Research on Intelligent Detection Technology of Surface Defects of Nuclear Fuel Rods Based on Machine Vision

Abstract

Nuclear fuel rods are the core of nuclear reactors. The main function of the fuel rod is to release heat and contain nuclear fission products. During operation condition, fuel rods are under long term infection of high temperature, high pressure and high radiation field. Nuclear fuel rods may occur surface defects such as scratches, pits, and scrubs during manufacture process. These surface defects may cause the fuel rods to break and leak during operation. The consequence is that nuclear fission products escape from the breakage into the nuclear reactor loop. Therefore, the detection of surface defects on fuel rods is important. Based on machine vision and deep learning theory, this paper studies the intelligent detection technology of fuel rod surface defects based on machine vision and deep learning. Through convolutional neural networks with multiscale algorithms to achieve sample identification and training. Finally, the experimental results show that the proposed method has high surface defect recognition accuracy and detection efficiency.