An intelligent multi-objective framework for optimizing friction-stir welding process parameters

Abstract

The comprehensive intention of this paper is to evaluate the optimal welding parameters for joining two dissimilar materials by friction stir welding (FSW) process which is termed as a green manufacturing technology in order to generate quality joints. Conventionally, the optimization of process parameters experimentally is carried out by a time-consuming trial and error technique. Also, the effect of two or more parameters cannot be considered at the same time experimentally. Due to this, mathematical modelling is carried out to determine the optimal welding parameters. This paper focuses on a theory that hybridizes the exploring capability of non-dominated sorting genetic algorithm-II (NSGA-II) and exploitation capability of Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS) method. The performance parameters of FSW process has a high level of nonlinearity due to which artificial neural network (ANN) is employed to mathematically model the performance measures. Full factorial design is employed to plan the experimentations. At random, ten dataset were used for testing the ANN structures and the remaining data were used to train the network for predicting the ultimate tensile strength (UTS), hardness and impact energy. The optimal networks had root mean square error (RMSE) and mean absolute error (MAE) of 0.7486 and 0.0074, 0.4045 and 0.003 and 0.1866 and 0.0354 respectively. The transfer equations developed from the ANN models is used as the fitness function for the NSGA-II algorithm. The optimal welding parameters obtained from the proposed hybrid algorithm are Tool rotation speed = 1693rpm, Traverse speed= 2.72 mm/s and Copper as advancing side material. Validation of the optimal result is done by carrying out experiments that are conducted at the simulated optimal parameter. Finally, a macro and microstructural study of the welded joint at the simulated optimal parameter is carried out in order to assess the behaviour of the weld.