Modelling and optimization of mean thickness of backward flow formed tubes using regression analysis, particle swarm optimization and neural network

Abstract

In this study, the relationship between the input process parameters and the output parameters of the backward metal flow forming process is established using a full factorial design of experiments. Both the linear and interaction terms of the process variables have been analyzed using statistical methods. Composite desirability and particle swarm optimization (PSO) is used to find an optimal set of process parameters for the desired responses. The composite desirability approach was found to provide more accurate results compared to the PSO. Neural network (NN) was used for modelling the flow forming process for the efficient prediction of output parameters. The performance of the NN-based predictive tool is better compared to regression analysis.