Researchers Stony Brook University, working with Ecosuite and using historical datasets from Ecogy Energy, developed a data-driven algorithm to detect physical anomalies in solar energy systems. The study focused on reducing solar project O&M costs by identifying long-term weather-related and inverter issues. The researchers trained self-supervised anomaly detectors using inverter and weather data and applied a unified data pipeline based on widely available datasets. The paper addressed long-term anomalies often missed by asset managers. The models were designed to predict and diagnose physical issues weeks or years in advance. The approach supported earlier maintenance planning, extended equipment lifetimes and reduced energy losses from unresolved system issues.