Research on defect identification method of oil and gas pipeline based on ultrasonic echo signal

As a vital infrastructure for oil and gas transportation, pipelines are susceptible to erosion, corrosion, and pressure fluctuations, potentially leading to defects like dents and cracks, which increase leakage risks. Traditional defect recognition methods often lack reliability. To tackle this issue, this paper presents a method for identifying defect types in oil and gas pipelines by combining variational mode decomposition (VMD) and wavelet packet decomposition (WPD) to extract features from ultrasonic echo signals. The approach begins with a hybrid strategy to optimize VMD parameters, followed by applying optimized VMD to decompose the original ultrasound echo signal. The Pearson correlation coefficient is used to select the optimal intrinsic mode component (IMF) for effective denoising. Subsequently, the denoised signal undergoes wavelet packet decomposition to extract defect energy features, which are then input into a random forest for recognition and classification. Validation through finite element simulation and actual ultrasound echo data shows that the proposed method achieves over 90% in accuracy, F1 score, precision, and recall for identifying defects such as dents, cracks, surface peeling, and perforations. Compared to other methods, it demonstrates significant advantages across all evaluation metrics, showcasing superior defect recognition performance.