Application of radial basis function neural networks in optimization of hard turning of AISI D2 cold-worked tool steel with a ceramic tool

In this work, the optimization of a finish hard turning process for the machining of D2 steel with ceramic tools is carried out. With the help of replicate experimental data at 27 different cutting conditions, radial basis function neural network models are fitted for predicting the surface roughness and tool wear as functions of cutting speed, feed, and machining time. A novel method for neural network training is proposed. The trained neural network models are used as a black box in the optimization routine. Two types of optimization goal are considered in this work: minimization of production time and minimization of the cost of machining. One novel feature of this work is that the surface roughness is considered in the tool life instead of as a constraint. This is possible owing to the availability of the relationship of surface roughness with time in the neural network model. The results of optimization will be dependent on the tool change time and the ratio of operating cost to tool change cost. The results have been presented for the possible ranges of these parameters. This will help to choose the appropriate process parameters for different situations, and a sensitivity analysis can be easily carried out.