Minimizing Hybrid Electric Power Generation Costs Using Smart Power Grid Solution

Increasing global electricity demand, coupled with resource constraints, necessitates strategic generation planning to optimize costs and curb emissions. This study leverages California's hourly electric generation data (renewable and non-renewable), demand consumption data, and cost of generation to develop Machine Learning models for forecasting electricity demand, resulting in mitigating costs and emissions. Moreover, we propose an innovative grid system, including trigger mechanisms, to redirect forecasted surplus electricity, back into methods of electric generation, minimizing operating costs while satisfying demand. The Random Forest Regressor model outperformed other models used to forecast electric demand. It achieved a Mean Absolute Percentage Error (MAPE) of 2.9%, Root Mean Squared Error (RMSE) of 1087.9, and R-squared (R2) of 95.5%. Using the forecasted demand data and leveraging our hybrid energy grid system, we applied a mathematical algorithm that resulted in significant cost savings and emission reductions during 8-hours of electricity overproduction in California during summer. By strategically redirecting excess electricity back into methods of generation, it facilitates savings of up to $525,596.68 in a single day and avoids emissions equivalent to 5252.38 tCO2e compared to traditional grid methods. The results achieved through our proposed methods offer a promising approach for optimizing generation costs and promoting sustainable electricity production.