Response analysis of an industrial power system with arc furnaces utilising artificial intelligence

The growing popularity of the electrical arc furnace in metallurgical industries causes significant impact on power grids, transient stability and quality of power supply. Being non-linear in nature, arc furnaces produce harmonics and inter-harmonics of arc voltage and current in electrical networks. Moreover, due to random load variation, the phenomenon of voltage fluctuation is found during arc furnace operation. Therefore, it is important to model the random behaviour of the arc furnace. The aims of this thesis are to investigate the applicability of the conventional and \black box\" modelling techniques in modelling the response of an arc furnace. Conventional modelling approaches are used to obtain the instantaneous voltage and current wave forms of the arc furnace while black box modelling approaches such as artificial neural networks (ANN) and adaptive neurofuzzy inference system (ANFIS) are used to capture the random and time varying pattern of arc furnace power consumption which is due to random changes of the bus voltage. The ANFIS model is also used for predicting the bus voltage of the arc furnace. To evaluate the performance of the proposed models several case studies are presented and outputs of the proposed models are compared with the actual data recorded on a metallurgical plant."