Application of multiobjective optimization and neural network techniques to process design

A major problem in product development is the selection of a set of conditions (parameters) which will result in a product with desirable performance. This problem is even more significant when optimizing multiple responses under a common set of constraints. This paper addresses the application of multiobjective optimization (MOP) techniques to process optimization where processes are represented using regression or neural network (NN) models. The application of MOP using regression and NN process modeling techniques is demonstrated through two examples.