Adaptive Neuro-Fuzzy Inference System for Prediction of Surface Roughness Under Biodegradable Nano-lubricant

Machining processes involve many nonlinear parameters which make them complex. Setting of machines is generally relying on decision-making skills based on intuition and common sense learned through experience. In this work, five variables namely the spindle speed, the feed rate, the length of the cut, the depth of cut and the helix angle were considered to predict the surface roughness during the end-milling machining of AA8112 alloy. The adaptive neuro-fuzzy inference system (ANFIS) is proposed in this study. A dataset made of 50 data was used. Each data corresponds to a specific configuration of the end-milling machine and the corresponding surface roughness. In order to assess the prediction performance of ANFIS, various types of membership functions. This includes π-shaped (PIMF), generalized bell shape (GBELLMF), triangular shape (TRIMF), trapezoidal shape (TRAPMF), and Gaussian curve (GAUSSMF). This study shows that the various outputs track the targets effectively irrespective of the membership functions adopted the deviations between the predicted results and the targets were within 7%. This study demonstrates the potential of ANFIS models for the prediction of surface roughness.