Ensuring the stability of modern power grids is paramount to achieving sustainable energy systems, particularly as renewable energy sources become more integrated. This study focuses on predicting smart grid stability using machine learning (ML), deep learning (DL) algorithms, and Explainable AI (XAI) methods to ensure model interpretability. Among the ML models, SVM outperformed others with an accuracy of 97.8%, while Multilayer Perceptron achieved the highest accuracy of 97.7% among DL models. Explainable AI techniques, particularly SHAP was employed to interpret model predictions, revealing key factors such power load (g3) and reaction times (tau1) significantly influence grid stability. Our approach ensures the grid’s stability by addressing fluctuations and system imbalances in real time. The study’s findings offer practical guidance for enhancing power grid performance, which advances the more general objectives of energy resilience and efficiency in the incorporation of renewable energy sources.