基于自组织拓扑映射图的发动机故障诊断研究

根据SOM自组织网络的高维数据矢量量化、快速聚类、信息融合和拓扑映射等特征,建立了发动机故障诊断可视拓扑映射图模型,给出了映射模式误差测度和拓扑误差测度,并通过拓扑映射图对发动机故障进行诊断,获得了很好的效果。同时,进一步通过建立的故障模式的测量参数灰度图,可以清晰地观察到故障模式矢量特征、测量参数相关性特征、参数变化趋势,并可进行更深层次故障机理分析。

Aeroengine Fault Diagnosis Based on Self-Organizing Topology Maps

This paper proposes a visualized aeroengine fault diagnosis system mode based on self organizing topology map which combines vector quantization,fast clustering,information fuse and topology preservation,estimation of topographic error and quantization error is also given.And then this method is applied to aeroengine gas path fault diagnosis.The results show that this diagnosis mode is effective and rational.Furthermore,the paper has set up component visualized gray plane in which characteristics of fault vector,correlation of measured parameters and tendency of components can be observed easily,and also we can analyze inherence of faults by this approach.