Soft sensor of bath temperature in an electric arc furnace based on a data-driven Takagi–Sugeno fuzzy model

Electric arc furnaces (EAFs) are intended for the recycling of steel scrap. One more important variables in process is tapping temperature steel. Due to nature process, continuous measurement melt complicated and requires sophisticated measuring equipment; therefore, most EAFs, separate samples taken several times before tapped, verify whether within prescribed range. The measurements obtained using disposable probes; when performed, furnace must be switched off, leading increased tap-to-tap time, unnecessary energy losses, consequently, lower efficiency. following paper presents a novel approach EAF bath estimation fuzzy model soft sensor Gustafson–Kessel input data clustering particle swarm optimization parameters. uses first as an initial condition, necessary inputs estimate continuously throughout refining stage process. results have shown that prediction accuracy proposed very high it fulfils required tolerance band. parallel implementation with aim achieving fewer measurements, shorter times, decreased losses. Furthermore, if information about accessible manner, operators can adjust control achieve optimal thus higher