ABSTRACT Magnetic flux leakage detection is a well‐established non‐destructive in‐line inspection for pipelines. In practical applications, the volume of inspection data is large, and manually labeling of defects is time‐consuming and inefficient. Automated processing of inspection data using machine learning methods can address these issues. This study evaluates the performance of four machine learning models in defect classification and defect depth prediction. The synthetic minority oversampling technique (SMOTE) is employed to increase the number of minority class samples, enhancing the model's generalization ability and thereby improving the classification accuracy for minority class defects. The prediction results showed that the Categorical Boosting (CatBoost) model had a defect classification accuracy of 0.9730, a mean squared error of 0.0256 for defect depth prediction, and a prediction residual range of −1 to 1.1 mm. The CatBoost model had the advantages of high classification accuracy, a small prediction error, and a residual range. The study results provided a reference for predicting pipeline defect types and defect depths using machine learning models.
Building similarity graph...
Analyzing shared references across papers
Loading...
Cong Chen
Rui Li
Guanwei Jia
Energy Science & Engineering
Henan University
Henan Bioengineering Technology Research Center
Line Corporation (Japan)
Building similarity graph...
Analyzing shared references across papers
Loading...
Chen et al. (Mon,) studied this question.
www.synapsesocial.com/papers/69af953870916d39fea4c8ba — DOI: https://doi.org/10.1002/ese3.70486
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: