Abstract: This paper proposes an intelligent fault diagnosis method based on rough set and BP neural network.First, for these equipments contain a huge number of redundant data, this method pretreated the preliminary data for fault diagnosis to reduce raw data attribute by use of rough set, and then analyzed the processed data with BP neural network technology which improved by using LM algorithms to implement the fault diagnosis.It established a model of fault diagnosis based on RS-LMBP algorithm, and the examples demonstrate that the method is feasibility and high efficiency.