Abstract: This paper discusses a interclass mapping network system IMNS using in machine design. IMNS consists of back propagation network (BPN) and recurrent network (RN). BPN is trained to form the corresponding class boundaries and obtaind the weights. In RN, using the weights tell from well-designed, and correct poor designed to well designed. As a example, in machine parts design, using IMNS can categorize if the design parameter is correction, if not, it will correct automatically the ill parameter to reasonable parameter. The study show that the IMN system will have application in automatic machine design.

Key words: Artificial neural network, Backpropagation network, Interclass mapping network system, Machine design, Recurrent network