基于主元核相似度免疫机制的故障诊断方法及应用

1引言目前,国内外已发展了多种应用于航空发动机的智能故障诊断方法,如专家系统方法、神经网络方法,以及新近提出的基于支持向量机的诊断方法。但这些方法需要有足够的已知故障模式的训练样本才能发挥出优异的性能,而且算法实时性较差,难以满足快速在线诊断的要求1]。传统的基

Fault diagnosis based on immunity strategy with similarity measurement of principle component kernel and its application

To solve the problem that exists in fault diagnosis based on the traditional method using distance discriminant function, an improved method based on immunity strategy with similarity measurement of principle component kernel was presented for aircraft engines. Based on the theory of principle component kernel and immunity strategy, the similarity measurement in immunity space is in the form of principle component kernel. Thus each certain sample is regarded as an antibody, and the sample to be recognized is regarded as an antigen. So the fault diagnosis can be converted into the process of antibody recognizing antigen. The practical application results show that the distribution structure of fault modes has little effect on the method proposed, and when the clustering samples are scattered greatly, the diagnosis results are still accurate.