A new method for flicker severity forecast

Precisely forecasting the flicker level is important for drastic voltage fluctuations associated with the rapid reactive power consumptions of electric arc furnace (EAF) loads. This paper presents a prediction model based on grey theory combined with radial basis function neural network (RBFNN) for the forecast of flicker severity caused by the operation of a dc and an ac EAF loads, respectively. Test results based on the proposed model are compared with two other neural network methods. It shows that more accurate forecast is achieved for the flicker prediction based on the proposed method.