Abstract: This paper is aimed at photovoltaic module failures in actual power stations. Through field investigation, some common fault types in polycrystalline silicon PV modules are collected,the reverse bias model of PV modules is established. Ⅰ-Ⅴ output characteristics under the mismatch of PV modules is analyzed,and the Ⅰ-Ⅴ characteristic curves of PV modules with different fault types are tested. The main reason for the convex function feature of the glass-cracked PV module Ⅰ-Ⅴ characteristics is analyzed. Fault characteristics of PV modules with different fault types are extracted. An online fault diagnosis method for PV modules based on Ⅰ-Ⅴ characteristics is proposed. The method diagnoses aging through Fill Factor（FF）. Open-circuit voltage ratio under high and low irradiance is used to diagnose potential-induced degradation（PID）. The step interval is detected according to the concavity and convexity of the Ⅰ-Ⅴ curve,and the maximum distance of the line-to-stage detection line on the Ⅰ-Ⅴ curve in the interval is used to diagnose hot spot and glass cracking fault. Furthermore,the power optimizer is combined to verify that the fault diagnosis method has high accuracy and high practical application value.