Chapter 14 - Supervised Learning-Based Fault Location in Power Grids

Abstract

In modern societies and with the introduction of “smart power grids,” customers are more sensitive to power outages. There are also more complex power transmission configurations to integrate renewable energy-based power generation at remote locations. Therefore, more efficient and accurate methods of fault location along these complex configurations are required, which target improving power supply restoration process, reducing the overall power outages time and costs, and enhancing end-users satisfaction. The availability of high-resolution/high-volume data, due to the proliferation of intelligent electronic devices in smart grids, paves ground to implement more accurate and intelligent fault location methods. This chapter presents a supervised-learning fault location method for complex power transmission lines by using high-resolution voltage and current measurements data. The fault location methods are developed for two complex high-voltage AC transmission systems, (1) three-terminal transmission lines, (2) hybrid transmission lines. The presented methodologies utilize discrete wavelet transform and support vector machine (SVM) as a supervised learning algorithm where the power system operating and fault conditions are taken into account through the learning steps of the SVM classifiers.