Multiresponse optimization of friction stir welding process parameters by an integrated WPCA-ANN-PSO approach

Abstract

Majority of conventional optimization techniques found in the literature deals with single response optimization problems. However, real life industrial processes are having multiple responses, which are conflicting with each other i.e, if we try to optimize one response we may have to compromise with the other. An attempt to optimize all the responses simultaneously to get all response in their acceptable range is called as Multiresponse Optimization (MRO) problem. This paper proposes an integrated WPCA-ANN-PSO approach to perform MRO of Friction Stir Welding (FSW) processes. It is a three stage process, in the first stage the multiple responses are converted into a single multiperformance index (MPI) using Taguchi-Weighted Principal Component Analysis (WPCA) approach. In the second stage an Artificial Neural Network (ANN) model is developed which is capable of predicting MPI for given set of control factors. Finally, in the third stage Particle Swarm Optimization (PSO) algorithm is used to search the global optimal solution using the developed ANN model. Further, to demonstrate the effectiveness of the proposed approach a case study on MRO of FSW process was carried-out to join two plates of AA2024-T4 aluminum using tool of HS steel of grade 304. The control factors considered were D/d ratio, tool rotation speed (TRS) and weld speed (WS). The responses considered were ultimate tensile strength (UTS) and hardness of the joint. The experiments were planned as per Taguchi orthogonal array L8. By WPCA approach the optimal parameters were found to be D/d ratio of 3, TRS of 670 rpm and WS of 17. This solution was further improved by WPCA-ANN-PSO approach and validated by confirmation experiments successfully.