Production of electric steel

Abstract

The artificial neural network (ANN) can be used as a model for friction stir welding (FSW), predicting the correlation between FSW parameters and weld properties. ANN for predicting properties of welded plates by FSW is explained and pervious works in this field are reviewed. Furthermore, some examples about applications of ANN in FSW are presented. The development of sound joints between materials is of great importance in FSW. One of the most challenging problems is choosing appropriate welding parameters in order to produce sound joint. Traditionally, a time-consuming trial and error development was carried out to determine welding parameters. Additionally, an optimized welding parameters combination was not achieved, because welds can be fabricated with very different ideal welding parameters. Different optimization techniques can be employed to determine the optimized output parameters by specifying the relation between the input and output variables. Applications of optimization methods in FSW are explained, and basic principles of these methods, such as Taguchi, genetic optimization and multi-objective optimization methods, are described