Marine pipeline inner wall defect detection and evaluation method and system, electronic device and storage medium

Abstract

The invention relates to the technical field of marine pipelines, in particular to a marine pipeline inner wall defect detection and evaluation method and system, an electronic device and a storage medium, and the marine pipeline inner wall defect detection and evaluation method comprises the following steps: collecting marine pipeline operation data and marine pipeline inner wall magnetic field data in real time; carrying out data preprocessing on the collected marine pipeline operation data to obtain accurate data; constructing and integrating a pipeline entity three-dimensional geometric model, a pipeline physical behavior model and a pipeline performance evaluation model to obtain a digital twin model, and outputting a pipeline performance evaluation result; constructing a defect identification model by adopting a convolutional neural network architecture, and outputting a pipeline inner wall defect evaluation result; and constructing a visual interface, and generating a pipeline inner wall defect coping scheme in combination with the pipeline entity three-dimensional geometric model, the pipeline performance evaluation result and the pipeline inner wall defect evaluation result. The marine pipeline inner wall defect detection and evaluation system can perform real-time monitoring and predictive maintenance on the state of the marine pipeline, and can simulate behaviors and performance of the pipeline under different working conditions, so that potential structural damage, corrosion risks and maintenance requirements are predicted, the maintenance basis is optimized, and the operation cost is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of offshore pipelines, and particularly to a method, a system, an electronic device, and a storage medium for detecting and evaluating inner wall defects of offshore pipelines. Background Art

[0002] With the development of offshore oil and gas resources towards the deep sea, the safety guarantee of offshore pipelines is of vital importance. Offshore pipelines are in a complex and harsh marine environment for a long time, and are easily affected by factors such as corrosion, fatigue, and external force impact, resulting in defects such as cracks and corrosion pits on the inner wall, threatening the structural integrity and operation safety of the pipeline, and may cause serious accidents such as oil and gas leakage and explosion, resulting in huge economic losses and environmental pollution. Traditional non-destructive testing technologies such as ultrasonic testing, magnetic particle testing, and eddy current testing have limitations in the detection of offshore pipelines. For example, ultrasonic testing is interfered by marine environmental noise, magnetic particle testing requires surface pretreatment and is not applicable to non-ferromagnetic material pipelines, and eddy current testing has low sensitivity to deep defects. Therefore, it is urgent to develop an efficient and accurate technology for detecting inner wall defects of offshore pipelines. Summary of the Invention

[0003] The purpose of the present invention is to provide a method, a system, an electronic device, and a storage medium for detecting and evaluating inner wall defects of offshore pipelines to solve the problems mentioned in the above background art.

[0004] The technical solution adopted by the present invention to solve its technical problems is: a method for detecting and evaluating inner wall defects of offshore pipelines, the method for detecting and evaluating inner wall defects of offshore pipelines includes the following steps:

[0005] S1. Real-time collect the operation data of the offshore pipeline and the magnetic field data of the inner wall of the offshore pipeline;

[0006] S2. Perform data preprocessing on the collected operation data of the offshore pipeline to obtain precise data;

[0007] S3. Construct and integrate a three-dimensional geometric model of the pipeline entity, a pipeline physical behavior model, and a pipeline performance evaluation model to obtain a digital twin model, input the precise data, and output the pipeline performance evaluation result;

[0008] S4. Construct a defect recognition model using a convolutional neural network architecture, input the magnetic field data of the inner wall of the offshore pipeline, and output the evaluation result of the inner wall defects of the pipeline;

[0009] S5. Construct a visualization interface, combine the three-dimensional geometric model of the pipeline entity, the pipeline performance evaluation result, and the evaluation result of the inner wall defects of the pipeline, and generate a countermeasure plan for the inner wall defects of the pipeline.

[0010] Preferably, the step S1 includes the following steps:

[0011] S11. Collect the operation data of the subsea pipeline through the pressure sensors, strain sensors, temperature sensors, corrosion monitoring sensors and flow sensors deployed on the subsea pipeline;

[0012] S12. Collect the magnetic field data of the inner wall of the subsea pipeline through the ACFM detection probes deployed inside the subsea pipeline.

[0013] The ACFM detection probe includes:

[0014] An excitation coil for generating an alternating magnetic field;

[0015] A magnetic sensor for collecting magnetic field electrical signals;

[0016] A conditioning circuit for amplifying and filtering the magnetic field electrical signals;

[0017] An analog-to-digital conversion circuit for converting the magnetic field electrical signals into magnetic field digital signals.

[0018] Preferably, the installation positions of the sensors and the ACFM detection probes include elbows, tees, flange connections and historical fault-prone areas.

[0019] Preferably, the step S2 includes the following steps:

[0020] S21. Identify and remove the outliers in the subsea pipeline data based on the statistical analysis method to obtain the outlier-removed data;

[0021] S22. Filter the outlier-removed data by using a Butterworth low-pass filter to obtain the noise-reduced data;

[0022] S23. Use the linear normalization method to unify the dimensions of the noise-reduced data to obtain the standardized data;

[0023] S24. Process the standardized data based on the data fusion algorithm of Dempster-Shafer evidence theory to obtain the refined data.

[0024] Preferably, the transfer function of the Butterworth low-pass filter is

[0025] where s is the complex frequency variable, ω c is the cut-off frequency, and n is the filter order;

[0026] Determine the appropriate cut-off frequency according to the Nyquist sampling theorem.

[0027] Preferably, in the step S24, let the identification framework Θ include the normal operation state, slight defect state and severe defect state of the pipeline, and define the BPA function as follows:

[0028]

[0029] In the formula,

[0030] Preferably, the step S3 includes the following steps:

[0031] S31. According to the marine pipeline design drawings, use 3D software to model and obtain the 3D geometric model of the pipeline entity;

[0032] S32. Based on the theory of elasticity and the finite element method, construct a pipeline mechanical analysis model;

[0033] S33. Use empirical formulas to calculate the pipeline corrosion rate and construct a corrosion damage model;

[0034] S34. Based on Miner's linear cumulative damage theory, evaluate fatigue damage and construct a material degradation model;

[0035] S35. Collect the historical operation data, fault case records and maintenance records of the marine pipeline, and perform data preprocessing to construct a pipeline historical database;

[0036] S36. Extract characteristic parameters from the pipeline historical database to construct a characteristic vector set;

[0037] S37. Use machine learning algorithms to construct a pipeline performance evaluation model, and use the characteristic vector set to train and optimize the pipeline performance evaluation model;

[0038] S38. Integrate the 3D geometric model of the pipeline entity, the pipeline physical behavior model and the pipeline performance evaluation model to obtain a digital twin model. The pipeline physical behavior model includes a pipeline mechanical analysis model, a corrosion damage model and a material degradation model;

[0039] S39. Input precise data into the digital twin model and output the pipeline performance evaluation result.

[0040] Preferably, the historical operation data of the marine pipeline includes pressure, flow rate and temperature time series data. The fault case records include fault types, occurrence locations, occurrence times and maintenance measures. The maintenance records include maintenance times, maintenance contents and details of replaced components.

[0041] Preferably, in step S4, the defect identification model includes:

[0042] An input layer for receiving the magnetic field data of the inner wall of the marine pipeline. The magnetic field data of the inner wall of the marine pipeline is two-dimensional image data;

[0043] Convolutional layer and pooling layer for extracting the characteristics of the magnetic field data of the inner wall of the marine pipeline;

[0044] A fully connected layer for integrating the characteristics of the magnetic field data of the inner wall of the marine pipeline;

[0045] An output layer for outputting the evaluation result of the inner wall defect of the pipeline, where the evaluation result of the inner wall defect of the pipeline includes the category, corresponding probability and distribution of the inner wall defect of the pipeline.

[0046] Preferably, the step S5 includes the following steps:

[0047] S51. Import the digital twin model into the game engine to construct a visualization interface;

[0048] S52. Locate the defect position in the visualization interface and mark the defect information;

[0049] S53. Construct a pipeline operation and maintenance strategy library, where the pipeline operation and maintenance strategy library includes maintenance and repair strategies and parameter adjustment rules under different defect types, size ranges and pipeline operation conditions;

[0050] S54. Screen the adapted strategy combination from the pipeline operation and maintenance strategy library through a rule matching algorithm to generate a solution for dealing with the inner wall defect of the pipeline.

[0051] An inner wall defect detection and evaluation system for an offshore pipeline, the inner wall defect detection and evaluation system for an offshore pipeline includes:

[0052] A data acquisition module for real-time acquisition of the operation data of the offshore pipeline and the magnetic field data of the inner wall of the offshore pipeline;

[0053] A data processing and transmission module for preprocessing the acquired operation data of the offshore pipeline to obtain refined data and transmitting it to the computing platform;

[0054] A digital twin model construction module for integrating and constructing a three-dimensional model of the pipeline entity, a pipeline physical behavior model and a pipeline performance evaluation model to form a digital twin model and output the pipeline performance evaluation result;

[0055] A defect intelligent recognition and evaluation module for constructing a defect recognition model and outputting the evaluation result of the inner wall defect of the pipeline;

[0056] A visualization and decision support module for visually outputting the three-dimensional geometric model of the pipeline entity, the pipeline performance evaluation result and the evaluation result of the inner wall defect of the pipeline, and generating a solution for dealing with the inner wall defect of the pipeline.

[0057] An electronic device, the electronic device includes a processor, a memory, a communication interface and a communication bus, and the processor, the memory and the communication interface complete mutual communication through the communication bus;

[0058] The memory is used to store the program for detecting and evaluating the inner wall defect of the offshore pipeline;

[0059] The processor is used to execute the program for detecting and evaluating the defects on the inner wall of the submarine pipeline. When the program for detecting and evaluating the defects on the inner wall of the submarine pipeline is executed, the steps of the method for detecting and evaluating the defects on the inner wall of the submarine pipeline as described above are implemented.

[0060] A computer-readable storage medium stores a program for detecting and evaluating the defects on the inner wall of the submarine pipeline. When the program for detecting and evaluating the defects on the inner wall of the submarine pipeline is executed by a processor, the steps of the method for detecting and evaluating the defects on the inner wall of the submarine pipeline as described above are implemented.

[0061] The beneficial effects of the present invention are as follows:

[0062] 1. Through the sensor network, the present invention can collect and analyze the submarine pipeline data in real time. By constructing a digital twin model and a defect recognition model, the state of the submarine pipeline can be monitored in real time and predictive maintenance can be carried out.

[0063] 2. The digital twin model of the present invention integrates marine environmental data, pipeline material characteristics and operating conditions, and can simulate the behavior and performance of the pipeline under different working conditions, so as to predict potential structural damage, corrosion risks and maintenance requirements, thereby optimizing the maintenance base and reducing the operation cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only the embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained according to the provided drawings without creative efforts.

[0065] Figure 1 It is a schematic diagram of the equivalent calculation model (a) of the non-defect area and the equivalent calculation model (b) of the defect area of the present invention;

[0066] Figure 2 It is a schematic structural diagram of the submarine pipeline of the present invention;

[0067] Figure 3 It is a cross-sectional view of the "H-H" of the submarine pipeline of the present invention;

[0068] Figure 4 It is a flowchart of the method for detecting and evaluating the defects on the inner wall of the submarine pipeline of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0069] It should be noted that the following detailed description is illustrative and is intended to provide further description of the present application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present application belongs.

[0070] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly indicates otherwise, the singular forms are also intended to include the plural forms. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, they specify the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0071] The present invention will be further described in conjunction with specific examples below. The following embodiments are only for explaining the present invention and do not constitute a limitation to the present invention. The test samples and test procedures used in the following embodiments include the following content (if the specific test conditions are not specified in the embodiments, they are usually in accordance with conventional conditions or the conditions recommended by the reagent company; the reagents, consumables, etc. used in the following embodiments, unless otherwise specified, can be obtained from commercial channels).

[0072] One of the objectives of the present invention is to provide a method for detecting and evaluating inner wall defects of marine pipelines. The specific steps of this method for detecting and evaluating inner wall defects of marine pipelines are as follows:

[0073] S1. Real-time collect the operation data of the marine pipeline and the magnetic field data of the inner wall of the marine pipeline;

[0074] S2. Perform data preprocessing on the collected operation data of the marine pipeline to obtain refined data;

[0075] S3. Construct and integrate a three-dimensional geometric model of the pipeline entity, a pipeline physical behavior model, and a pipeline performance evaluation model to obtain a digital twin model and output the pipeline performance evaluation result;

[0076] S4. Use a convolutional neural network architecture to construct a defect recognition model and output the evaluation result of the inner wall defects of the pipeline;

[0077] S5. Construct a visualization interface, combine the three-dimensional geometric model of the pipeline entity, the pipeline performance evaluation result, and the evaluation result of the inner wall defects of the pipeline, and generate a response plan for the inner wall defects of the pipeline.

[0078] Step S1 includes the following steps:

[0079] S11. Collect the operation data of the marine pipeline through the pressure sensors, strain sensors, temperature sensors, corrosion monitoring sensors, and flow sensors deployed on the marine pipeline;

[0080] S12. Collect the magnetic field data of the inner wall of the marine pipeline through the ACFM detection probes deployed inside the marine pipeline,

[0081] The ACFM detection probe includes:

[0082] An excitation coil for generating an alternating magnetic field;

[0083] A magnetic sensor for collecting magnetic field electrical signals;

[0084] A conditioning circuit for amplifying and filtering the magnetic field electrical signals;

[0085] An analog-to-digital conversion circuit for converting the magnetic field electrical signals into magnetic field digital signals.

[0086] According to the characteristics of the pipeline material and the properties of the transmission medium, determine the sampling frequencies of the above-mentioned various sensors, set a high sampling rate for parameters with fast dynamic changes, and set a low sampling rate for slow-changing parameters, balance data accuracy and storage and transmission burdens, and achieve all-round real-time monitoring of the pipeline operation status.

[0087] The mechanism of alternating current electromagnetic field testing is as follows:

[0088]

[0089] Where, is the induced current density at a certain depth in the conductor; is the induced current density on the surface of the conductor; z is the coordinate position of a certain point in the thickness direction of the conductor; f is the frequency of the coil excitation signal; μ is the magnetic permeability of the conductor, μ = μ 0 μ r μ 0 is the vacuum magnetic permeability, μ r is the relative magnetic permeability of the conductor; σ is the conductivity of the conductor; ω is the angular frequency of the excitation signal, ω = 2πf.

[0090] The phenomenon that the induced current gradually decays in the thickness direction of the conductor is called the skin effect, and the skin depth is inversely proportional to the conductivity, magnetic permeability of the material and the frequency of the excitation signal

[0091]

[0092] As Figure 1 shown, when the defect completely enters the simplified area, the magnetic flux density of the electromagnetic field signal of the inner wall defect at the detection position is:

[0093]

[0094] Where, μ 0 is the vacuum magnetic permeability, K is the current density, B 1 is the magnetic field generated in the non-defect area, B 2 is the magnetic field generated in the defect area.

[0095] Step S2 includes the following steps:

[0096] S21. Identify and eliminate outliers in the marine pipeline data based on statistical analysis methods to obtain outlier-removed data;

[0097] S22. Filter the outlier-removed data using a Butterworth low-pass filter to obtain noise-reduced data;

[0098] S23. Use the linear normalization method to unify the dimensions of the noise-reduced data to obtain standardized data;

[0099] S24. Process the standardized data based on the data fusion algorithm of Dempster-Shafer evidence theory to obtain refined data.

[0100] In step S21, for the pressure data, calculate its mean μ and standard deviation σ, and determine the data points that deviate from the mean by more than 3σ as outliers and remove them. For example, if a pressure data point P i satisfies |P i -μ|>3σ, then it is removed from the dataset. For the flow rate data, an outlier can be detected by constructing a time series model of the flow rate data, such as an autoregressive moving average model (ARMA), and using the model residuals. If the residuals exceed a preset threshold, the corresponding data points are regarded as outliers and cleared.

[0101] In step S22, the transfer function of the Butterworth low-pass filter is where s is the complex frequency variable, ω c is the cut-off frequency, and n is the filter order. Determine the appropriate cut-off frequency according to the Nyquist sampling theorem. For example, for data with a sampling frequency of f s , the cut-off frequency can be set to f c =0.1f s (adjusted according to the actual signal spectrum characteristics), filter the slowly varying signals such as pressure and temperature to effectively remove high-frequency noise interference and retain the effective components of the signals.

[0102] In step S23, for a certain data sequence x = [x 1 , x 2 , …, x n , its normalization formula is where x min and x max are the minimum and maximum values in the data sequence respectively. For example, normalize the pressure data to the [0, 1] interval. After this processing, the data collected by different sensors can be compared under the same dimension, which is convenient for subsequent fusion and analysis.

[0103] In step S24, assume that the identification framework Θ includes the normal operation state A, the minor defect state B, and the serious defect state C of the pipeline. For the evidence from different sensors, such as the evidence m 1 of the pressure sensor and the evidence m2 , calculate its basic probability assignment (BPA) function. Assume that the confidence level that the pressure sensor detects the data at a certain moment conforms to the normal state is m 1 (A) = 0.6, and the uncertainty is m 1 (Θ) = 0.4; the confidence level of the detection result of the strain sensor corresponding to the normal state is m 2 (A) = 0.7, and the uncertainty is m 2 (Θ) = 0.3. Use the Dempster combination rule to calculate the fused BPA function:

[0104]

[0105]

[0106] In the formula, m 1 and m 2 are the basic probability assignment functions regarding the pipeline state provided by the pressure sensor and the strain sensor respectively. They represent the confidence levels of each sensor for different pipeline states. m(A), m(B), m(C), m(θ) are the basic probability assignment values in different states calculated based on the evidence of the pressure sensor and the strain sensor. X and Y are intermediate variables in calculating the basic probability assignment, used to represent the state combinations involved in the evidence of different sensors. ∑ X∩Y＝C m 1 (X)m 2 (Y) represents the joint probability that the pipeline is in state A, represents the conflict degree between the evidences of the two sensors, that is, there are inconsistent parts in the judgments of different states, ∑ X∩Y＝α m 1 (X)m 2 (Y), represents the joint probability that the pipeline is in an unknown state.

[0107] Through the fusion calculation, integrating the information of each sensor, the pipeline state can be judged more accurately, the uncertainty can be reduced, and the data credibility can be improved.

[0108] The obtained refined data is packaged according to the TCP / IP protocol. At the sending end, the refined data is divided into data packets of appropriate sizes, and header information including source address, destination address, packet sequence number, and checksum is added.

[0109] The step S3 includes the following steps:

[0110] S31. According to the marine pipeline design drawings, use 3D software to model and obtain the 3D geometric model of the pipeline entity;

[0111] S32. Based on the theory of elasticity and the finite element method, construct a pipeline mechanical analysis model;

[0112] S33. Calculate the pipeline corrosion rate using empirical formulas and construct a corrosion damage model;

[0113] S34. Evaluate fatigue damage based on Miner's linear cumulative damage theory and construct a material degradation model;

[0114] S35. Collect the historical operation data, fault case records, and maintenance records of the subsea pipeline, and perform data preprocessing to construct a pipeline historical database;

[0115] S36. Extract characteristic parameters from the pipeline historical database to construct a feature vector set;

[0116] S37. Use machine learning algorithms to construct a pipeline performance evaluation model, and use the feature vector set to train and optimize the pipeline performance evaluation model;

[0117] S38. Integrate the pipeline entity three-dimensional geometric model, pipeline physical behavior model, and pipeline performance evaluation model to obtain a digital twin model. The pipeline physical behavior model includes a pipeline mechanical analysis model, a corrosion damage model, and a material degradation model;

[0118] S39. Input precise data into the digital twin model and output the pipeline performance evaluation results.

[0119] Among them, part of the structure of the subsea pipeline is as Figure 2 and Figure 3 shown. The subsea pipeline design drawings include the following information: pipeline outer diameter D, inner diameter d, length L, and wall thickness t. Use computer-aided design (CAD) software (such as SolidWorks, AutoCAD, etc.) to construct a pipeline entity three-dimensional geometric model. Taking a straight pipe section as an example, in CAD software, generate the basic shape of the pipeline by stretching a two-dimensional circular contour (radius D / 2), and determine the stretching distance according to the length L; for the elbow part, create a curved shape using surface modeling tools based on the elbow curvature radius R and angle; for pipe fittings such as tees and flanges, accurately model them according to the standard specification size parameters and then combine them with the pipeline main body. After the model construction is completed, convert it to a general three-dimensional model format (such as STL, STEP, etc.) for subsequent processing, ensuring that the model accurately reflects the actual geometric characteristics of the pipeline and providing a basic framework for physical behavior simulation and performance evaluation.

[0120] When constructing the mechanical analysis model, consider the complex loads in the marine environment, such as internal fluid pressure P, external seawater hydrostatic pressure P sw , ocean current impact force F c , wave load F w , and pipeline self-weight G, etc. Based on the theory of elasticity and the finite element method, construct the pipeline mechanical equilibrium equation:

[0121]

[0122] Among them, σ is the stress tensor, and F is the body force vector (including the spatial distribution of the above-mentioned various load terms). The pipeline is discretized into finite elements (commonly tetrahedral or hexahedral elements), the displacement field within the element is represented by the nodal displacement interpolation function, and combined with the material constitutive relationship (such as Hooke's law for linearly elastic materials σ = Eε, where E is the elastic modulus matrix and ε is the strain tensor), the nodal displacements and stress-strain distributions are solved. For example, for a four-node rectangular element in a two-dimensional plane stress problem, the nodal displacement vector u e = [u 1 , v 1 , u 2 , v 2 , u 3 , v 3 , u 4 , v 4 T .

[0123] In step S35, the historical operation data of the subsea pipeline includes pressure, flow rate, and temperature time series data, the fault case records include fault types, occurrence locations, occurrence times, and maintenance measures, and the maintenance records include maintenance times, maintenance contents, and details of replaced components. The above data is filtered by moving average to remove the fluctuation noise of pressure and flow rate data, and the missing values are filled by mean filling, linear interpolation, or multiple imputation methods according to the data distribution characteristics. The categorical data (such as fault types) is encoded and quantified to construct a complete pipeline historical database.

[0124] In step S36, the parameters reflecting the pipeline performance characteristics are extracted from the pipeline historical database, such as the standard deviation of pressure fluctuation σ p to measure pressure stability, the coefficient of variation of flow rate (σ f is the standard deviation of flow rate, and μ f is the mean flow rate) to evaluate flow uniformity, and the rate of change of temperature gradient to monitor abnormal heat transfer; the fault characteristics are extracted from the fault cases, such as the sudden change amplitude of pressure ΔP fault , the proportion of flow rate mutation , etc.; the maintenance cycle T maintenance . and maintenance cost C maintenance , etc. are extracted from the maintenance records. Key features are screened by methods such as principal component analysis (PCA) or correlation analysis, and the main information is retained while reducing the dimension to construct a feature vector set.

[0125] In step S37, a pipeline performance evaluation model is constructed using a support vector machine (SVM) or a neural network. Taking the SVM as an example, the labeled pipeline state data (normal, minor fault, severe fault) divides the feature vector set into a training set and a test set, and the kernel function is optimized (such as the Gaussian kernel function K(x​i , x j ) = exp(-γ||x i - x j || 2 ), where γ is the kernel parameter) and the penalty parameter C are optimized through grid search or genetic algorithm to minimize the classification error or prediction error function (such as the mean square error y i is the true value, is the predicted value), improve the prediction accuracy and generalization ability of the model, and achieve accurate identification and evaluation of the pipeline performance degradation trend and potential failure modes.

[0126] The above digital twin model can be deployed in a cloud computing platform (such as Amazon Web Services, Microsoft Azure, etc.). The core computing module of the model is deployed in the cloud to process large-scale data-intensive simulation analysis; a lightweight real-time response module is deployed at the edge computing node near the pipeline site (such as an Internet of Things gateway with computing capabilities) to process real-time data collection, local rapid analysis, and issuance of emergency control instructions. According to the pipeline operation conditions and data processing requirements, cloud computing and edge computing resources are dynamically allocated to collaboratively achieve the efficient operation of the digital twin model, ensuring the timeliness and accuracy of real-time pipeline status monitoring, predictive analysis, and generation of accurate control instructions.

[0127] In step S4, the defect identification model includes:

[0128] An input layer for receiving the magnetic field data of the inner wall of the submarine pipeline, where the magnetic field data of the inner wall of the submarine pipeline is two-dimensional image data;

[0129] A convolutional layer for extracting the features of the magnetic field data of the inner wall of the submarine pipeline;

[0130] A pooling layer for reducing the computational complexity and improving the operation efficiency of the defect identification model;

[0131] A fully connected layer for integrating the features of the magnetic field data of the inner wall of the submarine pipeline;

[0132] An output layer for outputting the evaluation results of the inner wall defects of the pipeline, where the output evaluation results of the inner wall defects of the pipeline include the category, corresponding probability, and distribution of the inner wall defects of the pipeline.

[0133] The two-dimensional image data is obtained by mapping the electromagnetic field signal intensity to the image pixel value according to the probe scanning position, with a size of n×n (determined according to the acquisition resolution); the output layer gives the category, corresponding probability, and distribution of the inner wall defects of the pipeline through the Softmax function.

[0134] The training and optimization steps of the above defect identification model are as follows:

[0135] S41. Collect ACFM detection data with known defect types, sizes, and location markings;

[0136] S42. Divide the ACFM detection data into a training set, a validation set, and a test set, with the ratio of the training set, the validation set, and the test set being 6:2:2;

[0137] S43. Train a defect recognition model using the backpropagation algorithm, derive and update the model parameters according to the loss function, and adjust the parameters using the stochastic gradient descent optimization algorithm;

[0138] S44. After multiple rounds of iterative training, the loss function converges and the model accuracy reaches the set target value.

[0139] In step S43, set the learning rate to 0.001, and adjust this learning rate according to the training convergence situation.

[0140] In step S44, set the target value to the accuracy of the test set, which can be set to 90%, and use the validation set early stopping method to prevent overfitting.

[0141] The said step S5 includes the following steps:

[0142] S51. Import the digital twin model into the game engine to build a visualization interface;

[0143] S52. Locate the defect position in the visualization interface and mark the defect information;

[0144] S53. Build a pipeline operation and maintenance strategy library, which includes maintenance and repair strategies and parameter adjustment rules under different defect types, size ranges, and pipeline operating conditions;

[0145] S54. Screen the appropriate strategy combinations from the pipeline operation and maintenance strategy library through a rule matching algorithm to generate a solution for dealing with defects on the inner wall of the pipeline.

[0146] In step S51, use professional game engines such as Unity or Unreal Engine to build a visualization environment. Import the pipeline digital twin model, render the three-dimensional geometric model of the pipeline entity, and set material properties (such as the reflectivity and roughness of the metal material) and lighting effects (simulating the light scattering and refraction in the ocean environment) to enhance the realism. Use model simplification and LOD (Level of Detail) technology to optimize the rendering performance, and dynamically switch the model detail level according to the user's perspective and distance. For example, when the user browses from a distance, the pipeline contour is displayed with a low-precision model, and when observing the defect part closely, it switches to a high-detail model. Implement user interaction with the model through the collision detection algorithm. The user can click on the pipeline model to query part parameters, rotate and scale the model to view it comprehensively, and use the ray casting algorithm to determine the model elements and data associations corresponding to the click position, providing an intuitive pipeline visualization management solution.

[0147] In step S52, according to the obtained evaluation results of the inner wall defects of the pipeline, accurate positioning and marking are carried out on the three-dimensional geometric model of the pipeline entity. Taking crack defects as an example, the crack shape is drawn at the corresponding inner wall position of the model according to the crack geometric parameters (length, width, orientation), and different color codes are used to represent the severity of the defects (for example, red for severe, yellow for moderate, and green for mild). The defect attribute information (type, size, detection time, remaining evaluation life, etc.) is associated, and an information box pops up to display details when the mouse hovers or clicks. For corrosion pits, in a similar way, parameters such as depth and area are drawn and marked on the model surface according to their shape and size, and the defect distribution density and gradient changes are intuitively presented by combining heat maps or contour maps, helping the operation and maintenance personnel quickly understand the overall situation and local details of pipeline defects, and improving the efficiency of defect identification and analysis.

[0148] In step S53, the pipeline operation and maintenance strategy library includes maintenance and repair strategies and parameter adjustment rules under different defect types, size ranges, and pipeline operation conditions. For example, for mild crack defects with a length less than 5mm, the strategy is to monitor regularly (interval of 1 month) and moderately reduce the operating pressure (10%); for moderate crack defects with a length of 5 - 10mm, the next maintenance time is planned within 1 week, and the repair measures are welding repair or installation of repair jigs, and the flow and pressure parameters are adjusted to ensure safe operation while reducing the impact on production.

[0149] In step S54, the rule matching algorithm can select the If-Then rule logic to select an appropriate strategy combination from the pipeline operation and maintenance strategy library to generate a response plan for pipeline inner wall defects. The response plan for pipeline inner wall defects includes a maintenance and repair plan, a scheduling plan, and suggestions for adjusting operating parameters. The maintenance and repair plan includes the breakdown of repair tasks (processes, man-hours, list of required tools and materials), the skill requirements of repair personnel, and task allocation; the scheduling plan includes the priority of repair tasks, the location and availability of resources to optimize the repair operation schedule, improve resource utilization and repair efficiency; the suggestions for adjusting operating parameters can be simulated through a digital twin model, and then appropriate operating parameters are adjusted, such as evaluating the flow field characteristics and force states of the pipeline after adjusting the flow and pressure based on fluid mechanics simulation.

[0150] When formulating a scheduling plan, a risk matrix can be introduced to evaluate the operation risk of pipeline defects. The risk level (high, medium, low) is divided according to the probability of defect occurrence (predicted by historical failure data and machine learning models, such as predicting the crack propagation probability based on the Bayesian network fusion of multiple factors) and the severity of consequences (evaluated according to the amount of oil and gas leakage caused by pipeline failure, the scope of environmental impact, and economic losses). High-risk defects are given priority for treatment, and decision-making strategies are adjusted to strengthen prevention and control measures (such as increasing the monitoring frequency, immediately stopping production for maintenance, and implementing redundant backups); for medium- and low-risk defects, the maintenance cost and risk are balanced according to resources and production conditions, and the decision-making is optimized to maximize the maintenance benefits. A multi-objective decision-making algorithm (such as the Analytic Hierarchy Process AHP combined with the Fuzzy Comprehensive Evaluation Method) is used to comprehensively consider the objective weights such as maintenance cost, maintenance time, production loss, and improvement of pipeline reliability, optimize the sorting of decision-making schemes, provide a scientific and quantitative decision-making basis for operation and maintenance personnel, improve the accuracy and rationality of decision-making, and ensure that the pipeline operation and maintenance decision-making takes into account the balance of safety, economy, and efficiency goals.

[0151] In one implementation, a remote monitoring and collaboration platform can be built based on a hybrid communication architecture of WebSocket and RESTful API. WebSocket enables real-time two-way data transmission, which is used to push the real-time obtained marine pipeline operation data, pipeline performance evaluation results, and pipeline inner wall defect evaluation results to ensure real-time synchronization of data between the remote end and the site; RESTful API is provided for the client to query historical data, obtain operation and maintenance reports, and issue control instructions (such as remotely adjusting equipment operation parameters and approving maintenance plans). A data caching and incremental update strategy is adopted to reduce the data transmission volume and improve communication efficiency. For example, recent high-frequency access data is cached at the edge node, and incremental updates are intelligently pushed according to the data change amount and timestamp, ensuring that remote users can obtain the latest data while reducing the network load and guaranteeing the timeliness and smooth interactivity of platform data.

[0152] A second object of the present invention is to provide a marine pipeline inner wall defect detection and evaluation system, which includes a data acquisition module, a data processing and transmission module, a digital twin model construction module, a defect intelligent identification and evaluation module, and a visualization and decision support module.

[0153] The process of building the above-mentioned marine pipeline inner wall defect detection and evaluation system is as follows:

[0154] Hardware Installation and Configuration: According to the actual working conditions of the pipeline and design requirements, install pressure sensors, strain sensors, temperature sensors, corrosion monitoring sensors, flow sensors, and ACFM detection probes at key positions along the pipeline (such as elbows, tees, flange connections, and areas with frequent historical failures). The sensor selection is determined based on factors such as measurement accuracy, environmental adaptability, and reliability. For example, high-precision seawater corrosion-resistant pressure sensors and high-temperature-resistant strain sensors are selected. After each sensor is installed according to the predetermined layout plan, it is connected to the local data acquisition unit for communication configuration (such as setting IP addresses, communication protocols, etc.) and calibration operations (calibrating parameters such as zero point and range according to standard metrology equipment) to ensure accurate and reliable data acquisition. The ACFM detection probe optimizes the installation parameters (such as quantity, spacing, angle) according to the pipeline material, diameter, and wall thickness to ensure high-sensitivity detection of inner wall defects.

[0155] Software Platform Construction and Debugging: Install the software for detecting and evaluating the inner wall defects of submarine pipelines on the data processing center server. Configure the software operating environment, including the operating system, database management system, middleware, etc., to ensure the stable operation of the software. Debug the data processing software and set the parameters of data cleaning, filtering, and normalization algorithms (such as wavelet basis functions for wavelet transform, filter cut-off frequencies, normalization ranges, etc.) to verify the accuracy and real-time performance of data processing. Import the pipeline design parameters, material properties, historical operation data, etc. into the digital twin model construction platform to initialize the model, and debug the model calculation modules (such as mechanical analysis algorithms, corrosion damage model parameters, machine learning model training and evaluation) to ensure that the model simulation results are consistent with the actual pipeline behavior. The visualization and decision support system software configures the connection parameters of virtual reality and augmented reality devices and develops a user interface to ensure intuitive and convenient operation and information acquisition for operation and maintenance personnel.

[0156] Data Acquisition and Transmission

[0157] Real-time Data Acquisition by Sensors: After the system runs, each sensor collects the pipeline operation data in real time according to the set sampling frequency. The pressure sensor monitors the pressure change of the fluid in the pipeline, the strain sensor measures the strain response of the pipeline structure, the temperature sensor records the temperature distribution of the pipeline, the corrosion monitoring sensor detects the corrosion state of the pipeline (such as electrochemical corrosion potential, corrosion current), the flow sensor obtains the oil transmission flow information, and the ACFM detection probe periodically scans the inner wall of the pipeline to collect electromagnetic field signals for defect detection. The pressure sensor collects data every 10 milliseconds, and the flow sensor dynamically adjusts the sampling rate according to the flow change (increasing the sampling rate to 10 times per second when the flow fluctuates greatly and decreasing it to 1 time per second when it is stable).

[0158] Data Preprocessing and Transmission Optimization: After receiving sensor data, the local data acquisition unit performs real-time preprocessing operations. The data cleaning module removes noise and abnormal data according to statistical analysis methods (such as calculating the mean and standard deviation to eliminate outliers) and time series models (such as using the ARMA model to detect abnormal flow data); data filtering uses a Butterworth low-pass filter (setting the cut-off frequency according to the signal spectrum characteristics) to smooth the data and eliminate high-frequency interference; normalization linearly normalizes data with different physical dimensions to a unified range (such as 0-1 or -1-1) to facilitate multi-source data fusion analysis. The preprocessed data is packaged, added with identification information such as timestamps and sensor numbers, and transmitted to the data processing center through a hybrid network mainly based on the TCP / IP protocol via a high-speed optical fiber communication network and supplemented by a wireless communication network. During the transmission process, the data sender monitors the network status in real time. In case of network congestion, it automatically adjusts the transmission strategy or switches to the backup communication link to ensure the real-time and integrity of the data. After receiving the data, the receiver verifies the data accuracy according to information such as checksum and packet sequence number, and automatically requests retransmission when an error occurs to ensure that the data is correctly delivered to the data processing center.

[0159] Digital Twin Model Construction and Update

[0160] Construction and Verification of the Three-dimensional Geometric Model of the Pipeline Entity: Extract parameters such as outer diameter, inner diameter, length, wall thickness, and fitting specifications from the pipeline design drawings and completion data, and accurately construct the three-dimensional geometric model of the pipeline entity in SolidWorks software. After the construction of the three-dimensional geometric model of the pipeline entity is completed, compare the design drawings with the actual pipeline dimensions for geometric accuracy verification, and control the deviation within ±0.5%. The model is converted to the common STL format and imported into the digital twin model platform, and integrated with the subsequent pipeline physical behavior model and pipeline performance evaluation model.

[0161] Real-time Simulation and Correction of the Pipeline Physical Behavior Model: Combining ocean environmental monitoring data (such as wave, current, and tidal information) and ocean pipeline operation data, based on the theories of elasticity mechanics, fluid mechanics, and the finite element method, use the pipeline physical behavior model to simulate in real time the physical behaviors such as the stress-strain distribution, fluid flow state, and corrosion damage evolution of the pipeline. For example, use finite element software such as ANSYS to simulate the mechanical responses of the pipeline under different working conditions (such as normal operation and extreme weather), and analyze the stress concentration areas and deformation conditions; predict the corrosion development trend according to the corrosion monitoring data and empirical corrosion models. Compare the simulation results with the on-site measured data (such as the actual strain values measured by strain gauges and the thickness data of regular corrosion inspections), and calculate the error index to evaluate the model accuracy. If the error exceeds the threshold, correct the model parameters (such as adjusting the material constitutive relationship parameters and corrosion rate coefficients) or improve the model algorithm according to the error feedback to ensure that the model can reliably reflect the physical behavior of the pipeline. At the same time, the model receives the updated sensor data in real time, dynamically adjusts the simulation calculation, and realizes synchronous operation with the actual pipeline.

[0162] Training and Optimization of Pipeline Performance Evaluation Model: Collect historical operation data, fault case records, and maintenance records of the pipeline to build a pipeline historical database. Extract feature parameters from the pipeline historical database, perform dimensionality reduction processing using methods such as principal component analysis (PCA), and screen key features to construct a feature vector set. Use machine learning algorithms (such as support vector machine SVM or neural network such as LSTM) to train the pipeline performance evaluation model. Classify and label historical data according to the pipeline state, and divide it into training set, validation set, and test set with a ratio of 7:2:1. During the training process, adjust the model hyperparameters (such as SVM kernel function parameters, penalty coefficients, neural network learning rate, number of hidden layer nodes) according to the performance indicators of the validation set (such as accuracy, recall rate, F1 value). Use cross-validation and grid search to optimize the parameter combination to improve the generalization ability and prediction accuracy of the model. After the model training is completed, evaluate the performance with the test set. After meeting the predetermined indicators (such as accuracy exceeding 90%), put it into use, and update the training regularly according to new data to adapt to the changes in pipeline performance.

[0163] Intelligent Identification and Evaluation of Defects

[0164] Intelligent Defect Identification Process Based on ACFM Data: The magnetic field data on the inner wall of the subsea pipeline collected by the ACFM detection probe is digital image data, which is input into a pre-trained convolutional neural network (CNN) model for defect identification. The model architecture can choose classic networks such as AlexNet, VGG, or be improved and optimized according to actual needs. The image data is first preprocessed by operations such as normalization, cropping, and enhancement to improve the robustness and generalization ability of the model. After input into the model, local features (such as edges, textures) are extracted through the convolutional layer, the pooling layer reduces the dimensionality to reduce the computational amount, and the fully connected layer integrates the features and classifies and outputs the defect type and probability. For example, if the model outputs a crack probability of 0.85, a corrosion pit probability of 0.1, and a non-defect probability of 0.05, it is determined as a crack defect. While the model identifies the defect, according to the mapping relationship between the position information of the feature map and the probe scanning coordinates, accurately locate the position of the defect in the pipeline model coordinate system (such as the distance from the pipeline starting point, circumferential angle), realizing intelligent defect identification and positioning.

[0165] Defect stress-strain and remaining strength assessment and calculation: For the identified defects, according to their types, sizes, positions, and pipeline operating parameters (pressure, axial force, bending moment), a finite element model of the pipeline with defects is constructed in the digital twin model. High-order solid elements (such as 20-node hexahedral elements) are used to accurately simulate the stress-strain field near the defects, and singular elements are set at the crack tip to simulate stress singularity. According to the finite element calculation results, obtain the stress-strain distribution nephogram around the defects, and calculate the stress-strain values at key positions (such as the stress intensity factor at the crack tip). Calculate the remaining strength of the pipeline according to fracture mechanics theories (Griffith criterion, J-integral criterion), and evaluate the impact of defects on pipeline safety. For crack-type defects, compare the calculated stress intensity factor with the material fracture toughness to judge the pipeline failure risk; for corrosion pit defects, calculate the remaining wall thickness bearing capacity according to the net section stress method, and determine the safe operating range in combination with pipeline design specifications, providing a key mechanical parameter basis for pipeline maintenance and repair decisions.

[0166] Defect fatigue life assessment and prediction: For crack-type defects, calculate the fatigue life according to the Paris crack growth law Calculate the fatigue life. Extract the sequence of stress intensity factor amplitude changes (ΔK) from the pipeline operation load spectrum, and integrate according to the material constants C and m (fitted from material test data) to calculate the number of cycles required for the crack to grow from the initial length to the critical length (fatigue life). Calculate and correct the parameters of the Paris formula considering the crack closure effect and environmental factors (such as seawater corrosion accelerating crack growth) to improve the prediction accuracy. For the fatigue life assessment of corrosion pits, combine the corrosion pit growth model (considering the interaction between corrosion rate and fatigue) and the fatigue damage accumulation theory (such as the Miner criterion) to estimate the number of cyclic loadings that the pipeline can withstand before fatigue failure caused by corrosion pits, comprehensively evaluate the impact of defects on the long-term operation fatigue life of the pipeline, and provide a scientific basis for optimizing the pipeline inspection cycle and safety management.

[0167] Visualization and decision support

[0168] Visual Interface Interaction Operations and Information Display: Operation and maintenance personnel access the visualization and decision support system through virtual reality (VR), augmented reality (AR) devices or ordinary computer terminals. According to the user's permissions and operation instructions, the system displays the pipeline digital twin model and related information on the three-dimensional visualization interface. Users can operate the model through interactive methods such as handles and gesture recognition, such as zooming in, rotating, and panning to view the overall and local details of the pipeline; clicking on the model part to query real-time operating parameters (pressure, temperature, flow rate), historical data trends, and sensor distribution locations; viewing the defect detection and evaluation results, including the defect list (type, location, size, severity), positioning and marking on the model, and a detailed information box (including detection time, remaining evaluation life, and treatment suggestions) pops up when the mouse hovers over or clicks on the defect. The system supports multi-perspective observation (such as viewing the inner wall condition of the pipeline from the internal perspective) and cross-section analysis (displaying the stress and strain distribution of the pipeline cross-section), helping operation and maintenance personnel comprehensively and intuitively understand the pipeline status.

[0169] Intelligent Decision Generation and Execution Feedback: The system automatically generates maintenance plans, scheduling plans, and operating parameter adjustment suggestions based on the defect evaluation conclusions and the pipeline operation and maintenance strategy library. For example, for moderately corroded defects, the strategy library recommends that the next maintenance time be within 2 weeks, and the maintenance measure is to replace the locally corroded pipe section. Based on this, a detailed maintenance plan (including construction procedures, required human and material resources, and construction period arrangements), a scheduling plan (maintenance personnel and equipment deployment plan), and operating parameter adjustment suggestions are generated. The generated decision-making plan is verified for safety and feasibility through pipeline system simulation (such as simulating the flow field stability of the pipeline after adjusting the flow rate and pressure based on computational fluid dynamics CFD, and evaluating the structural strength of the pipeline after maintenance based on finite element analysis) to ensure the continuous and stable operation of the pipeline. The decision-making plan is pushed to the operation and maintenance personnel's terminal, and after being reviewed and approved, it is issued for execution. Maintenance personnel carry out maintenance operations according to the plan. During the process, the maintenance progress and on-site situation (text, pictures, videos) are uploaded to the system in real time. The system dynamically adjusts the scheduling plan and resource allocation according to the feedback information to ensure that the maintenance task is completed on time and with high quality. After the task is completed, the system updates the pipeline operation and maintenance file to record the maintenance details, accumulating data and experience for subsequent decision-making analysis.

[0170] Remote monitoring and collaborative operation process: The remote monitoring and collaborative platform realizes real-time data interaction and remote control based on the WebSocket and RESTful API communication architectures. The on-site sensor data is pushed to the platform in real time through the network. Remote users (such as operation and maintenance personnel in the monitoring center, expert teams) can view the pipeline operation status and defect information in real time and receive system alarm notifications (such as pressure overlimit, defect expansion warning). The platform provides a remote operation function. Experts can remotely guide the maintenance operation based on real-time data and video images and adjust the device operation parameters (such as remotely controlling the rotation speed of the power device, starting and stopping the cleaning and filtering device). In the collaborative operation process, when the operation and maintenance personnel issue a maintenance task, the platform automatically matches the maintenance team or expert (according to skills, geographical location, workload) according to the preset rules and pushes the task notification. During the task execution, all parties communicate and collaborate in real time through the instant messaging and video conferencing modules of the platform, share document materials (such as design drawings, operation procedures, maintenance manuals), jointly solve technical problems, achieve efficient collaborative operation among multiple departments, and improve the efficiency and level of offshore pipeline operation and maintenance management.

[0171] A third object of the present invention is to provide an electronic device, which includes a processor, a memory, a communication interface, and a communication bus. The electronic device can be a device with intelligent computing capabilities such as a notebook computer, a stand-alone assistant, or an intelligent tablet; the processor can be a CPU or other final execution units with information processing and program running capabilities; the memory can be a device with storage and program reading capabilities such as a memory, a hard disk, or a storage card.

[0172] In this embodiment, the memory is used to store the program for detecting and evaluating the defects on the inner wall of the offshore pipeline, and the processor is used to execute the program for detecting and evaluating the defects on the inner wall of the offshore pipeline. When the program for detecting and evaluating the defects on the inner wall of the offshore pipeline is executed, the steps described in the method for detecting and evaluating the defects on the inner wall of the offshore pipeline are implemented.

[0173] A fourth object of the present invention is to provide a computer-readable storage medium. When each module in the system for detecting and evaluating the defects on the inner wall of the offshore pipeline of the present invention is implemented in the form of software functions and can be independently sold or used as a product, it can be stored in the computer-readable storage medium. The computer-readable storage medium can be a device such as a USB flash drive, a mobile hard disk, or an optical disc. When the computer executes the program for detecting and evaluating the defects on the inner wall of the offshore pipeline, it can read the program stored in the computer-readable storage medium and execute any one of the steps in the above method for detecting and evaluating the defects on the inner wall of the offshore pipeline through the processor.

[0174] In the present specification, the various embodiments are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. For the same or similar parts among the various embodiments, reference can be made to each other. For the devices disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple. For the relevant parts, reference can be made to the description in the method section.

[0175] The above has introduced in detail a method, system, electronic device and storage medium for detecting and evaluating defects on the inner wall of an ocean pipeline provided by the present invention. Specific examples are used herein to elaborate on the principle and implementation manner of the present invention. The description of the above embodiments is only used to help understand the method and its core idea of the present invention. It should be noted that for those of ordinary skill in the art in the technical field, without departing from the principle of the present invention, several improvements and modifications can be made to the present invention, and these improvements and modifications also fall within the protection scope of the claims of the present invention.

Claims

1. A method for detecting and evaluating defects on the inner wall of a marine pipeline, characterized in that: The method for detecting and evaluating defects on the inner wall of a marine pipeline comprises the following steps: S1. Real-time collection of marine pipeline operation data and marine pipeline inner wall magnetic field data; S2. Preprocess the collected marine pipeline operation data to obtain accurate data; S3. Build and integrate the pipeline entity 3D geometric model, pipeline physical behavior model and pipeline performance evaluation model to obtain a digital twin model and output the pipeline performance evaluation results; S4, using convolutional neural network architecture to build a defect recognition model and output the pipeline inner wall defect assessment results; S5. Build a visualization interface, combine the three-dimensional geometric model of the pipeline entity, the pipeline performance evaluation results and the pipeline inner wall defect evaluation results, and generate a response plan for the pipeline inner wall defects.

2. The method for detecting and evaluating inner wall defects of marine pipelines according to claim 1, characterized in that: The step S1 comprises the following steps: S11. Collecting marine pipeline operation data through pressure sensors, strain sensors, temperature sensors, corrosion monitoring sensors and flow sensors deployed on marine pipelines; S12, collecting magnetic field data of the inner wall of the marine pipeline by an ACFM detection probe deployed in the marine pipeline, wherein the ACFM detection probe comprises: An excitation coil for generating an alternating magnetic field; Magnetic sensor, used to collect magnetic field electrical signals; A conditioning circuit for amplifying and filtering the magnetic field electrical signal; The analog-to-digital conversion circuit is used for converting the magnetic field electrical signal into a magnetic field digital signal.

3. The method for detecting and evaluating inner wall defects of marine pipelines according to claim 1, characterized in that: The step S2 comprises the following steps: S21. Identify and eliminate outliers in the marine pipeline data based on a statistical analysis method to obtain outlier data; S22, filtering the noise-removed data using a Butterworth low-pass filter to obtain noise-reduced data; S23, using a linear normalization method to unify the dimension of the noise reduction data to obtain standardized data; S24. The data fusion algorithm based on Dempster-Shafer evidence theory processes the standardized data to obtain precise data.

4. The method for detecting and evaluating inner wall defects of marine pipelines according to claim 3, characterized in that: In step S24, it is assumed that the identification framework θ includes the normal operation state A, the slight defect state B and the serious defect state C of the pipeline, and the BPA function is defined as follows: Where m1 and m2 are the basic probability distribution functions of the pipeline state provided by the pressure sensor and the strain sensor, respectively; m(A), m(B), m(C), and m(θ) are the basic probability distribution values ​​under different states calculated based on the evidence of the pressure sensor and the strain sensor; X and Y are intermediate variables in calculating the basic probability distribution, which are used to represent the state combinations involved in the evidence of different sensors.

5. The method for detecting and evaluating inner wall defects of a marine pipeline according to claim 1, characterized in that: The step S3 comprises the following steps: S31. Based on the design drawings of the marine pipeline, a three-dimensional geometric model of the pipeline entity is obtained by using three-dimensional software modeling; S32. Based on elastic mechanics theory and finite element method, a pipeline mechanics analysis model is constructed; S33, using empirical formula to calculate pipeline corrosion rate and construct corrosion damage model; S34. Evaluate fatigue damage based on Miner linear cumulative damage theory and construct a material degradation model; S35, collecting historical operation data, fault case records and maintenance records of marine pipelines, and performing data preprocessing to build a pipeline history database; S36, extracting characteristic parameters from the pipeline history database and constructing a characteristic vector set; S37, using a machine learning algorithm to build a pipeline performance evaluation model, and using a feature vector set to train and optimize the pipeline performance evaluation model; S38, integrating a pipeline entity three-dimensional geometric model, a pipeline physical behavior model, and a pipeline performance evaluation model to obtain a digital twin model, wherein the pipeline physical behavior model includes a pipeline mechanical analysis model, a corrosion damage model, and a material degradation model; S39. Input the refined data into the digital twin model and output the pipeline performance evaluation results.

6. The method for detecting and evaluating inner wall defects of a marine pipeline according to claim 1, characterized in that: In step S4, the defect recognition model includes: An input layer is used to receive the magnetic field data of the inner wall of the marine pipeline, wherein the magnetic field data of the inner wall of the marine pipeline is two-dimensional image data; Convolutional layer and pooling layer are used to extract the magnetic field data features of the inner wall of the marine pipeline; The fully connected layer is used to integrate the magnetic field data characteristics of the inner wall of the marine pipeline; The output layer is used to output the pipeline inner wall defect assessment result, and the output pipeline inner wall defect assessment result includes the category, corresponding probability and distribution of the pipeline inner wall defect.

7. The method for detecting and evaluating inner wall defects of a marine pipeline according to claim 1, characterized in that: The step S5 comprises the following steps: S51. Import the digital twin model into the game engine and build a visualization interface; S52, locating the defect position in the visual interface and marking the defect information; S53, constructing a pipeline operation and maintenance strategy library, wherein the pipeline operation and maintenance strategy library includes maintenance and repair strategies and parameter adjustment rules under different defect types, size ranges, and pipeline operation conditions; S54. Filter the matching strategy combination from the pipeline operation and maintenance strategy library through the rule matching algorithm to generate a solution to deal with pipeline inner wall defects.

8. A marine pipeline inner wall defect detection and evaluation system, characterized by: The marine pipeline inner wall defect detection and evaluation system comprises: Data acquisition module, used to collect real-time marine pipeline operation data and marine pipeline inner wall magnetic field data; The data processing and transmission module is used to pre-process the collected marine pipeline operation data, obtain accurate data, and transmit it to the computing platform; The digital twin model construction module is used to integrate and construct the pipeline entity three-dimensional model, pipeline physical behavior model and pipeline performance evaluation model to form a digital twin model and output the pipeline performance evaluation results; Defect intelligent identification and assessment module, used to build a defect identification model and output the pipeline inner wall defect assessment results; The visualization and decision support module is used to visualize the three-dimensional geometric model of the pipeline entity, the pipeline performance evaluation results and the pipeline inner wall defect evaluation results, and generate a response plan for the pipeline inner wall defects.

9. An electronic device, characterized in that: The electronic device comprises a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other via the communication bus; The memory is used to store the marine pipeline inner wall defect detection and evaluation program; The processor is used to execute the marine pipeline inner wall defect detection and assessment program, and when the marine pipeline inner wall defect detection and assessment program is executed, the steps of the marine pipeline inner wall defect detection and assessment method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a marine pipeline inner wall defect detection and assessment program, and when the marine pipeline inner wall defect detection and assessment program is executed by the processor, the steps of the marine pipeline inner wall defect detection and assessment method according to any one of claims 1 to 7 are implemented.

Images