Modelling and predicting surface finish in turning using an adaptive neuro-fuzzy inference system

Abstract

Surface finish is a key factor in the machining process, and it is used to evaluate and determine the quality of a product. Therefore, modelling and predicting of surface finish of a workpiece in turning play an important role in manufacturing industry since turning is most common machining operation. This paper illustrates the application of adaptive neuro-fuzzy inference system ANFIS for modelling and predicting the surface finish in turning operation for set of given cutting parameters, namely spindle speed, feed rate and depth of cut. Three different membership functions, triangular, trapezoidal and generalised bell shaped, were adopted during the hybrid-training process i.e., combination of backpopagation gradient descent method and least square method of ANFIS in order to compare the prediction accuracy of surface finish by the three membership functions. The predicted surface finish values obtained from ANFIS were compared with experimental data. The comparison indicates that the adoption of triangular, trapezoidal and generalised bell shaped membership functions in proposed system achieved satisfactory accuracy. The generalised bell-shaped membership function in ANFIS achieves slightly higher prediction accuracy than other membership functions.