A method based on BP neural network is proposed for the diagnosis of UAV airborne power system in this paper. According to the mathematical model of onboard power system, the simulink model is set up on the basis of the structure and working principle of airborne power system combined with actual flight data and related parameters. According to the properties of the object, a failure feature database for generator of the onboard power system is established by using PCA (principal component analysis). And it realizes fault diagnosis by using BP neural network model. The results show that the system can diagnose UAV generator system effectively and has practicability.