Method, system and equipment for evaluating active corrosion defect of pipeline and storage medium

Abstract

The invention discloses a method, a system and equipment for evaluating active corrosion defects of a pipeline and a storage medium. The method comprises the following steps: acquiring multi-dimensional pipeline characteristic data of a pipeline detected by an inner detector; determining a defect type and an activity degree index corresponding to the pipeline based on the multi-dimensional pipeline features and a pre-constructed two-stage series machine learning model; wherein the two-stage series machine learning model comprises a first-stage convolutional neural network model trained based on a sample defect image and a second-stage extreme gradient boosting model trained based on sample multi-dimensional data; and determining a predicted development rate and a predicted residual life corresponding to the pipeline based on the defect type and the activity degree index, and generating an evaluation result of the active corrosion defect of the pipeline based on the predicted development rate and the predicted residual life. And the evaluation accuracy of the pipeline corrosion defect is improved.

Description

Technical Field

[0001] This invention relates to the field of oil and gas pipeline management technology, and in particular to a method, system, equipment and storage medium for assessing active corrosion defects in pipelines. Background Technology

[0002] Long-distance oil and gas pipelines serve as vital arteries for the nation's energy supply, and their safe operation is of paramount importance. Magnetic flux leakage (MFL) internal testing is currently the most widely used technology for detecting pipeline defects. However, in practical engineering applications, this technology still faces several prominent challenges:

[0003] (1) Difficulty in identifying defect types: MFL signals have similar response characteristics to different types of defects (such as corrosion pits, cracks, and manufacturing defects). It is difficult to achieve accurate and automated classification based solely on signal amplitude and morphology. It heavily relies on the subjective experience of analysts, resulting in low efficiency and poor consistency. (2) Poor comparability of multi-period data: Pipelines are usually inspected every few years. Different types of detectors may be used for different periods of inspection, with varying excitation intensity, sampling frequency, and detection speed, leading to systematic deviations in the signals. Directly comparing multi-period data and identifying defect growth becomes extremely difficult and prone to misjudgment. (3) Lack of defect activity evaluation methods: Traditional analysis mainly focuses on the current size of defects (such as depth and length) and lacks effective methods for evaluating defect "activity" (i.e., whether it is developing or has a tendency to develop). It is impossible to distinguish between high-risk active defects and stable inactive defects, resulting in inaccurate allocation of maintenance resources. (4) Weak prediction and causal analysis: Defect development rate prediction is mostly based on simple linear extrapolation, without considering the complexity of corrosion mechanisms. The prediction results are highly uncertain and cannot support risk-based maintenance decisions. At the same time, defect cause analysis often remains superficial, lacking a systematic, data-driven process to trace the root cause. Summary of the Invention

[0004] This invention provides a method, system, device, and storage medium for assessing active corrosion defects in pipelines, thereby achieving accurate assessment of active corrosion defects in pipelines and solving the technical problems of inaccurate defect type identification, insufficient quantification of activity level, and low reliability of life prediction in traditional assessment methods.

[0005] According to one aspect of the present invention, a method for evaluating active corrosion defects in pipelines is provided, the method comprising:

[0006] Acquire multidimensional pipeline feature data detected by an internal detector;

[0007] Based on the multidimensional pipeline features and a pre-built two-level cascaded machine learning model, the defect type and activity level index corresponding to the pipeline are determined; wherein, the two-level cascaded machine learning model includes a first-level convolutional neural network model trained based on sample defect images and a second-level extreme gradient boosting model trained based on sample multidimensional data;

[0008] Based on the defect type and the activity level index, the predicted development rate and predicted remaining life corresponding to the pipeline are determined, and the evaluation results of the active corrosion defects of the pipeline are generated based on the predicted development rate and predicted remaining life.

[0009] According to another aspect of the present invention, an assessment system for active corrosion defects in pipelines is provided, the system comprising:

[0010] The multidimensional feature acquisition module is used to acquire multidimensional pipeline feature data detected by the internal detector.

[0011] A cascaded model processing module is used to determine the defect type and activity level index corresponding to the pipeline based on the multidimensional pipeline features and a pre-built two-level cascaded machine learning model; wherein, the two-level cascaded machine learning model includes a first-level convolutional neural network model trained based on sample defect images and a second-level extreme gradient boosting model trained based on sample multidimensional data;

[0012] The defect assessment module is used to determine the predicted development rate and predicted remaining life corresponding to the pipeline based on the defect type and the activity level index, and to generate an assessment result of the active corrosion defects of the pipeline based on the predicted development rate and predicted remaining life.

[0013] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0014] At least one processor;

[0015] and memory that is communicatively connected to at least one processor;

[0016] The memory stores a computer program that can be executed by at least one processor, which enables the at least one processor to perform the method for evaluating active corrosion defects in pipelines according to any embodiment of the present invention.

[0017] According to another aspect of the present invention, a computer-readable storage medium is provided, which stores computer instructions for causing a processor to execute and implement a method for evaluating active corrosion defects in pipelines according to any embodiment of the present invention.

[0018] The technical solution of this invention acquires multi-dimensional pipeline feature data detected by an internal detector, and uses a two-level cascaded machine learning model consisting of a first-level convolutional neural network model trained based on sample defect images and a second-level extreme gradient boosting model trained based on sample multi-dimensional data to determine the defect type and activity level index. This, in turn, determines the predicted development rate and remaining lifespan and generates an evaluation result, achieving accurate assessment of active corrosion defects in pipelines. This solves the technical problems of inaccurate defect type identification, insufficient quantification of activity level, and low reliability of lifespan prediction in traditional assessment methods, thus improving the accuracy and scientific rigor of pipeline corrosion defect assessment and providing effective decision support for pipeline integrity management.

[0019] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 A flowchart illustrating a method for assessing active corrosion defects in pipelines, provided as an embodiment of the present invention;

[0022] Figure 2a A flowchart of another method for evaluating active corrosion defects in pipelines provided in an embodiment of the present invention;

[0023] Figure 2b A schematic diagram of signal normalization and DTW registration for an optional example of a method for evaluating active corrosion defects in pipelines provided in an embodiment of the present invention;

[0024] Figure 2c A schematic diagram illustrating multi-dimensional feature extraction of an optional example of a method for assessing active corrosion defects in pipelines provided by an embodiment of the present invention;

[0025] Figure 2d A schematic diagram of a two-layer machine learning model structure, which is an optional example of an evaluation method for active corrosion defects in pipelines provided in an embodiment of the present invention.

[0026] Figure 2e A predictive model library and Monte Carlo simulation calculation flowchart for an optional example of a pipeline active corrosion defect assessment method provided in this embodiment of the invention;

[0027] Figure 2f A flowchart of a Bayesian network-based causal diagnosis model, which is an optional example of an assessment method for active corrosion defects in pipelines provided by an embodiment of the present invention.

[0028] Figure 3 A schematic diagram of the structure of an assessment system for active corrosion defects in pipelines provided in an embodiment of the present invention;

[0029] Figure 4 A schematic diagram of the structure of an electronic device for implementing a method for evaluating active corrosion defects in pipelines according to an embodiment of the present invention. Detailed Implementation

[0030] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0031] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0032] Figure 1 This is a flowchart illustrating a method for assessing active corrosion defects in pipelines, provided by an embodiment of the present invention. This embodiment is applicable to the assessment of active corrosion defects in pipelines. The method can be executed by an assessment system for active corrosion defects in pipelines. This system can be implemented in hardware and / or software and can be configured in an electronic device. Figure 1 As shown, the method specifically includes the following steps:

[0033] S110. Obtain multi-dimensional pipeline feature data detected by the internal detector.

[0034] Internal detectors can be understood as detection devices that acquire information about pipeline defects. Multidimensional pipeline feature data can be understood as a multidimensional data set used to describe the pipeline's own attributes and defect status.

[0035] Specifically, an internal detector is used to inspect the target pipeline, and multi-dimensional pipeline feature data covering pipeline attributes and defect status is collected and processed.

[0036] Optionally, the multidimensional pipeline feature data includes pipeline geometric features, pipeline defect features, upstream and downstream features of defects, and temporal features; the pipeline geometric features include spiral weld reinforcement height, spiral weld spacing, and pipe section length; the pipeline defect features include defect amplitude, length, width, volume, and signal morphology indicators skewness and kurtosis; the upstream and downstream features of defects include the relative position of the defect and the weld, the surrounding signal background value, and the signal-to-noise ratio; the temporal features include the defect amplitude change rate, size expansion rate, and morphological change indicators.

[0037] Among these, pipeline geometric features can be understood as characteristic parameters characterizing the pipeline's manufacturing process and structural properties. Spiral weld reinforcement height can be understood as the height by which the weld metal extends beyond the surface of the pipeline base material after spiral weld welding. Spiral weld spacing can be understood as the axial distance between two adjacent spiral welds on the pipeline. Pipe section length can be understood as the axial length of a single pipeline segment. Pipeline defect features can be understood as characteristic parameters directly describing the shape and scale of the defect itself. Defect amplitude can be understood as the signal peak value corresponding to the defect region in the magnetic flux leakage detection signal. Defect length / width can be understood as the dimensions of the defect in the pipeline's axial (length) and circumferential (width) directions. Defect volume can be understood as the estimated total volume of defect metal loss obtained by dividing the defect region into grids along the axial and circumferential directions, and then performing a double integral on the signal amplitude (corresponding to the metal loss depth) at each grid point. Signal morphology indicators skewness can be understood as a parameter describing the degree of asymmetry in the defect signal distribution, and kurtosis can be understood as a parameter describing the sharpness of the signal peak.

[0038] The upstream and downstream characteristics of a defect can be understood as characteristic parameters that characterize the surrounding environment and signal background of the defect.

[0039] The relative position of the defect and the weld can be understood as the axial distance from the defect center to the nearest weld. The surrounding background signal value can be understood as the average flux leakage signal of the normal pipe section surrounding the defect area. The signal-to-noise ratio (SNR) can be understood as the ratio of the defect signal amplitude to the background noise amplitude. Temporal characteristics can be understood as characteristic parameters characterizing the historical change trend of the defect based on multi-period detection data. The defect amplitude change rate can be understood as the ratio of the change in defect amplitude between two adjacent detection periods to the detection time interval. The size expansion rate can be understood as the ratio of the change in defect length / width between two adjacent detection periods to the detection time interval. The morphological change index can be understood as parameters obtained by calculating cross-correlation coefficients and other methods based on defect signals from two adjacent detection periods.

[0040] Optionally, before acquiring the multidimensional pipeline feature data detected by the internal detector, the method further includes: normalizing the original magnetic flux leakage signal obtained by at least one type of internal detector to obtain a standardized magnetic flux leakage signal; and performing high-precision spatiotemporal registration on the standardized magnetic flux leakage signal to ensure the alignment of defect locations and signal features of multi-period detection data.

[0041] The original magnetic flux leakage signal can be understood as the magnetic flux leakage detection signal directly acquired by the internal detector.

[0042] Normalization can be understood as the process of eliminating systematic biases in signals from different types of detectors and different detection periods. Standardized magnetic flux leakage signal can be understood as the magnetic flux leakage signal after normalization. High-precision spatiotemporal registration can be understood as the process of aligning the time and spatial dimensions of multi-period standardized magnetic flux leakage signals based on signal feature points. Multi-period detection data can be understood as the detection data obtained from internal detection of the same pipeline at different times.

[0043] Specifically, for the raw magnetic flux leakage signal obtained by at least one type of detector, a normalization formula is used to eliminate the systematic bias caused by the difference in detector parameters, resulting in a standardized magnetic flux leakage signal. For the obtained multi-period standardized magnetic flux leakage signal, algorithms such as dynamic time warping are used to extract key feature points (such as peak points and zero crossings) of the signal, and the optimal warping path between the multi-period signals is calculated. Based on this path, the signal spatial coordinates are re-interpolated to achieve the alignment of defect locations and signal features of the multi-period data.

[0044] Optionally, the raw magnetic flux leakage signal obtained by at least one type of internal detector is normalized, as expressed by the following formula:

[0045] ;

[0046] in, The normalized leakage magnetic field signal is obtained after normalization processing. For the first The original magnetic flux leakage signal obtained from the internal detection of magnetic flux leakage in the secondary pipeline. and These are the mean and standard deviation of the background segment in the original magnetic flux leakage signal, calculated using a sliding window. The preset reference excitation intensity, For the first The current excitation intensity of the internal detector used in this test. For a pre-set reference sensor, For the first The current gain coefficient of the internal detector sensor used in this test.

[0047] S120. Based on the multidimensional pipeline features and the pre-built two-level cascaded machine learning model, determine the defect type and activity level index corresponding to the pipeline; wherein, the two-level cascaded machine learning model includes a first-level convolutional neural network model trained based on sample defect images and a second-level extreme gradient boosting model trained based on sample multidimensional data.

[0048] The two-level cascaded machine learning model can be understood as a model system consisting of two machine learning models with different functions working in sequence. The convolutional neural network model (level 1) can be understood as a deep learning model adept at processing image-like data. The extreme gradient boosting model (level 2) can be understood as a gradient boosting decision tree model based on ensemble learning. Sample defect images can be understood as signal images extracted from historical pipeline magnetic flux leakage detection data and verified through excavation to confirm defect types. These images are generated by converting the original magnetic flux leakage signal (one-dimensional time-series signal) into a two-dimensional pseudo-color or grayscale image, containing visual features of different defect types (such as external corrosion, pitting, and cracks). Sample multi-dimensional data can be understood as structured data compiled from historical pipeline detection data and associated with the actual state of defects (such as activity level and development trend). Each data sample is labeled with a corresponding defect activity tag (such as high activity, low activity), used to train the level 2 extreme gradient boosting model to quantitatively assess the degree of defect activity. Defect types can be understood as specific classifications of pipeline defects, including but not limited to external corrosion, internal corrosion, pitting, axial cracks, circumferential cracks, and manufacturing defects. Activity level indicators can be understood as parameters that quantitatively reflect the development trend of defects.

[0049] Specifically, the acquired multidimensional pipeline feature data is input into a pre-constructed two-level cascaded machine learning model; the first-level convolutional neural network model analyzes the image features in the data and outputs the defect type determination result; the second-level extreme gradient boosting model outputs the activity index of the defect based on the multidimensional feature data, realizing a dual analysis of the defect.

[0050] Optionally, determining the defect type and activity level index corresponding to the pipeline based on the multidimensional pipeline features and a pre-built two-level cascaded machine learning model includes:

[0051] The signal image features in the multidimensional pipeline features are input into the first-level convolutional neural network model, and the defect type corresponding to the pipeline is determined based on the output of the first model.

[0052] The feature vector of the multidimensional pipeline features is input into the second-level extreme gradient boosting model to evaluate the degree of defect activity. Based on the output of the second model, the activity index corresponding to the pipeline is determined.

[0053] In this context, signal image features can be understood as the spatial features extracted from a two-dimensional image after converting the raw magnetic flux leakage signal acquired by the internal detector into such an image. Feature vectors can be understood as high-dimensional numerical vectors formed by structuring multi-dimensional pipeline feature data. The output of the first model can be understood as the classification result of the defect type by the first-level convolutional neural network model. The output of the second model can be understood as the quantification output of the defect activity level by the second-level extreme gradient boosting model.

[0054] Specifically, signal image features are extracted from the multidimensional pipeline feature data and input into a pre-trained first-level convolutional neural network model. The model learns and matches image features to output a probability distribution of defect types, and determines the specific type of pipeline defect based on this distribution. The multidimensional pipeline feature data is then organized into structured feature vectors and input into a pre-trained second-level extreme gradient boosting model. The model performs fitting analysis on the feature vectors and outputs a defect activity index, thus completing a quantitative assessment of defect activity.

[0055] S130. Based on the defect type and the activity level index, determine the predicted development rate and predicted remaining life corresponding to the pipeline, and generate an evaluation result of the active corrosion defects of the pipeline based on the predicted development rate and predicted remaining life.

[0056] The predicted development rate can be understood as the rate of change of defect size (such as length, depth, and volume) over time, based on the current state and historical patterns of the defect, reflecting the future expansion speed of the defect. The predicted remaining life can be understood as the predicted development rate of the defect and a pre-defined failure criterion. The assessment result can be understood as a report on active corrosion defects in the pipeline.

[0057] Specifically, based on the obtained defect type and activity level indicators, an appropriate prediction method is selected to calculate the predicted development rate of the defects; combined with the preset failure criteria, the predicted remaining life is calculated; finally, the above information is integrated to generate the assessment results of pipeline active corrosion defects.

[0058] Optionally, determining the predicted development rate and predicted remaining lifetime corresponding to the pipeline based on the defect type and the activity level index includes:

[0059] Based on the defect type and the activity level index, a prediction model is adaptively selected from a pre-built model library, wherein the prediction model includes a linear growth model, an exponential growth model, a process probability model, and a neural network model.

[0060] The change of defect size over time is simulated and calculated based on the prediction model, pipeline defect characteristics, and the activity level index. The predicted development rate is determined based on the size change per unit time output by the prediction model.

[0061] Based on the predicted development rate and the preset failure threshold, the probability distribution of the defect size over time output by the prediction model is analyzed to determine the statistical value of the time distribution when the defect size reaches the preset failure threshold, and the predicted remaining lifetime is determined based on the statistical value of the time distribution.

[0062] The model library can be understood as a pre-built collection of prediction models applicable to different corrosion types and development patterns. A linear growth model is a prediction model that assumes defect size changes linearly over time, suitable for scenarios with stable defect development rates, such as uniform corrosion. An exponential growth model is a prediction model that assumes defect size changes exponentially over time, suitable for scenarios with accelerating defect development rates, such as localized pitting corrosion. The Wiener process probabilistic model is a probabilistic prediction model based on stochastic process theory, suitable for scenarios where the corrosion process exhibits random fluctuations. An LSTM neural network is a long short-term memory neural network, suitable for scenarios with sufficient multi-period detection data and complex defect development patterns. A preset failure threshold can be understood as a pre-set critical size for defect failure based on industry standards (such as ASME B31G) or pipeline operation requirements.

[0063] The probability distribution of defect size over time can be understood as the probability distribution of defect size at different time points calculated by a prediction model. The time distribution statistics can be understood as the characteristic values ​​obtained by performing probability statistics on the time when the defect size reaches a preset failure threshold, including but not limited to the expected value, median, and 90% confidence interval.

[0064] Specifically, based on the determined defect type and activity level index, a suitable prediction model is selected from a pre-built model library. For example, a linear growth model is selected for uniform corrosion, and an exponential growth model is selected for pitting corrosion. The selected prediction model, pipeline defect characteristics, and activity level index are used as inputs. The model simulates the dynamic change of defect size over time, outputting the change in defect size per unit time, i.e., the predicted development rate. Based on the predicted development rate and a preset failure threshold, the probability distribution of the defect size output by the prediction model over time is statistically analyzed to determine the time distribution at which the defect size reaches the failure threshold. Based on the time distribution, the statistical values ​​of the time distribution are extracted as the final predicted remaining lifetime.

[0065] The technical solution of this invention acquires multi-dimensional pipeline feature data detected by an internal detector, and uses a two-level cascaded machine learning model consisting of a first-level convolutional neural network model trained based on sample defect images and a second-level extreme gradient boosting model trained based on sample multi-dimensional data to determine the defect type and activity level index. This, in turn, determines the predicted development rate and remaining lifespan and generates an evaluation result, achieving accurate assessment of active corrosion defects in pipelines. This solves the technical problems of inaccurate defect type identification, insufficient quantification of activity level, and low reliability of lifespan prediction in traditional assessment methods, thus improving the accuracy and scientific rigor of pipeline corrosion defect assessment and providing effective decision support for pipeline integrity management.

[0066] Figure 2a This is a flowchart of another method for assessing active corrosion defects in pipelines provided by an embodiment of the present invention. Based on the above embodiments, this embodiment is a further optimization. Specific implementation details can be found in the technical solution of this embodiment. Technical terms that are the same as or corresponding to those in the above embodiments will not be repeated here. Optionally, after generating the assessment results of the active corrosion defects in the pipeline, the method further includes: diagnosing the cause of the defects by fusing multi-source data through a Bayesian network; wherein the multi-source data includes soil resistivity, cathodic protection potential, and activity level indicators; and determining the defect repair measures corresponding to the pipeline based on the cause diagnosis results and a pre-built pipeline repair measure library.

[0067] like Figure 2a As shown, the method specifically includes the following steps:

[0068] S210. Obtain multi-dimensional pipeline feature data detected by the internal detector.

[0069] S220. Based on the multidimensional pipeline features and a pre-built two-level cascaded machine learning model, determine the defect type and activity level index corresponding to the pipeline; wherein, the two-level cascaded machine learning model includes a first-level convolutional neural network model trained based on sample defect images and a second-level extreme gradient boosting model trained based on sample multidimensional data.

[0070] S230. Based on the defect type and the activity level index, determine the predicted development rate and predicted remaining life corresponding to the pipeline, and generate an evaluation result of the active corrosion defects of the pipeline based on the predicted development rate and predicted remaining life.

[0071] S240. Diagnose the causes of defects by fusing multi-source data through a Bayesian network; wherein the multi-source data includes soil resistivity, cathodic protection potential and activity level indicators.

[0072] Here, Bayesian networks can be understood as a reasoning tool based on probabilistic graphical models. Multi-source data can be understood as external data related to pipeline corrosion, in addition to internal detection data. Soil resistivity can be understood as the resistivity of the soil in the pipeline burial area. Cathodic protection potential can be understood as the potential value applied to the pipeline by the cathodic protection system.

[0073] Specifically, a Bayesian network is constructed that includes observation nodes (soil resistivity, cathodic protection potential, and activity level indicators) and hidden nodes (cathodic protection effectiveness, soil corrosivity, and anti-corrosion layer condition). Multi-source data of the target pipe section are input into the network, and the posterior probability of each potential cause is calculated through Bayesian probabilistic inference to obtain the cause diagnosis results.

[0074] S250. Based on the cause diagnosis results and the pre-built pipeline maintenance measure library, determine the defect maintenance measures corresponding to the pipeline.

[0075] The causal diagnosis results can be understood as the posterior probability distribution of each potential defect cause obtained through Bayesian network inference. Defect repair measures can be understood as targeted technical solutions used to address the root causes of defects.

[0076] Specifically, based on the obtained cause diagnosis results, the pre-built pipeline maintenance measures library is queried, the maintenance plan corresponding to the cause is matched, and the final defect maintenance measures are determined.

[0077] The technical solution of this invention achieves probabilistic diagnosis of defect causes by fusing multi-source data through Bayesian networks, and automatically matches targeted maintenance solutions with a pipeline maintenance measures library. This solves the problems of traditional defect cause analysis relying on human experience and lack of data support for maintenance measures. It provides data-driven decision support for accurate tracing and efficient treatment of pipeline active corrosion defects, and improves the scientific nature of pipeline preventive maintenance.

[0078] As an optional example of Embodiment 1 of the present invention, the method for evaluating pipeline active corrosion defects in this embodiment specifically includes the following steps:

[0079] Step 1. Normalization and high-precision registration of internal leakage magnetic field detection signal

[0080] First, a database covering calibration parameters (such as excitation intensity, sampling interval, number of channels, and nominal speed) for different types of internal detectors is established. Then, the raw MFL signal undergoes refined processing. Figure 2b A schematic diagram illustrating signal normalization and DTW registration is provided as an optional example of a method for assessing active corrosion defects in pipelines. (See diagram for example.) Figure 2b As shown, amplitude normalization:

[0081] ;

[0082] in, The normalized leakage magnetic field signal is obtained after normalization processing. For the first The original magnetic flux leakage signal obtained from the internal detection of magnetic flux leakage in the secondary pipeline. and These are the mean and standard deviation of the background segment in the original magnetic flux leakage signal, calculated using a sliding window. The preset reference excitation intensity, For the first The current excitation intensity of the internal detector used in this test. For a pre-set reference sensor, For the first The current gain coefficient of the internal detector sensor used in this test.

[0083] Spatiotemporal registration employs the Dynamic Time Warping (DTW) algorithm to achieve high-precision alignment of defects in multi-period data. This algorithm extracts significant feature points such as peak points and zero-crossing points of the signal, calculates the optimal warping path between two periods of signal, and re-interpolates the spatial coordinates of the signal, thereby compensating for spatial scale distortion caused by fluctuations in detection speed.

[0084] Step 2. Multi-dimensional feature extraction

[0085] Figure 2c A schematic diagram illustrating multi-dimensional feature extraction is provided as an optional example of a method for assessing active corrosion defects in pipelines. For example... Figure 2c As shown, four main categories of feature parameters are automatically extracted from the registered signal to construct a high-dimensional feature vector:

[0086] Geometric features include the reinforcement height of spiral welds, the spacing between spiral welds, and the length of pipe sections. These features reflect the manufacturing process characteristics of the pipeline and are closely related to the mechanical environment in which defects occur.

[0087] Defect characteristics include defect amplitude, defect length, defect width, defect volume (the defect region is divided into tiny grids along the axial and circumferential directions, and the signal amplitude (representing the metal loss depth) of each grid point is double-integrated to estimate the total volume and signal morphology indicators (such as skewness and kurtosis).

[0088] Upstream and downstream characteristics: including the relative position of defects and welds, surrounding signal background values, and signal-to-noise ratio.

[0089] Temporal characteristics (for defects with multiple detection data): including amplitude change rate, size expansion rate, and morphological change indicators (such as cross-correlation coefficients).

[0090] Step 3. Intelligent Defect Identification and Activity Evaluation

[0091] Figure 2d A schematic diagram of a two-layer machine learning model structure is provided as an optional example of a method for assessing active corrosion defects in pipelines. (See diagram for example.) Figure 2d As shown, this stage employs a two-stage cascaded machine learning model:

[0092] Level 1: Defect Type Classification. A Level 1 convolutional neural network model, such as a deep convolutional neural network (CNN), trained on sample defect images is used to classify the signal images of defects (such as pseudo-color images and grayscale images). The network input is a 128x128 image, which is processed through 5 convolutional blocks (each block contains Conv2D, BatchNorm, and MaxPooling) for feature extraction. Finally, a fully connected layer and a Softmax layer output the probability distribution of 6 types of defects (such as external corrosion, internal corrosion, pitting, axial cracks, circumferential cracks, and manufacturing defects). This model is trained using a large number of excavated and verified defect samples, and data augmentation (such as adding noise or slight stretching) can be used to expand the dataset when necessary.

[0093] Level Two: Activity Level Assessment. The extracted multi-dimensional features are input into a Level Two extreme gradient boosting model, such as the XGBoost ensemble learning model, trained based on multi-dimensional sample data. This model outputs a continuous activity score (ActivityScore) between 0 and 1. Based on engineering experience, thresholds are set to classify defects into four risk levels: High Risk (>0.8): Rapid development, requiring immediate attention; Medium Risk (0.6-0.8): Significant development, requiring planned attention within 6 months; Low Risk (0.4-0.6): Slow development, included in the annual monitoring plan; Harmless (<0.4): Minimal development, routine inspections are sufficient. This provides a direct and quantitative basis for prioritizing maintenance.

[0094] Step 4. Development Rate Prediction and Remaining Life Assessment

[0095] Figure 2e A library of predictive models and a flowchart of Monte Carlo simulation calculations for remaining life are provided as optional examples of methods for assessing active corrosion defects in pipelines. Figure 2e As shown, a hybrid modeling approach is adopted, providing a model library containing multiple models to adapt to different corrosion types and development patterns:

[0096] Model library: includes linear growth models ( Suitable for uniform corrosion), exponential growth model ( (Applicable to local pitting corrosion), Wiener process probability model ( , describing random erosion) and LSTM neural networks (suitable for defects with sufficient data and complex nonlinear development).

[0097] Remaining lifetime calculation: Based on failure criteria such as ASME B31G, the limit state function g(d, L, P, ...) is defined. Monte Carlo simulation is used to perform tens of thousands of random samplings on model parameters and future conditions to extrapolate the probability distribution of future defect sizes, and then calculate the probability distribution of the remaining lifetime (RUL). The final results are output in the form of expected value, median, and 90% confidence interval, quantifying the uncertainty of the prediction.

[0098] Step 5. Causal Diagnosis and Decision Support

[0099] Figure 2f A flowchart of a Bayesian network-based causal diagnostic model is provided as an optional example of an assessment method for active corrosion defects in pipelines. (See attached flowchart.) Figure 2f As shown, a Bayesian network model is constructed for multi-source data fusion and causal reasoning.

[0100] Network construction: Network nodes include observed nodes (such as soil resistivity, cathodic protection potential, and activity score) and hidden nodes (such as "cathodic protection effectiveness" and "soil corrosivity"). Edges represent causal relationships between nodes.

[0101] Inference and diagnosis: When new observation data is input (such as a section of pipeline with dense active defects, low soil resistivity, and substandard cathodic protection potential), the network performs probabilistic inference to calculate the posterior probability of each potential cause (such as "CP ineffective" and "soil corrosion"), thereby determining the most likely root cause and its confidence level.

[0102] Recommended measures: Based on the cause diagnosis results, the system automatically recommends targeted measures (such as "inspect the cathodic protection system" or "excavate and repair the anti-corrosion layer").

[0103] For example, taking a 100-kilometer-long crude oil pipeline as the evaluation object, we collected its internal magnetic flux leakage detection data in 2018 and 2021.

[0104] Data preprocessing: The data is imported into the system, and the signal preprocessing module automatically calls the detector parameters in the database, executes the amplitude normalization formula and DTW algorithm, and generates a high-precision registered standard dataset.

[0105] Feature extraction and intelligent evaluation: The system automatically traverses all defects and calculates their 13 multi-dimensional feature vectors. Subsequently, the pre-trained CNN model determines that one of the defects is "external corrosion" (probability 92%); the XGBoost model calculates an activity score of 0.76 based on its features, and the system automatically marks it as "medium risk".

[0106] Lifetime prediction: The system call exponential model and Wiener process predict the defect. After 10,000 Monte Carlo simulations, the 90% confidence interval for its remaining lifetime is [3.2 years, 5.1 years].

[0107] Cause diagnosis: The system automatically retrieves soil data (resistivity = 10 Ω·m) and cathodic protection data (potential fluctuation between -0.7V and -1.1V CSE) for this pipe section. After Bayesian network calculation, the diagnostic results are output: the main cause is "insufficient cathodic protection effectiveness" (probability 68%), followed by "soil corrosivity" (probability 25%).

[0108] Decision output: The system automatically generates a report in the visualization platform and highlights the recommended measures: "Prioritize the investigation and commissioning of the cathodic protection system of this pipeline section."

[0109] The technical solution of this invention, through high-precision normalized registration and multi-dimensional feature fusion combined with a two-layer machine learning model, greatly improves the accuracy of defect identification, classification, and activity evaluation. Employing a hybrid model library and Monte Carlo probabilistic prediction, it provides prediction results that better reflect actual corrosion patterns and quantifies uncertainty, making risk management decisions more scientific. By fusing multi-source data through Bayesian networks, it achieves systematic, data-driven diagnosis of the root causes of defects, changing the one-sided analysis mode that relies on human experience. It automates the entire process from raw data to cause reports, significantly reducing reliance on the experience of analysts and improving work efficiency and consistency.

[0110] Figure 3 This is a schematic diagram of a pipeline active corrosion defect assessment system provided in an embodiment of the present invention. Figure 3 As shown, the system includes: a multi-dimensional feature acquisition module 310, a serial model processing module 320, and a defect assessment module 330.

[0111] The system includes a multi-dimensional feature acquisition module 310, which acquires multi-dimensional pipeline feature data detected by an internal detector; a cascaded model processing module 320, which determines the defect type and activity level index corresponding to the pipeline based on the multi-dimensional pipeline features and a pre-built two-level cascaded machine learning model; wherein the two-level cascaded machine learning model includes a first-level convolutional neural network model trained based on sample defect images and a second-level extreme gradient boosting model trained based on sample multi-dimensional data; and a defect evaluation module 330, which determines the predicted development rate and predicted remaining life corresponding to the pipeline based on the defect type and the activity level index, and generates an evaluation result of the active corrosion defect of the pipeline based on the predicted development rate and predicted remaining life.

[0112] The technical solution of this invention acquires multi-dimensional pipeline feature data detected by an internal detector, and uses a two-level cascaded machine learning model consisting of a first-level convolutional neural network model trained based on sample defect images and a second-level extreme gradient boosting model trained based on sample multi-dimensional data to determine the defect type and activity level index. This, in turn, determines the predicted development rate and remaining lifespan and generates an evaluation result, achieving accurate assessment of active corrosion defects in pipelines. This solves the technical problems of inaccurate defect type identification, insufficient quantification of activity level, and low reliability of lifespan prediction in traditional assessment methods, thus improving the accuracy and scientific rigor of pipeline corrosion defect assessment and providing effective decision support for pipeline integrity management.

[0113] Optionally, the concatenated model processing module includes:

[0114] The first processing unit is used to input the signal image features in the multi-dimensional pipeline features into the first-level convolutional neural network model, and determine the defect type corresponding to the pipeline based on the output result of the first model.

[0115] The second processing unit is used to input the feature vector of the multi-dimensional pipeline features into the second-level extreme gradient boosting model to evaluate the degree of defect activity, and determine the activity index corresponding to the pipeline based on the output result of the second model.

[0116] Optionally, the multidimensional pipeline feature data includes pipeline geometric features, pipeline defect features, upstream and downstream features of defects, and temporal features; the pipeline geometric features include spiral weld reinforcement height, spiral weld spacing, and pipe section length; the pipeline defect features include defect amplitude, length, width, volume, and signal morphology indicators skewness and kurtosis; the upstream and downstream features of defects include the relative position of the defect and the weld, the surrounding signal background value, and the signal-to-noise ratio; the temporal features include the defect amplitude change rate, size expansion rate, and morphological change indicators.

[0117] Optionally, the defect assessment module includes:

[0118] The prediction model selection unit is used to adaptively select a prediction model from a pre-built model library based on the defect type and the activity level index, wherein the prediction model includes a linear growth model, an exponential growth model, a process probability model, and a neural network model.

[0119] The development rate prediction unit is used to simulate and calculate the change of defect size over time based on the prediction model, pipeline defect characteristics and the activity level index, and to determine the predicted development rate based on the size change per unit time output by the prediction model.

[0120] The remaining lifetime prediction unit is used to analyze the probability distribution of the defect size over time output by the prediction model based on the predicted development rate and the preset failure threshold, determine the time distribution statistics of the defect size reaching the preset failure threshold, and determine the predicted remaining lifetime based on the time distribution statistics.

[0121] Optionally, the device further includes:

[0122] The cause diagnosis module is used to diagnose the cause of defects by fusing multi-source data through a Bayesian network after generating the assessment results of active corrosion defects in the pipeline; wherein, the multi-source data includes soil resistivity, cathodic protection potential and activity level indicators.

[0123] The maintenance measures determination module is used to determine the defect maintenance measures corresponding to the pipeline based on the cause diagnosis results and the pre-built pipeline maintenance measures library.

[0124] Optionally, the device further includes:

[0125] The normalization processing module is used to normalize the original magnetic flux leakage signal obtained by at least one type of internal detector before acquiring the multi-dimensional pipeline feature data of the pipeline detected by the internal detector, so as to obtain a standardized magnetic flux leakage signal.

[0126] A special alignment module is used to perform high-precision spatiotemporal registration of the standardized magnetic flux leakage signal to ensure the alignment of defect locations and signal features in multi-period detection data.

[0127] Optionally, the normalization processing module is specifically used for:

[0128] The raw magnetic flux leakage signal obtained by at least one type of internal detector is normalized, and the result is expressed by the following formula:

[0129] ;

[0130] in, The normalized leakage magnetic field signal is obtained after normalization processing. For the first The original magnetic flux leakage signal obtained from the internal detection of magnetic flux leakage in the secondary pipeline. and These are the mean and standard deviation of the background segment in the original magnetic flux leakage signal, calculated using a sliding window. The preset reference excitation intensity, For the first The current excitation intensity of the internal detector used in this test. For a pre-set reference sensor, For the first The current gain coefficient of the internal detector sensor used in this test.

[0131] The pipeline active corrosion defect assessment system provided in this embodiment of the invention can execute the pipeline active corrosion defect assessment method provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method.

[0132] Figure 4This is a schematic diagram of an electronic device used to implement the pipeline active corrosion defect assessment method according to embodiments of the present invention. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0133] like Figure 4 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 can also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0134] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0135] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as the evaluation of active corrosion defects in pipelines.

[0136] In some embodiments, the assessment of active corrosion defects in the method piping can be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or mounted on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the assessment of active corrosion defects in the method piping described above can be performed. Alternatively, in other embodiments, processor 11 can be configured to perform the assessment of active corrosion defects in the method piping by any other suitable means (e.g., by means of firmware).

[0137] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations may include: implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0138] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0139] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0140] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0141] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or middleware components (e.g., application servers), or frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0142] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0143] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0144] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for assessing active corrosion defects in pipelines, characterized in that, include: Acquire multidimensional pipeline feature data detected by an internal detector; Based on the multidimensional pipeline features and a pre-built two-level cascaded machine learning model, the defect type and activity level index corresponding to the pipeline are determined; wherein, the two-level cascaded machine learning model includes a first-level convolutional neural network model trained based on sample defect images and a second-level extreme gradient boosting model trained based on sample multidimensional data; Based on the defect type and the activity level index, the predicted development rate and predicted remaining life corresponding to the pipeline are determined, and the evaluation results of the active corrosion defects of the pipeline are generated based on the predicted development rate and predicted remaining life.

2. The method according to claim 1, characterized in that, The method for determining the defect type and activity level index corresponding to the pipeline based on the multidimensional pipeline features and a pre-built two-level cascaded machine learning model includes: The signal image features in the multidimensional pipeline features are input into the first-level convolutional neural network model, and the defect type corresponding to the pipeline is determined based on the output of the first model. The feature vector of the multidimensional pipeline features is input into the second-level extreme gradient boosting model to evaluate the degree of defect activity. Based on the output of the second model, the activity index corresponding to the pipeline is determined.

3. The method according to claim 1, characterized in that, The multidimensional pipeline feature data includes pipeline geometric features, pipeline defect features, upstream and downstream features of defects, and temporal features; the pipeline geometric features include spiral weld reinforcement height, spiral weld spacing, and pipe section length; the pipeline defect features include defect amplitude, length, width, volume, and signal morphology indicators skewness and kurtosis; the upstream and downstream features of defects include the relative position of the defect and the weld, the surrounding signal background value, and the signal-to-noise ratio; the temporal features include the defect amplitude change rate, size expansion rate, and morphological change indicators.

4. The method according to claim 1, characterized in that, The determination of the predicted development rate and predicted remaining lifetime corresponding to the pipeline based on the defect type and the activity level index includes: Based on the defect type and the activity level index, a prediction model is adaptively selected from a pre-built model library, wherein the prediction model includes a linear growth model, an exponential growth model, a process probability model, and a neural network model. The change of defect size over time is simulated and calculated based on the prediction model, pipeline defect characteristics, and the activity level index. The predicted development rate is determined based on the size change per unit time output by the prediction model. Based on the predicted development rate and the preset failure threshold, the probability distribution of the defect size over time output by the prediction model is analyzed to determine the statistical value of the time distribution when the defect size reaches the preset failure threshold, and the predicted remaining lifetime is determined based on the statistical value of the time distribution.

5. The method according to claim 1, characterized in that, Following the assessment results of the active corrosion defects in the generated pipeline, the following is also included: The causes of defects are diagnosed by fusing multi-source data through Bayesian networks; wherein the multi-source data includes soil resistivity, cathodic protection potential and activity level indicators. Based on the causal diagnosis results and the pre-built pipeline maintenance measure library, the corresponding defect maintenance measures are determined for the pipeline.

6. The method according to claim 1, characterized in that, Before acquiring multidimensional pipeline feature data detected by the internal detector, the following steps are also included: The raw magnetic flux leakage signal obtained by detecting at least one type of internal detector is normalized to obtain a standardized magnetic flux leakage signal. High-precision spatiotemporal registration is performed on the standardized magnetic flux leakage signal to ensure the alignment of defect locations and signal characteristics in multi-period detection data.

7. The method according to claim 6, characterized in that, The normalization process of the raw magnetic flux leakage signal obtained by at least one type of internal detector is expressed by the following formula: ； in, The normalized leakage magnetic field signal is obtained after normalization processing. For the first The original magnetic flux leakage signal obtained from the internal detection of magnetic flux leakage in the secondary pipeline. and These are the mean and standard deviation of the background segment in the original magnetic flux leakage signal, calculated using a sliding window. The preset reference excitation intensity, For the first The current excitation intensity of the internal detector used in this test. For a pre-set reference sensor, For the first The current gain coefficient of the internal detector sensor used in this test.

8. A system for assessing active corrosion defects in pipelines, characterized in that, include: The multidimensional feature acquisition module is used to acquire multidimensional pipeline feature data detected by the internal detector. A cascaded model processing module is used to determine the defect type and activity level index corresponding to the pipeline based on the multidimensional pipeline features and a pre-built two-level cascaded machine learning model; wherein, the two-level cascaded machine learning model includes a first-level convolutional neural network model trained based on sample defect images and a second-level extreme gradient boosting model trained based on sample multidimensional data; The defect assessment module is used to determine the predicted development rate and predicted remaining life corresponding to the pipeline based on the defect type and the activity level index, and to generate an assessment result of the active corrosion defects of the pipeline based on the predicted development rate and predicted remaining life.

9. An electronic device, characterized in that, The electronic device includes: At least one processor; and a memory communicatively connected to the at least one processor; The memory stores a computer program that can be executed by the at least one processor, which enables the at least one processor to perform the assessment method for active corrosion defects in pipelines according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that, when executed by a processor, implement the method for evaluating active corrosion defects in pipelines as described in any one of claims 1-7.

Images