Cutting force prediction in high-speed turning of Ti-6Al-4V alloy using RSM, ANN and ANFIS.

In modern engineering field, data modelling and prediction capability of different techniques are to be evaluated in order to find the most accurate technique. In this work, the prediction capability of RSM (Response Surface Methodology), ANN (Artificial Neural Networks), ANFIS (Adaptive Neuro Fuzzy Inference System) have been compared in case of dry machining of Ti-6Al-4V Alloy. Ti-6Al-4V being one of the difficult to machine material, the machining of such alloys becomes necessary for the aerospace/manufacturing industries. Total 15 experimental runs were finalized and performed by using Response Surface Methodology (Box-Behnken Design). The machining was carried out using carbide inserts and the output cutting forces are measured using Kistler Dynamometer. The above mentioned three techniques have been employed in prediction of cutting force and all had a good predicting capability. However, ANN showed good results than RSM and ANFIS technique. Percentage of error in prediction for RSM, ANFIS and ANN are 10.10, 4.56 and 2.56 respectively. The R² values of RSM, ANFIS and ANN are 93.86, 95.96, 97.91 respectively. The performance of ANN also depends on the algorithm and parameters used for training. [ABSTRACT FROM AUTHOR]