A Reasoning Framework for Knowledge Learning in Photovoltaics Degradation Science Studies

Degradation science studies in photovoltaics (PV) are vital for assessing the impact of environmental conditions on the performance and longevity of PV components. Traditional approaches, often limited by poor metadata practices and manual data analysis, hinder reproducibility and efficiency. This work introduces a data-driven reasoning framework utilizing ontologies, knowledge graphs, and semantic reasoning to address these challenges. Using a large-scale study dataset which includes accelerated degradation data on PV minimodules, we developed domain-specific ontologies to create FAIR (Findable, Accessible, Interoperable, Reusable) knowledge graphs. These graphs enable advanced reasoning and inference, uncovering key relationships between process parameters and degradation pathways. This framework unlocks new insights and optimizes research in PV degradation studies.