The operation modes of rectifier with different switches open-circuit faults are analyzed. The analysis result shows that the fly-wheel diode open-circuit faults will cause the most damage for rectifier. The fault diagnostic system based on neural network is constructed. The DC component in three-phase input current and the AC ripple in output voltage of rectifier are proposed as fault characteristic values and training data for neural network. The training and testing results demonstrate that the fault diagnostic system using the proposed neural network has the excellent fault diagnostic ability that can locate the fault power devices in the rectfier except the fly-wheel diodes.