Intelligent Expert System for Power Quality Improvement Under Distorted and Unbalanced Conditions in Three-Phase AC Microgrids

Abstract

This paper presents an expert system (ES) based on decoupled power/current decomposition and the k-nearest neighbor pattern recognition method to identify and choose the correct mitigation solution for power quality improvement in three-phase ac microgrids under non-sinusoidal current and voltage operations. By using power/current terms, load conformity factors and a k-nearest neighbor classifier, the proposed ES achieved 99.98% classification accuracy. Simulation studies were carried out in a PSCAD/EMTDC environment, where the IEEE 13-bus feeder test system was in a grid connected microgrid mode. The obtained results indicate that the proposed ES is robust and able to easily select an appropriate/adequate compensation solution.