Optimizing the auto-brazing process quality via a Taguchi-neural network approach in the automotive industry

Many parameters affect the quality of the auto-brazing process. It is not easy to obtain optimal parameters of this process. This paper applies an integrated approach using the Taguchi method and a neural network (NN) to optimize the lap joint quality of air conditioner parts. The proposed approach consists of two phases. First phase executes initial optimization via Taguchi method to construct a database for the NN. In second phase, we use a NN with the Levenberg-Marquardt back-propagation (LMBP) algorithm to provide the nonlinear relationship between factors and the response based on the experimental data. Then, a well-trained network model is applied to obtain the optimal factor settings. The experimental results showed that the tensile strength of specimens of the optimal parameters via the proposed approach is better than apply Taguchi method only.