Predictive analytics for sustainable energy: An in-depth assessment of novel Stacking Regressor model in the off-grid hybrid renewable energy systems

The growing need for Hybrid Renewable Energy Systems (HRES), particularly for off-grid areas, has increased the need for efficient strategies to meet energy needs and minimize environmental impacts. Advanced predictive models are crucial for improving the performance of such systems to ensure their efficient and reliable operation. Nevertheless, despite considerable advances, there are imbalances in the precise prediction of renewable energy sources like Direct Normal Irradiance (DNI) and Wind Speed (WS) that are crucial in HRES optimization and design. Current approaches do not completely capture intricate and non-linear relationships among these parameters, which impose a limit on model precision and overall system efficiency. This study aims to overcome these shortcomings by suggesting an innovative stacking regression model. The novel model significantly improves the accuracy of DNI and WS forecasts with a coefficient of determination value of 0.990 in forecasting DNI and 0.972 in forecasting WS, surpassing traditional forecasting approaches. The accurate predictions provided by the stacking model were then used to optimize the proposed HRES, which consists of solar photovoltaic panels, wind turbines, a proton exchange membrane fuel cell, battery storage systems, hydrogen tanks, an electrolyzer, and a converter, delivering a sustainable and economically viable energy solution. The proposed HRES is applied to the Bayin Wusu community near Wuhai City, China. The optimized system has a Levelized Cost of Energy (LCOE) of 0.345 $/kWh and a Net Present Cost (NPC) of 27,342,795 $ for 20 years, providing a low surplus or deficit of energy of 10,617,155 kWh/year. The proposed system minimizes the use of fossil fuels and carbon dioxide emissions significantly, thereby providing a sustainable and economically viable alternative for rural communities. This study underscores the importance of using data-driven approaches to maximize the use of renewable energy and additionally, it can offer a scalable framework for designing sustainable, off-grid energy systems.