Abstract: In order to improve the ability of condition monitoring and fault diagnostic system, a hybrid intelligent diagnostic system based on multilevel flow model (MFM) and information fusion was proposed. This method utilized information fusion technique to improve the rapidness and veracity of fault diagnosis, and made use of MFM to explain the alarm propagation path, which could enhance the comprehension of diagnostic result. The emulation test proves that the hybrid intelligent diagnostic system can identify fault and propose the alarm analysis quickly.