A Machine Learning based Fault Identification Framework for Smart Grid Automation

Abstract

A smart grid is an automated electric grid that frequently monitors the working of a power system to control it and one of the most essential parts of a smart grid is a transmission line, which is used to carry a large amount of generated power in the power system. However, due to their exposure to the environment, these lines may experience a flow of anomalous electric current or faulty current, which can disrupt the regular operation of the power system and cause equipment failure. using machine learning, this research provides a novel automated framework for identifying which type of fault is occurring in the system without having to visit the actual fault location. With an area under the curve (AUC) score of 99.95 % and an accuracy of 99.15 %, the suggested model combines quadratic discriminant analysis coupled with pre-processing techniques like feature engineering to detect if the system has defects and the type of faults. The suggested model generates all of these results in less than 0.015 seconds. Knowing which type of problem is occurring in the power system using voltage and current data can improve power system cost savings by lowering the use of relays and creating effective ways to automate fault handling for smart grids.