Using Machine Learning to Predict Heat Affected Zone in Steel Induction Heating

Abstract

The uncertain relationship between the size of a steel plate's heat-affected zone and the heating parameters related during the heating process in induction heating means that this causes a different deformation of the steel plate. The electromagnetic induction approach has been used to control plate deformation in the steel manufacturing ship hull process instead of torch heating. This research investigates the heat flux distribution over a steel plate during the induction process numerically first, with the assumption that the process is quasi-stationary around the inductor and that the heat flux is significantly dependent on the temperature of the workpiece. Following the analysis of the heat flux, a commercial FEM program with heating settings is used to perform thermo-mechanical and heat flow studies on the plate. The heat-affected zone's size and after that, the supplied heating parameters are synthesized using machine learning to develop simplified regression that can quickly reveal the relationship between the heating parameters and the heat-affected zone's size. The formula findings are quite similar to the simulation results.