Predicting Transient Stability of Power Systems Using Machine Learning: A Case Study on the IEEE New England 39-Bus Test System

The need to evaluate the transient stability of power systems is inevitable and crucial in order to ensure that they will continue to operate efficiently after interruptions. In the present investigation, an attempt is made to use machine learning methods, particularly the XGBoost and the Random Forest models, with the objective of predicting the stability of the power systems after a fault has occurred. Thus, the dataset used by the models has several generator and bus parameters, as well as pre and post-fault conditions; the objective is to identify if the system stability is stable or unstable. In general, it is possible to conclude that the use of a hybrid model, combining the XGBoost and Random Forest techniques, outperforms each model separately. This is the case because it has the merits of both methods as it combines the two methods to identify the similarities of the two plans. In this case, the effectiveness of the proposed approach is assessed using evaluation parameters like accuracy, precision, recall rate, and F1-score. Furthermore, the study gives an understanding of the stability measures most affected by the characteristics and those that affect stability predictions. By applying more complex modes of predictive modeling, this work contributes to advancing the reliability and efficacy of power grid management.