Multiple distributed generators islanding detection using GBDT-JS techniques

Abstract

Photovoltaic arrays and wind Distributed Generators (DG) have become integral to our renewable energy landscape. However, when a DG disconnects from the grid, the risk of islanding arises, necessitating detection within the stringent of two-second timeframe mandated by IEEE standards. Machine-level algorithms play a pivotal role in enhancing power system reliability and safety by swiftly identifying and isolating isolated segments, thereby preventing potential hazards and ensuring efficient grid operation. This study introduces an algorithm-based islanding detection approach for distributed generating systems employing both Solar Photo Voltaic (SPV) and wind systems. The GBDT-JS algorithm, a combination of Gradient Boosting Decision Trees and Jelly Fish Techniques, emerges as an intelligent solution. The focus of this technique is based on the Rate of Change in Phase Angle (RCPP) at the target DG position, offering a characterized approach to islanding detection. In addressing the major difficulties and challenges, the GBDT-JS method proves instrumental in categorizing islanding situations and grid disturbances. This classification aids in determining the system’s adoptability based on various loading and switching capabilities. The achievements in overcoming these challenges lie in the algorithm’s ability to provide a comprehensive solution, ensuring the reliability and safety of distributed generating systems.