摘要
某型号液体火箭故障仿真过程中涉及大量描述发动机状态的参数,因此在对该设备进行故障诊断前,需要对监测或仿真数据进行特征提取,以减少存储空间,缩短故障诊断时间。采用主分量分析法及其改进算法核主分量分析法对其故障仿真数据进行特征提取,从多个描述该型号一级火箭发动机故障状态的变量中选取了少量特征,采用这些特征进行故障诊断时,诊断结果正确,同时显著提高了故障诊断的实时性能。
There are a lot of parameters used to describe the situation of a liquid rocket engine in its fault simulation process, so it is necessary to extract the main features of the variables in fault diagnosis process in order to minimize the storage space and to shorten the time complexity. The Principal Component Analysis （PCA） method and its improving algorithm, the Kernel-based Principal Component Analysis （KPCA）, are applied in the feature extraction of the simulated fault data. Fewer characteristics are obtained from the numerous variables which described the first-stage liquid rocket engine. The judgements made in the fault diagnosis process are right by this method, and the real-time ability is prominently improved.
出处
《导弹与航天运载技术》
北大核心
2009年第2期5-7,36,共4页
Missiles and Space Vehicles
关键词
核主分量分析法
特征提取
故障特征
Kernel-based principal component analysis
Fault characteristics
Feature extraction
