Smart Distributed Generation Systems Using Improved Islanding Detection and Event Classification

摘要: Distributed Generation (DG) sources have become an integral part of modern decentralized power systems. However, the interconnection DG systems to grid can present several operational challenges. One such major challenge is islanding detection. Islanding occurs when a system disconnected from rest grid. serious safety hazards and therefore accurate fast detection technique mandated by standards as IEEE 1547 UL 1741. Conventional techniques passively monitor local parameters voltage frequency detect islanding. These large non-detection zones are prone nuisance tripping. Therefore, two improved computationally inexpensive passive for inverter-based were proposed. The ripple content in rate change amplitude waveforms using time domain-spectral analysis. proposed tested modeled according 929-2000 standard. Results indicated that both not only capable detecting islanding, but also able accurately distinguish between non-islanding events under wide range operating conditions. Furthermore, novel Smart which classify was This added intelligence has considerable impact on operation feature will help operator develop clear understanding requirements needed mitigate effects events. event classification been implemented artificial neural networks (ANN) with set input parameters. Five parallel ANNs designed majority vote final stage represent output. A total 310 cases generated test performance technique. classified within 10 cycles their occurrence 98% average accuracy. INDEX WORDS— Passive detection, grid, Anti-islanding, Voltage RMS, Rate frequency, generation, Event classification, Artificial Neural Networks SMART DISTRIBUTED GENERATION SYSTEMS USING IMPROVED ISLANDING DETECTION AND EVENT CLASSIFICATION