AI Defect Recognition System for Epoxy Resin Pipeline in Factory

In petrochemical, oil refining and other factories, process pipelines are important elements of production and are often loaded with flammable, explosive, and toxic fluids. These pipelines, which cannot be effectively and comprehensively inspected, are often leaked or damaged during operation and resulting in unplanned shutdown, industrial safety accidents or casualties. This research adopts external visual inspection (VT) and infrared thermal image inspection (IRT) and other inspection technologies to design a process pipeline defect inspection and information system. Moreover, for defects such as pipeline blockage, thinning, small cracks and leakage that cannot be discovered by VT inspection, the possible location of the defect can be detected through the temperature change of the thermal imager and further inspection can be carried out. This research mainly uses the deep learning algorithm to automatically identify the pipeline problems in the captured pipeline images to achieve the goal of automatic VT detection.