Application of hierarchical neural-network-based control to transient stabilization of interconnected power systems

Flexible AC Transmission Systems (FACTS) research is aimed at increasing transmission capabilities of existing power systems. FACTS devices provide means to dynamically control parameters of the transmission lines. In particular thyristor-controlled series capacitors have been proposed to increase inter-area transient stability margins. Design of a suitable control algorithm for adjusting capacitor compensation level is a challenging task. The power system dynamics, particularly during transients caused by severe faults, are highly nonlinear. A series capacitor affects the system dynamics in nonlinear fashion, further increasing the difficulty of applying traditional linear control. There is also an inherent system uncertainty due to fluctuations of operating conditions and to sudden structural changes caused by faults. A successful controller should be robust with respect to these uncertainties and must accommodate the nonlinear dynamics of the system. The above issues are addressed here using neural-network methodology