A comparative analysis of tool wear prediction using response surface methodology and artificial neural networks

Abstract

The research reported herein is to study the comparison between a response surface methodology (RSM) and artifi cial neural network (ANN) in the modelling and prediction of tool wear during face milling of hybrid composites. Aiming to achieve this goal, several milling experiments were performed with polycrystalline diamond (PCD) inserts at different machining parameters namely feed rate, cutting speed, depth of cut and weight fraction of alumina (Al 2 O 3 ). The experiment was carried out using 6061 aluminium alloy reinforced with alumina of size 65 m and graphite of size 60 m particles which are prepared using stir casting method. Mathematical model is created using central composite face centred second-order RSM and the adequacy of the model was verifi ed using analysis of variance. With regard to the machining test, it was observed that feed rate is the dominant parameter that affects tool wear of PCD inserts. The comparison results show that models provide accurate prediction of tool wear in which ANN perform better than RSM. The data predicted from ANN is very nearer to experimental results compared to RSM, therefore we can use this ANN model to determine the tool wear for various composites and also for various machining parameters.