فهرست نشریات علمی دسترسی آزاد ایران

This paper presents a comprehensive methodology to improve solar irradiance forecasting and optimize its integration into Hybrid Electric Vehicle (HEV) Energy Management Systems (EMS). Accurate solar irradiance prediction is critical for maximizing renewable energy utilization and ensuring reliable operation of hybrid storage devices. To achieve this, the proposed framework combines Deep Neural Networks (DNNs) with the Horizontal Diffuse K Clearness Index Regression (HDKR) technique for day-ahead forecasting of Global Horizontal Irradiance (GHI). This hybrid approach not only predicts GHI with high precision but also decomposes it into beam, diffuse, and reflected components on tilted surfaces, thereby enabling more realistic modeling of Photovoltaic (PV) generation under varying conditions. The forecasting model achieved an R-squared value of 0.8818 and a Root Mean Square Error (RMSE) of 0.269, confirming its robustness and accuracy. These forecasts are incorporated into a real-time EMS that coordinates energy distribution among the battery, ultracapacitor, and PV subsystems while accounting for seasonal variability, auxiliary system demands, and dynamic driving cycles. Unlike traditional static State of Charge (SOC) targets, the proposed dynamic SOC control adapts to daily, monthly, and seasonal variations, allowing efficient energy sharing between storage devices. This strategy reduces stress on critical battery components by 47.6%, thereby improving both performance and lifespan. Furthermore, the integration of ultracapacitors effectively mitigates transient load demands, ensuring higher system reliability under fluctuating conditions. Overall, the research demonstrates how HDKR-based estimation combined with adaptive EMS strategies can significantly enhance efficiency, sustainability, and durability in solar-assisted HEVs.