Image Classification Method Based on Improved Deep Convolutional Neural Networks for the Magnetic Flux Leakage (MFL) Signal of Girth Welds in Long-Distance Pipelines

Girth weld defects in long-distance oil and gas pipelines are one of the main causes pipeline leakage failure serious accidents. Magnetic flux (MFL) is most widely used inline inspection methods for pipelines. However, it impossible to determine type girth defect via traditional manual analysis due complexity MFL signal. Therefore, an automatic image classification method based on deep convolutional neural networks was proposed effectively classify signals. Firstly, data set welds signal established with radiographic testing results as labels. Then, generative adversarial network (DCGAN) enhancement algorithm enhance set, residual (ResNet-50) address challenge presented by sets. The after randomly selected train test improved (ResNet-50), ten validation exhibiting accuracy over 80%. indicated that model displayed a strong generalization ability robustness could achieve more accurate welds.