A Comprehensive Review of Harmonic Issues and Estimation Techniques in Power System Networks Based on Traditional and Artificial Intelligence/Machine Learning

Abstract

Modern industrial and commercial devices that are fed by power electronics circuits and behave text non-linearly tend to produce power quality issues in power systems including harmonics and interharmonics, swell, flicker, spikes, notches, and transient instabilities. Among them, harmonic emission is the most significant challenge to be overcome by the distribution networks. Unwanted current, overheating motors and transformers, equipment failure, and circuit breaker misoperation are some of the harmonic consequences. While it is important to employ the best methods to mitigate or suppress the harmonic distortions in power systems, it is even more essential to estimate these harmonics at the outset by developing smart, efficient, and accurate techniques. Due to their capability for learning, predicting, and identifying, researchers have turned to Artificial Intelligence technologies for harmonic estimation in distribution networks. Although the power system parameters (impedance/admittance model) and many harmonic monitors are prerequisites for traditional harmonic estimation methods, by utilizing Artificial Intelligence, these requirements are minimized. In this paper, a comprehensive review of traditional and modern (smart) harmonic estimation techniques are discussed.

Keywords

• Power quality
• harmonic distortion
• artificial intelligence
• machine learning
• neural network
• power systems