| The research presented in this dissertation concerns the application of artificial neural networks to the Taguchi type experimental design for robust product or process design.;In Quality Engineering, and in most experimental design cases, understanding the relationship between design factors and product or process performance is essential for improving quality. Because most manufacturing conditions are so complicated, a large number of experiments are generally required. Taguchi's approach can systematically reduce the number of experiments required. But in certain cases, where interaction exist between design factors, the level average analysis will not select optimum condition. To overcome this weakness, after orthogonal array experiments are performed, an artificial neural network is used to select the optimum conditions instead of traditional level average analysis.;Actual applications, design of a Wheatstone Bridge, design of a four-bar linkage and design of an ultrasonic plastic welding process, proved that this approach can select real optimum points both within and between factor levels. |
