Future-ready smart grids harnessing predictive analytics and optimization for sustainable energy

Due to increasing energy demands around the world and rising environmental concerns, there is a call for the addition of renewable resources to the grid. In general, smart grids allow for real-time monitoring of the power system, but do not have the ability to forecast and predict generation from variable solar or wind generation and effectively allocate the optimal energy mix. Literature tends to focus on either machine learning (ML) forecasting or linear programming (LP) allocation, but lacks an integrated article. This paper integrates ML models (Random Forest Regressor (RF), Support Vector Regression (SVR), and Multilayer Perceptron (MLP)) for forecasting unknown generation into LP for allocation. The RF model outperformed the others, achieving an RMSE of 0.0110 and an R² of 0.9998. The LP optimization successfully allocated energy to meet demand while maximizing the renewable use. Results from the tests indicated that the optimized grid could meet approximately 100% of demand with renewables in the best conditions, making significant advances to sustainability despite limitations in computing power.