Affiliations: Faculty of Engineering, Osaka Electric-Communication University, 18-8, Hatsu-cho, Neyagawa, 572-8530, Japan | College of Integrated Arts and Sciences, Osaka Prefecture University, 1-1, Gakuen-cho, Sakai, 599-8531, Japan | Faculty of Engineering, Osaka Sangyo University, 3-1-1, Nakagaito, Daito, 574-8530, Japan | Faculty of Engineering, Setsunan University, 17-8, Ikedanaka-machi, Neyagawa, 572-8508, Japan
Abstract: A computer aided evaluation system that can be used to discriminate the joint performance (joint strength) of SUS304 stainless steel friction welded joints is proposed. The data used for analyze are four input factors of the friction welding conditions, two shape parameters of the upset burr created during the welding process and the total burn-off quantity. The input factors are set before welding, and after welding the parameter of burr shape and the burn-off quantity are measured by using a vernier caliper. The learning of the synapse weights of the neural network is performed using the extended Kalman filtering algorithm. The results of experiments and the analysis of the joint performance of stainless steel joints suggest that the proposed method is superior to a trial and error and other conventional statistical methods. More, it is suggested a method which can be obtained the optimum welding conditions based on the tensile strength estimated.
