Computational Intelligence-Based Modeling of a UAV-Integrated PV Module in Icing Conditions

Solar UAVs utilize solar energy to extend flight endurance and reduce maintenance compared to traditional UAVs. However, in-flight icing presents significant challenges, impacting both aerodynamic performance and the operational reliability of UAV-integrated photovoltaic (PV) systems. Ice accumulation on wings degrades mechanical properties, while icing on PV modules obstructs sunlight, adversely affecting their parameters. Partial shading from ice is more harmful than uniform shading, complicating PV module behavior analysis. This study presents a novel approach to model the behavior of UAV-integrated PV modules incorporating the impact of nonuniform in-flight icing into irradiance calculations. By analyzing how ice formation acts as an obstacle and considering its effect on the PV module governing equations, this research develops a computational framework alongside with segmenting the PV module's operational curve into two zones: ice-covered and normal. The parameters for these zones are determined using advanced computational intelligence methods. The proposed method enables predictions of PV module performance using trained machine learning models, enhanced by the minimum redundancy maximum relevance technique, under dynamic and adverse conditions such as movement-induced sunlight variations and partial shading. The trained models’ performance is validated through experimental tests, demonstrating the reliability and effectiveness of the approach.