Study of different machine learning methods in welded seam width prediction

Abstract

As an important new laser processing technique, the high-power disk laser welding has been increasingly widely used in the manufacturing area. Aiming at the strong coupling multi-variable and real- time feedback requirements of the welding process, a new method using support vector machine is proposed to predict the width of the molten pools. The performance of this model is validated by the test data. Meanwhile, analysis and comparison between the support vector machines and the BP neural network are conducted. Experiment results show that the support vector machine and the BP neural network both have a good predictive ability. However, in comparison with the BP neural network, the support vector machine is more suitable for high-power disk laser welding process.