Object Detection Based Power Quality Expert System for an Electric Vehicle Infrastructure

Abstract

The shift in power systems from large power stations towards smaller decentralized generation units imposes new challenges on grid operators regarding power quality. The steady increase of power electronics connected to the grid results in elevated harmonic distortion levels, inducing additional losses and causing malfunctioning of control devices. This development has led to a rising number of monitoring systems, in order to ensure adequate power quality within the mandatory limits and boundaries. Consequently, detection and classification of power quality disturbances is a vital component of the mentioned systems. This paper proposes a new method for detecting power quality issues and identifying their underlying causes, based on historical data. The implemented automated image classification algorithm continuously analyses the FFT spectrogram of disturbances and simultaneously determines the type of disturbance and its cause without the need of a segmentation and feature extraction process. The results are validated using measurement from an electric vehicle car charging station.