A Comparative Study of Various Machine Learning Algorithms to Detect Power Plant Stability

Abstract

Smart grid’s concept has emerged in past decade as a key transition for the power system. The continuous integration of renewable energy sources and increasing demand affects the stability of the grid. A power grid that is not stable can lead to blackouts, brownouts, and other types of disruption. This can cause a great deal of inconvenience and financial losses, so it is important to ensure that the power grid stays stable. This study focuses on the identification of the smart grid stability by applying various machine learning algorithms i.e., Light Gradient Boosting Machine, Extra Trees Classifier, Extreme Gradient Boosting, CatBoost Classifier, Random Forest Classifier, Gradient Boosting Classifier, K Neighbors Classifier, Decision Tree Classifier, Naive Bayes, Logistic Regression all of which are evaluated using Stratified K-fold cross validation to assess their performance. Finally, this work compares the various machine learning models on the basis of their characteristics for better perspective. Of all the models tested CatBoost classifier had the best scores as it handles categorical variables effectively and works efficiently even on non-preprocessed data and provided an accuracy of 0.9506, Area under the curve (AUC) of 0.9913, recall value of 0.9725, precision value of 0.9512 and F1 score of 0.9617.