A NEURAL-NETWORK SYSTEM FOR PREDICTING MACHINING BEHAVIOR

This paper describes a case study the aim of which was to apply neural-network technology to the selection of machining parameters and to the prediction of machining performance in metal cutting. The project involved the development of a back-propagation neural network using Neuralwork Professional II, a multi-paradigm, prototyping and development system. The network was trained using data from the Machining Data Handbook, after training the network being able to select appropriate machining conditions and to predict cutting forces and surface finish for a given work material. The results so obtained are analysed and discussed. A novel feature of this work is the development of an on-line implementation of the trained network using the C programming language. Current limitations of neural-network technology with respect to engineering application are discussed also.