Abstract:

In order to troubleshoot the aircraft pneumatic system's problem fast and accurately,CBR methodology is applied to the fault diagnosis and case base is made based on practical experience from airlines and manufacture experts. Firstly,different types of knowledge representation methods are proposed. Secondly,the similarity in case retrieval is discussed in details. Aiming at the disadvantages of k-NN algorithm similarity as a reference,the dynamic fault ratio is established and the reference degree is formed based on it. Experimental results show that the proposed case base could detect faults effectively,and it is a robust foundation of fault reasoning algorithms.

Key words: aircraft pneumatic system, fault diagnosis, case base, dynamic fault ratio, reference degree