Machine Learning -Based Data-Driven Fault Classification on Electric Power Transmission System

Abstract

This paper proposes a novel scheme to improve the performance metrics of machine learning (ML) models for the classification of the short circuit fault (SCF) on electric power transmission line (TL) by enhancing the quantity and quality of dataset. The dataset consists of 24 features and 55289 observations, where each observation represents the maximum and minimum values of the signals during the fault and post fault conditions. A comparative analysis is conducted against several ML to showcase the efficacy of proposed dataset. The ability to classify the faults using proposed dataset is also compared with Phasor Measurement Unit (PMU) based dataset on Kundur two-area four-Machine Power System. The various simulations, dataset collection and ML algorithm have been performed in MATLAB environment. The performance metrics such as accuracy, precision, recall and F1-score and training time of various ML algorithm trained by proposed dataset is much superior than PMU based dataset for the classification of short circuit fault on Transmission Line.