Neural-net based control structure with FACTS devices

This paper deals mainly with the load side voltage stability. A neural-network based dynamic load model is incorporated into voltage stability analysis. FACTS (flexible AC transmission systems) devices are applied for power system stability enhancement. The use of dynamic load model and FACTS devices for control may sometimes lead to excitation of generator dynamics, resulting in the whole power system more complex. Conventional methods often neglect either the load dynamics or generator dynamics, while the proposed methods deal with both. For the convenience of control design, proper system models are developed. Methods are presented for thyristor controlled series capacitor and static var compensator control cases to represent the controlled system through three sets of equations: generator dynamics, load dynamics and control constraints. The control is then synthesized in the form of neural networks, trained by using the pre-specified optimal trajectories.