Development of an artificial intelligence-based adaptive digital twin method for early fault detection and energy loss reduction in solar photovoltaic systems

Abstract

This article examines the scientific and conceptual foundations for developing an artificial intelligence-based adaptive digital twin method aimed at early fault detection and the reduction of associated energy losses in solar photovoltaic systems. An adaptive digital twin model is proposed by integrating real-time monitoring data, environmental parameters, electrical indicators, technical condition characteristics, and time-varying operational data into a unified information environment. The model is designed to enable early detection of operational deviations, forecasting of potential fault development, assessment of their impact on energy generation, and formulation of optimal technical decisions. The scientific novelty lies in dynamically updating the virtual model and integrating the sequence of “monitoring → early anomaly detection → fault forecasting → energy-loss assessment → technical decision” into a unified adaptive digital twin architecture.