Machining Performance Analysis in End Milling: Predicting Using ANN and a Comparative Optimisation Study of ANN/BB-BC and ANN/PSO

Abstract

End milling machining performance indicators such as surface roughness, tool wear and machining time are the principally indicators of machine tool industrial productivity, cost and competitiveness. Since accurate predictions and optimisations are necessary for control purposes, new merit-driven approaches are for good results. The aim of this work is two folds: prediction of machining performance for surface roughness, tool wear and machining time with ANN and the optimisation of these performance indicators using the combined models of ANN-BB-BC and ANN-PSO. However, the optimisation platform is hinged on the fuzzy goal programming model, which facilitates comparisons between the performance of the BB-BC and the PSO algorithms. To demonstrate the approach, optimal tool wear and surface roughness were obtained from a fuzzy goal programme, then converted to a bi-objective non-linear programming model, and solved with the BB-BC and the PSO algorithms. The outputs of the artificial neural network (ANN) were integrated with the optimisation models. The effectiveness of the method was ascertained using extensive literature data. Thus, prediction and optimisation of complex end milling parameters was attained using appropriate selection of parameters with high quality outputs, enhanced by precise prediction and optimisation tools in this proposed approach.