This article presents an effective means of applying the neural network to optimize a multiresponse system. Two neural networks are constructed. The first network is utilized to determine the optimal factor combinations, and the second one is used to predict the values of multiple responses. Two numerical examples, attributed separately to conventional experiments and to Taguchi's orthogonal array, demonstrate the effectiveness of the proposed approach. Analysis results indicate that the neural network is a very effective and flexible means to identify the real optimal parameter combination-for a multiresponse system.
- Artificial neural network
- Experimental design