Abstract: Distributed photovoltaic systems are affected by multiple factors such as shadow occlusion, temperature and humidity fluctuations, and equipment aging and so on, resulting in highly uncertain and concealed operational states, which pose significant challenges for their operation, maintenance, and troubleshooting. This paper analyzes the types of distributed PV failures, proposes methods for extracting anomalous features of PV systems, and conducts simulations and characteristic analyses of different failure types based on Matlab/Simulink. The study establishes a robust data foundation for improving detection accuracy, effectively supporting PV fault detection practices.