Machine learning-based real-time power stability optimization for photovoltaic systems using hybrid inductor-capacitor patterns

Abstract

Photovoltaic (PV) systems often face real-time power stability challenges due to rapid fluctuations in solar irradiance and varying load conditions, which conventional control strategies struggle to manage effectively. Addressing this limitation, the present study proposes a novel machine learning-based control framework integrated with a hybrid inductor-capacitor (LC) network to enhance dynamic power regulation. The proposed system employs predictive algorithms to adjust LC parameters in real time, enabling adaptive voltage and current stabilization during transient conditions. Simulation results validate the model's effectiveness, showing a 58% reduction in power fluctuation (from 12% to 5%) and consistent improvement in voltage stability index (VSI), maintaining values above 0.95 compared to 0.88-0.93 in traditional systems. Moreover, the approach reduces computation time by 66% (150 ms versus 450 ms for PID-based systems), supporting faster and more efficient control actions. These outcomes demonstrate that the proposed intelligent control strategy significantly improves energy efficiency, voltage stability, and responsiveness in PV systems, offering a scalable solution for reliable grid integration of renewable energy sources.