Fault Detection in Renewable Energy Systems Using Large Time Series Models

Renewable energy systems have become an important part of the modern power grid, but they also face more complex fault situations than traditional energy sources. This paper presents a novel fault detection framework for photovoltaic (PV) and wind power systems, utilizing large time series models to capture complex temporal dependencies and enhance diagnostic accuracy. The proposed framework integrates historical operational data with real-time data streams to effectively address the challenges posed by fault detection in renewable energy systems, particularly in the presence of rare and diverse fault types. Comprehensive experiments on synthetic and real-world datasets demonstrate that the inclusion of large time models significantly improves model performance, reducing error rates and enhancing the robustness of fault detection. The results underscore the potential of large time series models, particularly in zero-shot learning contexts, for improving fault diagnosis and operational reliability in renewable energy systems.