Future distribution systems are faced with more challenges on islanding detection due to the increasing penetration level of inverter-based distributed generators (DGs). Different DG technologies, inverter control as well as… Click to show full abstract
Future distribution systems are faced with more challenges on islanding detection due to the increasing penetration level of inverter-based distributed generators (DGs). Different DG technologies, inverter control as well as other advanced inverter functions, such as fault ride through are challenging the capability of islanding detection schemes. On the other hand, for multiple feeders network, topological change of feeders and islanding at adjacent feeder increase the vulnerability of islanding detection devices. Furthermore, available islanding detection schemes are suffering from notable non-detection zones (NDZs) under reduced power mismatches. Therefore, to mitigate these issues, this paper proposes an effective methodology for building decision trees-based intelligent relay (IR). This methodology utilizes the NDZ boundaries of existing standard relays and applies a comprehensive training/testing strategy, which effectively reduces the NDZ while maintaining a superior dependability and security performance. To validate the applicability of the proposed methodology, the hardware-in-the-loop simulations are realized by programming the generated IR logics into a real commercial relay.
               
Click one of the above tabs to view related content.