Prediction of Machinability characteristics of Ti6Al4V alloy using Neural Networks and Neuro-Fuzzy techniques

Abstract

Titanium alloys are familiar for their capability to withstand extreme environmental conditions. However, high fabrication and machinability cost of titanium limits their applications to defense and aerospace sectors. In order to enhance their domestic applicability, either their fabrication or machining costs have to be controlled. With an intention to reduce machinability cost; tool wear and surface roughness of Ti6Al4V titanium alloy are predicted through different computational techniques like Neural Network and fuzzy logics in this investigation. A full factorial design set of experiments are conducted on CNC turning machine to envisage the clout of input parameters like cutting speed, cutting feed, cutting time. The output responses emphasized are the tool wear of carbide tool inserts with coated TiN layer and surface roughness of Titanium Alloy. Neural network and fuzzy logic models are generated by training on this data to prediction output responses. The models are further validated by addition 8 different experimental data sets. This revealed the superiority of ANFIS over neural network with good prediction accuracy.