Enhancing the Performance of Solar Panel during Partial Shading Condition

Abstract

Partial shading in photovoltaic (PV) systems poses a significant challenge to energy harvest efficiency, often causing disproportionate power losses that far exceed the physical area under shade. This paper presents a novel approach to mitigate these effects through an intelligent dynamically reconfigurable PV array system. The proposed solution integrates advanced hardware architecture using silicon carbide switching matrices with a multi-layered control system that combines computer vision for shadow detection, long short-term memory (LSTM) networks for shadow movement prediction, and hybrid particle swarm optimization with artificial neural network (PSO-ANN) algorithms for maximum power point tracking. Experimental results from a 3 kW test system demonstrate that the reconfigurable array achieves a 20.9% increase in annual energy yield compared to conventional string inverter systems and a 5.1% improvement over module-level power electronics solutions. The performance advantage is most pronounced during challenging operating conditions such as winter months, early morning/late afternoon periods, and partially cloudy days. Economic analysis reveals that despite higher initial costs, the reconfigurable array system achieves the lowest levelized cost of electricity ($0.085/kWh) with an acceptable payback period of 8.86 years. Thanks to built-in machine learning tools, solar power systems are now performing impressively well even outside the lab—in actual outdoor environments—without much drop in efficiency. This study shows that using smart, predictive reconfiguration methods makes solar setups more adaptable to problems like partial shading. What does that mean? We can now install solar panels in places that were once considered less ideal, and still get great results. It also means better returns on investment for existing solar installations, making solar energy more practical and profitable than ever.