Abstract: With the large-scale popularization of photovoltaic (PV) power generation, the problem of fault detection in PV system has become a research hotspot. With the continuous technological innovation, it is particularly important to be able to predict and prevent the occurrence of PV system faults in advance and ensure the reliable operation of the system. In this paper, based on the working condition, module structure, aging phenomenon and corresponding equivalent circuit model parameter changes of PV modules, the health state of PV modules was divided, and three major indicators affecting the sub-health state of PV modules were summarized, namely, light transmittance, the series resistance and the parallel resistance. A fuzzy algorithm was proposed to diagnose the health state of PV modules, including health, sub-health, partial shadow and fault state. Firstly, Fuzzy C-Means (FCM) clustering was performed on the normalized PV module sample data set to obtain the clustering center. Then, the clustering center and test sample was substituted into Gaussian membership function to diagnose the health status. Finally, the feasibility of the proposed method is verified by simulation and experiment, which provides a reference for fault warning and aging detection in PV system.