Machinability data representation with artificial neural network

Machinability data selection is a crucial step in a manufacturing environment. It plays an important role in efficient utilization of machine tools and significantly influences the overall manufacturing cost. This paper is devoted to studying the feasibility of using neural network in representing machinability data. The feed-forward neural network is used to predict optimum machining parameters under different machining conditions. The back propagation (BP) method is used to optimize the network component representation. An object-oriented neural network-handling library is developed and implemented in the turning process. The authors introduce a new type of artificial neuron in the design of neural network for turning process, namely the Product neuron, which has multiplication instead of summation. The characteristics and applications of the new neuron are explained in this paper. The proposed network unveils the possibility of developing an expert system for machinability data selection based on neural networks. Comparisons with fuzzy logic representation in the literatures are made. (C) 2003 Published by Elsevier Science B.V.