A soft computing-based study on WEDM optimization in processing Inconel 625
Tatjana V. Sibalija,Sandeep Kumar,G C Manjunath Patel,Jagadish
DOI: https://doi.org/10.1007/s00521-021-05844-8
2021-03-22
Neural Computing and Applications
Abstract:This work presents an examination of the wire electric discharge machining (WEDM) in processing Inconel 625. The major WEDM variables (pulse-on time, pulse-off time, servo voltage, wire feed rate) were experimentally investigated to address multiple aspects of this process, namely to reduce the gap current and surface roughness and increase the cutting speed. After testing the statistical significance of the major WEDM parameters for the three responses, an advanced statistical procedure was utilized for tackling correlations among outputs and their integration into the WEDM output performance in a fully objective manner. The Bayesian regularized neural network established a highly accurate process model that was engaged as the objective function for the evolutionary algorithms to identify the optimal machining conditions. The following algorithms were employed: particle swarm optimization, teaching-learning based optimization, grey wolf optimization and Jaya algorithm. Their results were thoroughly analyzed in terms of accuracy, i.e., repeatability of the resulting solutions, convergence speed and computational time, including assessment of the hyperparameter effects. The obtained optimal solution was highly convincing and it was successfully validated in a confirmation run. Therefore, the benefits of these findings are twofold, offering: (i) a thorough analysis of the four metaheuristics effectiveness in dealing with a real industrial problem, (ii) useful insights for controlling WEDM variables to enhance the technological, environmental and economic aspects at the same time.
computer science, artificial intelligence