solution result of the simulation model; A sample library establishment module, based on the three-axis defect magnetic leakage signal, extracts the characteristic quantity of the three-axis defect magnetic leakage signal and establishes a defect feature sample library; A quantization model building module, which builds an ELMAN defect quantization model and optimizes the ELMAN defect quantization model using a genetic algorithm; The defect size acquisition module quantifies the defects of the small-diameter oil and gas pipeline by using the optimized ELMAN defect quantification model to obtain the defect size of the small-diameter oil and gas pipeline.

10. The small-diameter oil and gas pipe defect quantification system based on multi-modal magnetic flux leakage signals according to claim 9, characterized in that: The quantitative model building module includes: A data processing unit, which divides the original data in the defect feature sample library into a training set and a test set, and performs normalization processing on the original data; A structure determination unit, which determines the number of indicators of the input layer, hidden layer and output layer of the ELMAN defect quantification model; A parameter configuration unit, configured with respect to the network parameters of the ELMAN defect quantization model, and randomly initialized the weights of the input layer and the hidden layer, and the threshold of the hidden layer; A model training unit is used to train the ELMAN defect quantification model using the training set, and to evaluate the ELMAN defect quantification model using the test set.

Images