Bending deformation pipe section identification method and device, electronic equipment and storage medium

Abstract

The application provides a bending deformation pipe section identification method and device, electronic equipment and storage medium. The method comprises the following steps: obtaining first strain data measured by an inertial measurement unit (IMU); obtaining a bending strain curve according to the first strain data; determining a wave crest in the bending strain curve, wherein the bending strain value of the wave crest is greater than a preset value; obtaining a to-be-identified pipe section, wherein the to-be-identified pipe section is a pipe section corresponding to a preset interval on both sides of the wave crest in a pipeline; inputting second strain data corresponding to the to-be-identified pipe section into a pipe section type identification model to obtain a pipe section abnormal type of the to-be-identified pipe section. According to the method, the to-be-identified pipe section is determined according to the obtained first strain data, and the second strain data corresponding to the to-be-identified pipe section is input into the pipe section type identification model, so that the pipe section abnormal type of the to-be-identified pipe section can be obtained. The judgment of the pipe section abnormal type does not need to be manually performed, time and manpower can be saved, a unified judgment standard is provided, and the judgment efficiency of the pipe section abnormal type is improved.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to a method, apparatus, electronic device and storage medium for identifying bent and deformed pipe sections. Background Technology

[0002] Long-distance oil and gas pipelines are characterized by long-distance transport, shallow burial depth, vast traversal areas, and complex and variable geological conditions, making them susceptible to damage from natural disasters such as earthquakes, subsidence, landslides, and mudslides. Geological disasters are one of the main causes of damage and failure of buried oil and gas pipelines. Changes in surface shape and soil displacement caused by geological disasters can lead to bending deformation of the pipeline. After bending deformation under external forces, stress concentration areas easily appear at the points of most severe deformation, welds, or deformation constraint points. In severe cases, this can lead to pipeline failure, causing incalculable economic losses, casualties, and environmental damage. Therefore, pipeline inspection technology is needed to monitor pipelines for anomalies to avoid various losses caused by pipeline failure.

[0003] Pipeline inspection technology relies on the pressure of the transported medium within the pipeline to drive the operation of an internal detector, thereby measuring defects and anomalies in oil and gas pipelines. Inertial Measurement Unit (IMU) inspection technology is a commonly used pipeline inspection technique and an effective method for detecting pipeline bending strain. Its core components are a three-dimensional orthogonal gyroscope and accelerometer. After completing the measurement of the entire pipeline under test, the IMU collects and records data, performs integration and other calculations, and obtains the velocity, position, and attitude data of the internal detector at any given time. Based on the attitude data obtained from the IMU, the deformation and strain state of each section of the pipeline under test can be calculated, and long-term monitoring can identify and control the risks caused by pipeline displacement and bending strain.

[0004] Existing methods for identifying and analyzing IMU strain data pre-identify all pipe segments in the IMU strain data with bending strain exceeding 0.125% using Matlab programming. These segments include elbows, depressions, bending deformation sections, and abnormal circumferential weld sections. Based on aligned geometric detection data, the mileage locations of given elbows and depressions are marked, and strain data within the influence range of geometric feature points such as depressions and elbows are deleted to eliminate interference from elbows and depressions. For the remaining unmarked pipe segments with bending strain values ​​greater than 0.125% through geometric detection, experienced technicians determine the segment type segment by segment using graphical methods and manually label them as suspected elbows, suspected depressions, abnormal circumferential weld sections, and bending deformation sections. This method uses geometric detection and manual identification to segmentally identify and classify IMU data to determine the location of pipelines prone to geological hazards.

[0005] Current methods for identifying and analyzing IMU strain data still rely on manual, segment-by-segment identification of high-risk pipe sections for geological hazards. This approach suffers from problems such as being time-consuming and having low accuracy. Summary of the Invention

[0006] This application provides a method, device, electronic device, and storage medium for identifying bent and deformed pipe sections, in order to solve the problems of long time consumption and low accuracy in existing identification and analysis of IMU strain data.

[0007] In a first aspect, this application provides a method for identifying bent and deformed pipe sections, including:

[0008] The first strain data is obtained from the inertial measurement unit (IMU), which is the pipeline strain data obtained by the IMU when it completes the pipeline inspection.

[0009] Based on the first strain data, the bending strain curve is obtained.

[0010] The peaks in the bending strain curve where the bending strain value is greater than a preset value are identified to obtain the pipe segments to be identified. The pipe segments to be identified are the pipe segments in the pipeline corresponding to the preset intervals on both sides of the peaks.

[0011] Input the second strain data corresponding to the pipe segment to be identified into the pipe segment type identification model to obtain the pipe segment anomaly type of the pipe segment to be identified.

[0012] Optionally, the pipe segment type identification model is trained based on a one-dimensional convolutional neural network model.

[0013] Optionally, before inputting the second strain data corresponding to the pipe segment to be identified into the pipe segment type identification model, the following steps are included:

[0014] The initial pipe segment type identification model is trained based on the sample database and the corresponding labels of the sample data to obtain the pipe segment type identification model. The sample database includes the pipe segment type, the absolute mileage range of the pipe segment strain data, the range of circumferential weld numbers, the inspection date, the pipe number, and the pipe segment strain value.

[0015] Optionally, an initial pipe segment type identification model is trained based on the sample database and the labels corresponding to the sample data to obtain the pipe segment type identification model, including:

[0016] Obtain sample data corresponding to four types of pipe section anomalies in the third strain data: indentation, bend, bending deformation, and circumferential weld anomaly. The sample database includes sample data corresponding to the four types of pipe section anomalies.

[0017] An initial pipe segment type identification model was trained based on sample data and labels corresponding to four pipe segment types, resulting in a pipe segment type identification model. The labels corresponding to the sample data include four types of pipe segment anomalies: dents, elbows, bending deformation, and circumferential weld anomalies.

[0018] Optionally, an initial pipe segment type recognition model is trained based on sample data corresponding to the four pipe segment types and the labels corresponding to the sample data, resulting in a pipe segment type recognition model, including:

[0019] The sample data is input into the initial pipe segment type identification model. After feature processing through the input layer, convolutional layer, pooling layer, flattening layer, fully connected layer and output layer, the predicted pipe segment type is obtained.

[0020] Based on the predicted pipe segment type and label, the network parameters of the initial pipe segment type identification model are updated to obtain the pipe segment type identification model.

[0021] Optionally, based on the predicted pipe segment type and label, the network parameters of the initial pipe segment type identification model are updated to obtain the pipe segment type identification model, including:

[0022] Based on the pipe segment prediction type and label, construct the loss function corresponding to the sample data.

[0023] Based on the loss function corresponding to the sample data, update the network parameters of the initial pipe segment type identification model to obtain the pipe segment type identification model.

[0024] Secondly, this application provides a device for identifying bent and deformed pipe sections, comprising:

[0025] The acquisition module is used to acquire the first strain data measured by the inertial measurement unit (IMU). The first strain data is the pipeline strain data obtained by the IMU when it completes the pipeline inspection.

[0026] The acquisition module is also used to obtain the bending strain curve based on the first strain data.

[0027] The determination module is used to determine the peaks in the bending strain curve where the bending strain value is greater than a preset value, and to obtain the pipe segments to be identified. The pipe segments to be identified are the pipe segments in the pipeline corresponding to the preset intervals on both sides of the peak.

[0028] The identification module is used to input the second strain data corresponding to the pipe segment to be identified into the pipe segment type identification model to obtain the pipe segment anomaly type of the pipe segment to be identified.

[0029] Optionally, the pipe segment type identification model is trained based on a one-dimensional convolutional neural network model.

[0030] Optionally, the bending deformation pipe segment identification device also includes a training module.

[0031] This training module is used to train an initial pipe segment type identification model based on a sample database and the labels corresponding to the sample data before inputting the second strain data corresponding to the pipe segment to be identified into the pipe segment type identification model. The sample database includes the pipe segment type, the absolute mileage range of the pipe segment strain data, the range of circumferential weld numbers, the inspection date, the pipe number, and the pipe segment strain value.

[0032] Optionally, this training module is specifically used for:

[0033] Obtain sample data corresponding to four types of pipe section anomalies in the third strain data: indentation, bend, bending deformation, and circumferential weld anomaly. The sample database includes sample data corresponding to the four types of pipe section anomalies.

[0034] An initial pipe segment type identification model was trained based on sample data and labels corresponding to four pipe segment types, resulting in a pipe segment type identification model. The labels corresponding to the sample data include four types of pipe segment anomalies: dents, elbows, bending deformation, and circumferential weld anomalies.

[0035] Optionally, this training module is specifically used for:

[0036] The sample data is input into the initial pipe segment type identification model. After feature processing through the input layer, convolutional layer, pooling layer, flattening layer, fully connected layer and output layer, the predicted pipe segment type is obtained.

[0037] Based on the predicted pipe segment type and label, the network parameters of the initial pipe segment type identification model are updated to obtain the pipe segment type identification model.

[0038] Optionally, this training module is specifically used for:

[0039] Based on the pipe segment prediction type and label, construct the loss function corresponding to the sample data.

[0040] Based on the loss function corresponding to the sample data, update the network parameters of the initial pipe segment type identification model to obtain the pipe segment type identification model.

[0041] Thirdly, this application provides an electronic device, including: a memory and a processor;

[0042] Memory is used to store computer programs.

[0043] The processor is used to read the computer program stored in the memory and execute the bending deformation pipe segment identification method of the first aspect described above according to the computer program in the memory.

[0044] Fourthly, this application provides a readable storage medium having a computer program stored thereon, the computer program storing computer execution instructions, which, when executed by a processor, are used to implement the bending deformation pipe segment identification method as described in the first aspect above.

[0045] Fifthly, embodiments of this application also provide a computer program product, including a computer program that, when executed by a processor, implements the bending deformation pipe segment identification method of the first aspect described above.

[0046] The bending deformation pipe segment identification method, device, electronic equipment, and storage medium provided in this application acquire first strain data measured by an inertial measurement unit (IMU). This first strain data is the pipe strain data obtained by the IMU during pipe inspection. Based on the first strain data, a bending strain curve is obtained. The peaks in the bending strain curve where the bending strain value exceeds a preset value are identified, thus determining the pipe segment to be identified. The pipe segment to be identified is the pipe segment corresponding to the preset interval on both sides of the peak within the pipe. The second strain data corresponding to the pipe segment to be identified is input into a pipe segment type identification model to obtain the pipe segment anomaly type. This method only requires acquiring the IMU strain data and identifying the pipe segment to be identified, and inputting the second strain data corresponding to the pipe segment to be identified into the pipe segment type identification model to obtain the pipe segment anomaly type. It eliminates the need for manual judgment of pipe segment anomaly types, saving time and manpower, standardizing the judgment criteria, and thus improving the efficiency of pipe segment anomaly type judgment. Attached Figure Description

[0047] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0048] Figure 1 A flowchart illustrating a method for identifying bent and deformed pipe sections provided in an embodiment of this application;

[0049] Figure 2 This is a schematic diagram of a pipe segment to be identified, provided as an embodiment of this application.

[0050] Figure 3 A schematic diagram of a one-dimensional convolutional neural network structure provided in this application embodiment;

[0051] Figure 4 A flowchart illustrating another method for identifying bent and deformed pipe segments provided in this application embodiment;

[0052] Figure 5 This is a schematic diagram of the structure of a bending deformation pipe section identification device provided in an embodiment of this application;

[0053] Figure 6This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.

[0054] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0055] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0056] First, let me explain the terms used in this application:

[0057] Machine learning: Machine learning is a method that involves training on large amounts of data to build a machine learning model, and then using the model to classify and predict new data. Based on different learning methods, machine learning algorithms can be divided into three categories: supervised learning, unsupervised learning, and reinforcement learning. Machine learning involves multiple interdisciplinary fields and mainly studies how computers can simulate or implement the learning behavior of the human brain to acquire new knowledge or skills, reorganize existing knowledge structures to continuously improve model performance, and thus solve for optimal parameters.

[0058] Supervised learning: Using samples with known characteristics as sample data, an optimal model is trained to achieve data classification and prediction.

[0059] Feature engineering, also known as variable selection, attribute selection, or variable subset selection in machine learning or statistics, is the process of selecting relevant features and forming a feature subset during model building.

[0060] The technical solutions provided in this application can be applied to the bending strain detection of buried oil and gas pipelines, especially the use of IMU (In-Mechanical Unit) internal inspection technology to detect defects and anomalies on the pipeline. my country has developed internationally leading internal inspection technologies, such as IMU and magnetic flux leakage. With the increasing prevalence of pipeline internal inspection technology, the amount of IMU strain data obtained will continue to grow. Therefore, conducting research on intelligent identification of high-risk pipeline sections for geological disasters based on machine learning is of great significance for identifying high-strain pipeline sections affected by geological disasters and for conducting integrity assessments of pipelines in geological disaster areas.

[0061] The existing method is a semi-automatic identification of high-risk pipe sections for geological hazards based on a Matlab algorithm. It aligns IMU strain data and geometric inspection data of the same pipeline using the circumferential weld number as a reference. Matlab programming pre-identifies all pipe sections in the IMU strain data with bending strain exceeding 0.125%, including elbows, depressions, bending deformation sections, and abnormal circumferential weld sections. Based on the aligned geometric inspection data, the mileage locations of the given elbows and depressions are marked. Strain data within the influence range of geometric feature points such as depressions and elbows are deleted to eliminate interference from elbows and depressions. For the remaining unmarked pipe sections with bending strain values ​​greater than 0.125% in the geometric inspection, experienced technicians use graphical methods to mark the pipe sections within a certain range on both sides of the large strain point as abnormal sections to be investigated. Based on the characteristics of the pipe sections within this range, the pipe section type is determined segment by segment and labeled as suspected elbow, suspected depression, abnormal circumferential weld section, and bending deformation section. This method can mark some pipe segments as bends or indentations using aligned geometric detection data, but a large number of segments still require manual identification. This is time-consuming and labor-intensive, and requires staff with ample practical experience, as non-professionals find it difficult to discern the characteristic differences between different pipe segment types. Furthermore, due to the large dataset, subtle characteristic differences still exist between pipe segment types, making it difficult to maintain a consistent standard during manual identification. This easily leads to misjudgments, and accurate judgments are difficult to make for ambiguous pipe segment features.

[0062] To address the above issues, this application proposes a method for identifying bent and deformed pipe segments. This method involves acquiring first strain data measured by an inertial measurement unit (IMU), which is the pipe strain data obtained by the IMU during in-pipe inspection. Based on this first strain data, a bending strain curve is generated. Peaks in the bending strain curve where the bending strain value exceeds a preset value are identified. Preset intervals on both sides of these peaks correspond to pipe segments to be identified. Second strain data corresponding to these pipe segments is then input into a pipe segment type identification model to determine the pipe segment anomaly type. In this application, the pipe segment to be identified is determined based on the first strain data obtained from the IMU, and the corresponding second strain data is input into the pipe segment type identification model to obtain the pipe segment anomaly type. This eliminates the need for manual judgment of pipe segment anomaly types, saving time and manpower, standardizing the judgment criteria, and thus improving the efficiency of pipe segment anomaly type identification.

[0063] The technical solution of this application and how it solves the above-mentioned technical problems will be described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will be described below with reference to the accompanying drawings.

[0064] Figure 1 This is a flowchart illustrating a method for identifying bent and deformed pipe segments according to an embodiment of this application. This method can be executed by software and / or hardware devices, such as electronic devices like terminals or servers. For example, please refer to [link to example]. Figure 1 As shown, the method for identifying bent and deformed pipe sections may include:

[0065] S101. Obtain the first strain data measured by the inertial measurement unit (IMU).

[0066] The first strain data is the pipeline strain data obtained by the IMU when it completes the pipeline inspection.

[0067] In this step, the Inertial Measurement Unit (IMU) is a device that measures the three-axis attitude angles (or angular rates) and acceleration of an object. Its core components are a three-dimensional orthogonal gyroscope and accelerometer. After completing the measurement of the entire pipeline under test, the IMU performs calculations such as integration on the data collected and recorded by the IMU to obtain the velocity, position, and attitude data of the IMU at any given time. Based on the attitude data detected by the IMU, the deformation and strain state of the pipeline segment can be calculated. The first strain data can be understood as the IMU strain data obtained by measuring the pipeline under test through IMU in-system detection technology, including information such as the velocity, position, and attitude data of the IMU at any given time within the pipeline.

[0068] Specifically, the inertial measurement unit (IMU) installed inside the pipeline takes measurements to obtain the velocity, position, and attitude data of the IMU at any given moment within the pipeline, which is the first strain data.

[0069] S102. Based on the first strain data, the bending strain curve is obtained.

[0070] In this step, the bending strain curve can be represented as the relationship curve between the bending strain value and the absolute mileage. The bending strain value can be calculated based on the IMU strain data.

[0071] Specifically, the corresponding bending strain value can be obtained from the first strain data, and then a bending strain curve can be plotted based on this bending strain value and its corresponding absolute mileage. The absolute mileage can be understood as the distance from the strain point to the starting position of the measurement along the entire pipeline under test.

[0072] S103. Determine the peak in the bending strain curve where the bending strain value is greater than the preset value to obtain the pipe section to be identified.

[0073] In this step, the pipe segment to be identified is the pipe segment in the pipeline corresponding to the preset interval on both sides of the wave crest.

[0074] The preset value can be set according to actual conditions or experience, for example, it can be set to 0.125%. The specific value of the preset value is not limited in this embodiment. When determining the peak in the bending strain curve where the bending strain value is greater than the preset value, the peak can be determined using the `findpeak` function in Python, or it can be implemented programmatically. The `findpeak` function can be expressed as:

[0075] peaks=scipy.signal.find_peaks(x,threshold,distance) (1)

[0076] In equation (1): x is a signal with a peak value, threshold is the set recognition threshold, and distance is the minimum horizontal distance between two peak values.

[0077] Specifically, based on the preset values, the peaks in the bending strain curve where the bending strain value is greater than the preset value can be found. Then, two strain points on the bending strain curves on the left and right sides of the peak, where the bending strain value is the preset value, can be found respectively. The absolute mileage intervals corresponding to these two strain points are the preset intervals on both sides of the peak. Thus, the pipe segments within the absolute mileage intervals in the pipeline can be found, which are the pipe segments to be identified.

[0078] in, Figure 2 This is a schematic diagram of a pipe segment to be identified, such as... Figure 2 As shown, the preset value is 0.125%. Based on this preset value of 0.125%, peaks A, B, and C with bending strain values ​​greater than 0.125% can be found in the bending strain curve. Taking peak A as an example, the two strain points with bending strain values ​​of 0.125% on the bending strain curves on the left and right sides of peak A are E and F, respectively. The absolute mileage corresponding to strain point E is 3330m, and the absolute mileage corresponding to strain point F is 3340m. Therefore, the absolute mileage interval corresponding to the two strain points is (3330, 3340). Thus, the preset interval (3330, 3340) on both sides of peak A can be obtained. At this time, the pipe segment corresponding to the preset interval (3330, 3340) is the pipe segment to be identified 1.

[0079] S104. Input the second strain data corresponding to the pipe segment to be identified into the pipe segment type identification model to obtain the pipe segment anomaly type of the pipe segment to be identified.

[0080] In this step, the second strain data can be understood as the IMU strain data obtained by measuring the pipe segment to be identified through IMU in-sight technology.

[0081] For example, the pipe segment type identification model can be trained based on a one-dimensional neural network model, such as a one-dimensional convolutional neural network model. Training the initial pipe segment type identification model using a one-dimensional neural network model eliminates the need for feature engineering to extract feature values, thereby reducing computational load and simplifying the feature extraction process.

[0082] The pipe segment type identification model can also be trained by other network models, such as multidimensional convolutional neural network models, recurrent neural network models, or random forest models.

[0083] For example, before inputting the second strain data corresponding to the pipe segment to be identified into the pipe segment type identification model, an initial pipe segment type identification model is first trained. The specific steps can be: train the initial pipe segment type identification model based on the sample database and the labels corresponding to the sample data to obtain the pipe segment type identification model.

[0084] The sample database includes pipe segment type, absolute mileage range of pipe segment strain data, circumferential weld number range, inspection date, pipe number, and pipe segment strain value. The sample database can be a MySQL database.

[0085] Specifically, when training the initial pipe segment type recognition model, the sample data and the corresponding labels in the sample database are input into the initial pipe segment type recognition model, and the pipe segment type recognition model is trained based on the sample data and the corresponding labels in the sample database.

[0086] Through years of pipeline data monitoring practice, a large amount of pipeline IMU strain data has been obtained. However, despite this wealth of data, corresponding data processing methods are still lacking. This solution establishes a sample database for different pipe segment types, forming a big data-driven machine learning method. Furthermore, as pipeline inspection technology becomes more widespread and applied, the database is continuously expanded, which in turn influences the machine learning method, improving the accuracy of pipe segment type identification.

[0087] For example, the sample database includes sample data corresponding to four types of pipe segment anomalies: dents, elbows, bending deformation, and circumferential weld anomalies. When training the initial pipe segment type identification model based on the sample data and corresponding labels in the sample database, sample data corresponding to the four types of pipe segment anomalies (dents, elbows, bending deformation, and circumferential weld anomalies) are obtained from the third strain data. The initial pipe segment type identification model is then trained based on the sample data and corresponding labels for the four types of pipe segment anomalies, resulting in the pipe segment type identification model.

[0088] The sample data is tagged with four types of pipe section anomalies: dents, bends, bending deformation, and circumferential weld anomalies. The third type of strain data can be understood as historical IMU strain data obtained based on IMU in-system detection technology; it can be obtained from a local database or downloaded from a server.

[0089] For example, Table 1 is a sample database. As shown in Table 1, the sample database includes pipe segment type, absolute mileage range of pipe segment strain data, circumferential weld number range, inspection date, pipeline number, and pipe segment strain value. The pipe segment type includes four types of pipe segment anomalies: depression, elbow, bending deformation, and circumferential weld anomaly. These four types are represented by numerical labels 1, 2, 3, and 4, respectively. Taking pipe segment type 1 as an example, this pipe segment type is depression, corresponding to label 1, absolute mileage range is 600.9-620.5m, circumferential weld number range is 400-420, inspection date is May 2016, pipeline number is segment A, and pipe segment strain value is [-0.86%, ..., 0.23%].

[0090] Table 1 Sample Database

[0091]

[0092] Specifically, firstly, sample data corresponding to four pipe segment anomaly types—dents, bends, bending deformation, and circumferential weld anomalies—are obtained from historical IMU strain data. Specifically, sample data corresponding to dents and bends in the third strain data are extracted using geometric detection methods, while sample data corresponding to bending deformation and circumferential weld anomalies in the third strain data are extracted using manual identification methods. Based on the sample data corresponding to the four pipe segment types and their corresponding labels, an initial pipe segment type identification model is trained, resulting in the pipe segment type identification model.

[0093] In this scheme, the establishment of sample databases for different pipe segment types lays a data foundation for the application and continuous improvement of machine learning methods. Based on the existing database, detailed information on the required pipe segments can be quickly retrieved, and a basis for judgment can be provided for subsequent pipe segment development. Furthermore, the pipe segment type recognition model, trained using sample data and labels corresponding to four pipe segment anomaly types—dents, bends, bending deformations, and circumferential weld anomalies—only requires geometric detection data and manual identification results when establishing the sample database. After training the pipe segment type recognition model, the pipe segment type can be predicted solely based on IMU strain data, eliminating the need for manual identification methods. This saves time and manpower, standardizes judgment criteria, and improves the efficiency of pipe segment anomaly type identification.

[0094] For example, the pipe segment type recognition model can be trained based on a one-dimensional convolutional neural network model. This model can include six parts: an input layer, a convolutional layer, a pooling layer, a flattening layer, a fully connected layer, and an output layer. When training the initial pipe segment type recognition model, sample data is first input into the initial pipe segment type recognition model. After feature processing by the input layer, convolutional layer, pooling layer, flattening layer, fully connected layer, and output layer, the predicted pipe segment type is obtained. Based on the predicted pipe segment type and label, the network parameters of the initial pipe segment type recognition model are updated to obtain the pipe segment type recognition model.

[0095] Specifically, the basic framework of a one-dimensional convolutional neural network model can be built using the TensorFlow module in Python. Figure 3 This is a schematic diagram of a one-dimensional convolutional neural network structure. The model can include six parts: an input layer (not shown), a convolutional layer, a pooling layer, a flattening layer, a fully connected layer, and an output layer (not shown).

[0096] The input layer of a one-dimensional convolutional neural network model can receive one-dimensional or two-dimensional arrays. Sample data is input to the input layer of the initial pipe segment type identification model. This sample data consists of IMU strain data, including components in both horizontal and vertical directions, i.e., a 2×n pipe segment strain feature matrix, which can be represented as:

[0097]

[0098] In equation (2): T is the strain value matrix; Let be the horizontal strain component at the nth strain point; Let be the vertical strain component at the nth strain point.

[0099] The input layer is the input port for sample data of the neural network model. The preset input dimension of this layer needs to be consistent with the dimension of the sample data. The one-dimensional convolutional neural network needs to transform the 2×n pipe segment strain feature matrix corresponding to the multi-dimensional sample data, i.e. the second strain data, into one-dimensional data in the input layer.

[0100] Convolutional layers are feature extraction layers. These layers reduce the dimensionality of the original sample data through convolution operations, while extracting the main features of the samples, thus preventing the model from overfitting due to excessive parameters.

[0101] The role of pooling layers is to further reduce the size of feature data, which can significantly reduce the number of parameters in the network and remove redundant information. They mainly include two forms: mean sampling and maximum sampling.

[0102] Because the data passes through a filter, the dimensionality of the data changes. The role of the flattening layer is to convert the data to be processed into a one-dimensional vector before it is input into the fully connected neural network, and then map it onto the neurons in the fully connected layer.

[0103] A fully connected layer consists of multiple neurons that are fully connected to each other. It reassembles the local features from the pooling layer using a weight matrix to form complete global features.

[0104] The output layer consists of a regression classifier, which can use the softmax function to map the outputs of multiple neurons to a fixed interval, thereby enabling the identification of various pipe segment anomaly types by the pipe segment type recognition model.

[0105] After feature processing through the input layer, convolutional layer, pooling layer, flattening layer, fully connected layer, and output layer, the predicted pipe segment type is obtained. Based on the predicted pipe segment type and label, the network parameters of the initial pipe segment type recognition model are continuously updated to obtain the pipe segment type recognition model.

[0106] In this scheme, the initial pipe segment type recognition model is trained by a one-dimensional convolutional neural network model, which eliminates the need for feature engineering to extract feature values, thereby reducing the amount of computation and simplifying the feature extraction process.

[0107] For example, when updating the network parameters of the initial pipe segment type identification model based on the predicted pipe segment type and label, the loss function corresponding to the sample data is first constructed based on the predicted pipe segment type and label, and the network parameters of the initial pipe segment type identification model are updated based on the loss function corresponding to the sample data to obtain the pipe segment type identification model.

[0108] Specifically, when constructing the loss function corresponding to the sample data based on the pipe segment prediction type and label, the loss function can be the cross-entropy loss function, and the formula can be expressed as:

[0109]

[0110] In equation (3): n is the number of samples, x is the sample number, y represents the true value, and a represents the predicted value.

[0111] For example, all sample data in the sample database are divided into training dataset and test dataset according to a certain ratio, such as 4:1. During the training process, the batch size and the number of iterations (epochs) are set, and the trained pipe segment type recognition model and its parameters are stored.

[0112] In this scheme, by constructing a loss function corresponding to the predicted pipe segment type and the label corresponding to the sample data, the network parameters of the initial pipe segment type identification model are continuously updated, thereby improving the accuracy of the pipe segment type identification model training.

[0113] When evaluating the overall performance of a model, parameter tuning is performed on the network model to determine the optimal hyperparameters. In supervised learning classification models, commonly used evaluation metrics include accuracy, precision, recall, and the fiat index (FI). The formulas can be expressed as:

[0114]

[0115]

[0116]

[0117]

[0118] In the above formula, Accuracy represents accuracy, Precision represents precision, Recall represents recall, FIScore represents FI value, TP represents correctly classified as a positive sample, TN represents correctly classified as a negative sample, FP represents misclassified as a positive sample, and FN represents misclassified as a negative sample.

[0119] For example, when evaluating the performance of a pipe segment type identification model, if the accuracy of the pipe segment type identification model is greater than or equal to a preset value, then the pipe segment type identification model can accurately identify the abnormal type of the pipe segment.

[0120] The bending deformation pipe segment identification method provided in this application acquires first strain data measured by an inertial measurement unit (IMU). This first strain data is the pipe strain data obtained by the IMU during pipe inspection. Based on the first strain data, a bending strain curve is obtained. The peaks in the bending strain curve where the bending strain value exceeds a preset value are identified. The pipe segments corresponding to the preset intervals on both sides of the peaks are the pipe segments to be identified. The second strain data corresponding to the pipe segment to be identified is input into a pipe segment type identification model to obtain the pipe segment anomaly type. In this application, the pipe segment to be identified is determined based on the acquired IMU strain data, and the second strain data corresponding to the pipe segment to be identified is input into the pipe segment type identification model to obtain the pipe segment anomaly type. This eliminates the need for manual judgment of pipe segment anomaly types, saving time and manpower, standardizing the judgment criteria, and thus improving the efficiency of pipe segment anomaly type judgment.

[0121] Figure 4 For a flowchart illustrating another method for identifying bent and deformed pipe sections provided in this application, please refer to [link / reference]. Figure 4 As shown, this method for identifying bent and deformed pipe segments can include a training process and an application process:

[0122] When training the initial pipe segment type identification model, the third strain data is first preprocessed. This involves extracting sample data corresponding to depressions and bends from the third strain data using geometric detection methods, and extracting sample data corresponding to bending deformation and circumferential weld anomalies from the third strain data using manual identification methods. Then, a sample database is established based on the extracted sample data corresponding to the four pipe segment types. The initial pipe segment type identification model is trained based on the sample database and the labels corresponding to the sample data, resulting in the pipe segment type identification model. Finally, the performance of the pipe segment type identification model is evaluated. When the accuracy of the pipe segment type identification model is greater than or equal to a preset value, the pipe segment type identification model can accurately identify the pipe segment anomaly type.

[0123] When using the pipe segment type identification model to identify pipe segment anomaly types, the first strain data obtained by the inertial measurement unit (IMU) is first acquired. The first strain data is the pipe strain data obtained by the IMU when it completes the pipe inspection. Based on the first strain data, the bending strain curve is obtained, and the peak of the bending strain value in the bending strain curve that is greater than the preset value is determined to obtain the pipe segment to be identified. The second strain data corresponding to the pipe segment to be identified is then input into the pipe segment type identification model to obtain the pipe segment anomaly type of the pipe segment to be identified, thereby realizing the intelligent identification and classification of high-risk sections of geological disasters.

[0124] The bending deformation pipe segment identification method provided in this application uses a pipe segment type identification model trained with sample data and labels corresponding to four pipe segment anomaly types: dents, bends, bending deformation, and abnormal circumferential welds. It acquires first strain data measured by an inertial measurement unit (IMU), which is the pipe strain data obtained by the IMU during in-pipe inspection. Based on the first strain data, a bending strain curve is obtained. The peaks in the bending strain curve where the bending strain value exceeds a preset value are identified. The pipe segments corresponding to the preset intervals on both sides of the peaks are the pipe segments to be identified. The second strain data corresponding to the pipe segment to be identified is input into the pipe segment type identification model to obtain the pipe segment anomaly type. In this application, geometric detection data and manual identification results are required when establishing the sample database. After training the pipe segment type identification model, the pipe segment anomaly type can be predicted solely based on IMU strain data, eliminating the need for manual identification. This saves time and manpower, standardizes the identification criteria, and improves the efficiency of pipe segment anomaly type identification.

[0125] Figure 5 This application provides a schematic diagram of the structure of a bending deformation pipe section identification device 50. For an example, please refer to [link to example diagram]. Figure 5 As shown, the bending deformation pipe section identification device 50 includes:

[0126] The acquisition module 501 is used to acquire the first strain data measured by the inertial measurement unit (IMU). The first strain data is the pipeline strain data obtained by the IMU when it completes the pipeline inspection.

[0127] The acquisition module 501 is also used to obtain the bending strain curve based on the first strain data.

[0128] The determination module 502 is used to determine the peak in the bending strain curve where the bending strain value is greater than the preset value, and obtain the pipe segment to be identified. The pipe segment to be identified is the pipe segment in the pipeline corresponding to the preset interval on both sides of the peak.

[0129] The identification module 503 is used to input the second strain data corresponding to the pipe segment to be identified into the pipe segment type identification model to obtain the pipe segment anomaly type of the pipe segment to be identified.

[0130] Optionally, the pipe segment type identification model is trained based on a one-dimensional convolutional neural network model.

[0131] Optionally, the bending deformation pipe segment identification device also includes a training module 504.

[0132] The training module 504 is used to train an initial pipe segment type identification model based on a sample database and the labels corresponding to the sample data before inputting the second strain data corresponding to the pipe segment to be identified into the pipe segment type identification model. The sample database includes the pipe segment type, the absolute mileage range of the pipe segment strain data, the range of circumferential weld numbers, the inspection date, the pipe number, and the pipe segment strain value.

[0133] Optionally, this training module 504 is specifically used for:

[0134] Obtain sample data corresponding to four types of pipe section anomalies in the third strain data: indentation, bend, bending deformation, and circumferential weld anomaly. The sample database includes sample data corresponding to the four types of pipe section anomalies.

[0135] An initial pipe segment type identification model was trained based on sample data and labels corresponding to four pipe segment types, resulting in a pipe segment type identification model. The labels corresponding to the sample data include four types of pipe segment anomalies: dents, elbows, bending deformation, and circumferential weld anomalies.

[0136] Optionally, this training module 504 is specifically used for:

[0137] The sample data is input into the initial pipe segment type identification model. After feature processing through the input layer, convolutional layer, pooling layer, flattening layer, fully connected layer and output layer, the predicted pipe segment type is obtained.

[0138] Based on the predicted pipe segment type and label, the network parameters of the initial pipe segment type identification model are updated to obtain the pipe segment type identification model.

[0139] Optionally, this training module 504 is specifically used for:

[0140] Based on the pipe segment prediction type and label, construct the loss function corresponding to the sample data.

[0141] Based on the loss function corresponding to the sample data, update the network parameters of the initial pipe segment type identification model to obtain the pipe segment type identification model.

[0142] The bending deformation pipe segment identification device 50 shown in this application embodiment can execute the technical solution of the bending deformation pipe segment identification method in the above embodiment. Its implementation principle and beneficial effects are similar to those of the bending deformation pipe segment identification method. Please refer to the implementation principle and beneficial effects of the bending deformation pipe segment identification method. It will not be repeated here.

[0143] Figure 6 This is a schematic diagram of the structure of an electronic device 60 provided in an embodiment of this application. For example, please refer to [link to example diagram]. Figure 6 As shown, the electronic device 60 may include a processor 601 and a memory 602; wherein,

[0144] Memory 602 is used to store computer programs.

[0145] The processor 601 is used to read the computer program stored in the memory 602 and execute the bending deformation pipe segment identification method in the above embodiment according to the computer program in the memory 602.

[0146] Optionally, the memory 602 can be either standalone or integrated with the processor 601. When the memory 602 is a device independent of the processor 601, the electronic device 60 may further include a bus for connecting the memory 602 and the processor 601.

[0147] Optionally, this embodiment also includes a communication interface, which can be connected to the processor 601 via a bus. The processor 601 can control the communication interface to realize the acquisition and transmission functions of the aforementioned electronic device 60.

[0148] For example, in this embodiment of the application, the electronic device 60 can be a terminal or a server, which can be set according to actual needs.

[0149] The electronic device 60 shown in this application embodiment can execute the technical solution of the bending deformation pipe segment identification method in the above embodiment. Its implementation principle and beneficial effects are similar to those of the bending deformation pipe segment identification method. Please refer to the implementation principle and beneficial effects of the bending deformation pipe segment identification method. It will not be repeated here.

[0150] This application also provides a computer-readable storage medium storing computer-executable instructions. When a processor executes the computer-executable instructions, it implements the technical solution of the bending deformation pipe segment identification method in the above embodiments. Its implementation principle and beneficial effects are similar to those of the bending deformation pipe segment identification method, and can be found in the implementation principle and beneficial effects of the bending deformation pipe segment identification method, which will not be repeated here.

[0151] This application also provides a computer program product, including a computer program. When the computer program is executed by a processor, it implements the technical solution of the bending deformation pipe segment identification method in the above embodiments. Its implementation principle and beneficial effects are similar to those of the bending deformation pipe segment identification method. Please refer to the implementation principle and beneficial effects of the bending deformation pipe segment identification method, which will not be repeated here.

[0152] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the mutual coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0153] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated units described above can be implemented in hardware or in a combination of hardware and software functional units.

[0154] The integrated modules described above, implemented as software functional modules, can be stored in a computer-readable storage medium. These software functional modules, stored in a storage medium, include several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute some steps of the methods of the various embodiments of this application.

[0155] It should be understood that the aforementioned processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. A general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly manifested as being executed by a hardware processor, or executed by a combination of hardware and software modules within the processor.

[0156] The memory may include high-speed random access memory (RAM) and may also include non-volatile memory (NVM), such as at least one disk storage device, and may also be a USB flash drive, external hard drive, read-only memory, disk or optical disc, etc.

[0157] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, the buses shown in the accompanying drawings are not limited to a single bus or a single type of bus.

[0158] The aforementioned computer-readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random-Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read-Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The storage medium can be any available medium accessible to general-purpose or special-purpose computers.

[0159] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

Claims

1. A method for identifying bent and deformed pipe sections, characterized in that, include: Acquire the first strain data measured by the inertial measurement unit (IMU), wherein the first strain data is the pipeline strain data obtained by the IMU when completing the pipeline inspection; Based on the first strain data, the bending strain curve is obtained; The peak in the bending strain curve with a bending strain value greater than a preset value is identified, and two strain points on the bending strain curves on the left and right sides of the peak with bending strain values ​​of the preset value are identified. The absolute mileage intervals corresponding to the two strain points are the preset intervals on both sides of the peak, and the pipe segments in the pipeline within the preset intervals are identified as the pipe segments to be identified. The second strain data corresponding to the pipe segment to be identified is input into the pipe segment type identification model to obtain the pipe segment anomaly type of the pipe segment to be identified; the pipe segment type identification model is obtained by training a one-dimensional convolutional neural network model. The pipe segment type identification model was trained based on the following method: Obtain sample data corresponding to four types of pipe section anomalies in the third strain data: indentation, elbow, bending deformation, and circumferential weld anomaly. The sample database includes sample data corresponding to the four types of pipe section anomalies. The sample database includes pipe section type, absolute mileage range of pipe section strain data, circumferential weld number range, inspection date, pipe number, and pipe section strain value. The initial pipe segment type identification model is trained based on the sample data corresponding to the four pipe segment types and the labels corresponding to the sample data to obtain the pipe segment type identification model; wherein, the labels corresponding to the sample data include four types of pipe segment anomalies: dents, elbows, bending deformations, and abnormal circumferential welds.

2. The method according to claim 1, characterized in that, The initial pipe segment type identification model is trained based on sample data corresponding to the four pipe segment types and the labels corresponding to the sample data to obtain the pipe segment type identification model, including: The sample data is input into the initial pipe segment type identification model. After feature processing through the input layer, convolutional layer, pooling layer, flattening layer, fully connected layer and output layer, the predicted pipe segment type is obtained. Based on the predicted pipe segment type and the label, the network parameters of the initial pipe segment type identification model are updated to obtain the pipe segment type identification model.

3. The method according to claim 2, characterized in that, The step of updating the network parameters of the initial pipe segment type identification model based on the predicted pipe segment type and the label to obtain the pipe segment type identification model includes: Based on the predicted pipe segment type and the label, construct the loss function corresponding to the sample data; Based on the loss function corresponding to the sample data, the network parameters of the initial pipe segment type identification model are updated to obtain the pipe segment type identification model.

4. A device for identifying bent and deformed pipe sections, characterized in that, include: The acquisition module is used to acquire the first strain data measured by the inertial measurement unit (IMU), wherein the first strain data is the pipeline strain data obtained by the IMU when it completes the pipeline inspection. The acquisition module is further configured to obtain a bending strain curve based on the first strain data; The determination module is used to determine the peak in the bending strain curve where the bending strain value is greater than a preset value, and to determine two strain points on the bending strain curve on the left and right sides of the peak where the bending strain value is a preset value. The absolute mileage intervals corresponding to the two strain points are the preset intervals on both sides of the peak, and the pipe segment in the pipeline within the preset interval is determined as the pipe segment to be identified. The identification module is used to input the second strain data corresponding to the pipe segment to be identified into the pipe segment type identification model to obtain the pipe segment anomaly type of the pipe segment to be identified; the pipe segment type identification model is trained based on a one-dimensional convolutional neural network model. The pipe segment type identification model was trained based on the following method: Obtain sample data corresponding to four types of pipe section anomalies in the third strain data: indentation, elbow, bending deformation, and circumferential weld anomaly. The sample database includes sample data corresponding to the four types of pipe section anomalies. The sample database includes pipe section type, absolute mileage range of pipe section strain data, circumferential weld number range, inspection date, pipe number, and pipe section strain value. The initial pipe segment type identification model is trained based on the sample data corresponding to the four pipe segment types and the labels corresponding to the sample data to obtain the pipe segment type identification model; wherein, the labels corresponding to the sample data include four types of pipe segment anomalies: dents, elbows, bending deformations, and abnormal circumferential welds.

5. An electronic device, characterized in that, include: Memory and processor; The memory is used to store computer programs; The processor is configured to read the computer program stored in the memory and execute the bending deformation pipe segment identification method according to any one of claims 1-3 based on the computer program in the memory.

6. A readable storage medium having a computer program stored thereon, characterized in that, The computer program stores computer execution instructions, which, when executed by a processor, are used to implement the bending deformation pipe segment identification method as described in any one of claims 1-3.

7. A computer program product comprising a computer program, which, when executed by a processor, implements the bending deformation pipe segment identification method according to any one of claims 1-3.

Images