Forward and reverse mapping of multiresponses turning process using adaptive network-based fuzzy inference system and simulated annealing algorithm

Abstract

Input–output relationships of a turning process have been established both in forward as well as reverse directions using adaptive network-based fuzzy inference system. Four input parameters, namely insert type, lubrication strategy, feed rate, and cutting velocity and two outputs, namely cutting force and surface roughness have been considered for the aforementioned mappings. Training and testing of the network in the forward direction were adopted by use of experimental data, which derived from high-speed turning of Monel K500 super alloy. For multiattributes reverse mapping problem, grey relational grade was firstly used to convert the cutting force and surface roughness in a single attribute problem. Then the reverse mapping was performed by the use of simulated annealing algorithm by minimizing the absolute difference between grey relational grade of specified cutting force and surface roughness and adaptive network-based fuzzy inference system model of grey relational grade, which was derived from the forward mapping. The confirmation was performed in eight benchmark tests. Results indicated that the proposed methodology can predict the input–output relationship of high-speed turning process in both forward and backward directions with error goal below 8%. The developed model was further used to find the parametric influence of the process factors on the cutting force and surface roughness.