Abstract: A fault fuzzy diagnosis method based on correlation matrix and probability model is proposed to deal with complicated system fault diagnosis problems. A fault-test correlation matrix is established and the fault criterion in probability space is obtained. By investigating the probabilities of fault features in non-ideal test conditions, and using the principles of reasoning, a fault fuzzy diagnosis can be done. Faults filtering according to minimum norm can further reduce the false alarm, which contributes to achieve an efficient and convenient diagnosis for complicated system. The proposed method is validated by using the test data of Apollo before launching. Experimental results show that the proposed method has sufficient diagnostic accuracy, as well as low computational complexity, which can support rapid diagnosis and be convenient for on-site emergency use.