A procedure for training an artificial neural network with application to tool wear monitoring

An artificial neural network (ANN) has been used for on-line tool wear monitoring. Training the ANN for tool wear data has been done by reducing the dimension of the input of the training pattern from six dimensions to two dimensions. Reduction of the input pattern from six dimensions to two dimensions is done by using an optimal discriminant plane technique. Two projection vectors phi(1), phi(2) are calculated for reducing the dimension of the input pattern. During training and testing of the ANN, the number of nodes in the input layer is two. Thirty patterns for training and 83 patterns for testing the ANN are used. Results of the ANN trained without reducing the dimensions of the input patterns and with reduced dimensions of the input patterns are compared.