A hybrid method for islanding stability detection of distributed generators using wavelet transform and artificial neural networks

Although in many cases islanding is considered an undesired situation, the intentional islanding can reduce the interruption indices of the distribution power systems. However the distributed synchronous generator must mantain both voltage and frequency in the operational limits of the islanded system. This study proposes the use of artificial neural networks and Wavelet Transform to detect stable islanding. Then, the distributed generator keeps powering the loads in the island. Otherwise the passive anti-islanding protection (frequency or voltage relays) disconnect the distributed generator to prevent any damage to the loads. The islanding stability detection method proposed in this paper perform high success rates on identifying the system's stability, even in conditions of white noise in the voltage and current measurements.