A wavelet feature-based neural network approach to estimate electrical arc characteristics

Electric Arc Furnaces (EAFs) account for almost half of the North American steel production. Arc furnaces draw high and dynamic electrical power to melt scrap metal loads and as the result, they are highly non-linear and time-varying in nature. Because of the power system problems associated with EAFs, there is an ongoing need to accurately model EAFs to properly assess the impact of these types of nonlinear loads. This paper proposes a wavelet feature-based neural network approach for modeling the dynamic voltage-current (v-i) characteristic of the EAF. This method uses the data from the operational electric arc furnaces to describe the underlying processes, and unlike conventional mathematical techniques does not rely on presumed model structures, or simplified assumptions. The simulation results obtained by using the proposed method are compared with the actual measured data, and it is shown that the proposed method accurately models the EAF dynamic v-i behavior.