Evaluating and forecasting stability across different conditions is essential since smart grid stabilization is among the most significant characteristics that could be employed to assess the functionality of smart grid design. Some intelligent methods to foresee stability are required to mitigate unintended instability in a smart grid design. This is due to the rise in domestic and commercial constructions and the incorporation of green energy into smart grids. It is currently hard to forecast the stability of the smart grid. In this framework, a smart grid with reliable mechanisms is being implemented to meet the fluctuating energy demands as well as providing more availability. The involvement of consumers and producers is one of the many factors influencing the grid's stability. This study suggested a novel approach for locating stability statistics in smart grid systems utilizing machine learning frameworks was presented. This paper outlined the multi-layer perceptron-Extreme Learning Machine (MLP-ELM) methodology to predict the sustainability of the smart grid. Additionally, this utilized the principal component analysis (PCA) approach for extracting features. In addition to an empirical assessment and a comparison to various approaches, this article presents an implementation result for smart grid stability. Simulation findings demonstrate that the suggested MLP-ELM approach outperforms traditional machine learning techniques, with accuracy reaching up to 95.8%, precision at 90%, recall at 88%, and F-measure at 89%.