Neural identification of SSSC based on average model using GAMMA, DNN, RBF and MLP for steady state calculations

The use of exact model of FACTS devices in steady state calculations is complex, due to their switching behavior. However, applying very simple models such as pure inductor/capacitor for FACTS devices leads to inaccurate results in power system studies. In addition in a power market the use of inaccurate models for power system components affects the electrical energy pricing system; while this is very crucial when FACTS devices are used for congestion management of the transmission systems. Average technique provides an appropriate time-domain representation of FACTS devices in which high frequency switching ripples are vanished. But average model can not be directly applied to the power system steady states. Thus this paper extends average model and presents an average-neural model of SSSC as a series FACTS device, which is well-suited for analytical purposes in power system applications. To this extend, design and development of four neural identifiers are performed using the GAMMA, DNN, RBF and MLP. To verify the developed models, the exact solutions obtained from the average model of SSSC are compared with the outcomes of the identifiers.