Improvement of power system transient stability using Wavelet Neural Networks and FACTS devices

In this paper we improve transient stability of power system by using the Wavelet Neural Network or “Wavenet” as a controller for this system. This wavenet controller is used as a FACTS device in this power system. The control signal of this controller is current or voltage that is injected to system to improve the transient stability. One of ordinary neural networks disadvantages is the need of off-line training for initializing their weights. Since the power system in transient mode is a non-linear system, its off-line training is very difficult. In this paper because of eliminating the off-line training, this problem will not exist. The robustness of controller will be investigated by additional fault after the main fault. Our goal is to control the rotor acceleration by means of reactive power injection by FACTS devices.