Performance Analysis of Machine Learning-Based Traditional and Ensemble Techniques for Smart Grid Stability Prediction

A smart grid is integral to the digitalized transformation of the electricity sector, employing self-sufficient systems that integrate information, telecommunication, and advanced power technologies. Artificial Intelligence (AI), particularly Machine Learning (ML), plays a crucial role in overcoming the limitations of traditional modeling techniques. ML enables intelligent decision-making in response to dynamic factors like changing customer energy demands or disruptions in power supply within smart grids. This paper compares Machine Learning-Based Traditional and Ensemble techniques for predicting smart grid stability, utilizing an augmented dataset from Kaggle. For optimization, various classifiers and Ensemble Techniques, including Bagging, Boosting, Stacking, and Voting, were implemented with hyperparameter tuning. The experimental study highlights that ML-based Ensemble Techniques, with optimized parameters, outperform individual traditional techniques, showcasing higher accuracy and overall superior performance. A comprehensive comparative analysis based on evaluation metrics such as accuracy, precision, recall, F1-score, and ROC emphasizes the potential of these techniques to enhance the efficiency of predicting smart grid stability.