Real-time recognition of weld defects based on visible spectral image and machine learning

The quality of Tungsten Inert Gas welding is dependent on human supervision, which can’t suitable for automation. This study designed a model assessing the tungsten inert gas with potential application in real-time. used K-Nearest Neighborhood (KNN) algorithm, paired images visible spectrum formed by high dynamic range camera. Firstly, projecting image weld defects training set into two-dimensional space using multidimensional scaling (MDS), so similar was aggregated blocks and distributed hash, among different has overlap. Secondly, establishing models including KNN, CNN, SVM, CART NB classification, to classify recognize defect images. results show that KNN best, recognition accuracy 98%, average time recognizing single 33ms, common hardware devices. It can be applied system automatic robot improve intelligent level robot.