Real-Time Implementation of Optimized Power Quality Events Classifier

Abstract

The future's smart grid consists of an increasing number of dispersed energy generation and consumption devices. Due to its complexity, the energy sector yields the development of decentralized monitoring and control framework, where each node in the grid will be a potential location for power quality devices. In order to continually improve the stability of the energy system, those devices need a tool for accurate and real-time detection and classification of the power quality disturbances. However, despite the continuous progress in the field, the development of that kind of tool is still a challenge. This article presents a real-time implementation of an optimized power quality events classifier for the detection and classification of 21 classes of single and combined disturbances. The focal point of the presented classifier is the real-time implementation of optimal feature extraction, optimized classification, accurate zero-crossing detection, and efficient handling of different noise levels present in the voltage signal. The implementation is performed on myRIO-1900 using the LabVIEW interface. Testing and validation results show that this implementation exhibits high classification accuracy, even in cases when the classification relies on classes obtained as the combination of four power quality disturbances.