摘要
通常情况下利用传统的主元分析方法虽然可以对系统进行故障检测和诊断,但是如果数据标准化以后呈“均匀”分布时,由于很难选取主元,或者选取出主元时没有考虑随机向量分量的物理意义,使得主元没有代表性。在分析了主元分析方法的基础上,我们提出了一种相对主元分析的方法,给出了相对主元的几何意义,同时还提出了相对化变换,分布“均匀”等概念。在处理分布“均匀”数据时,应用新概念和新方法,可有效地克服传统的主元分析(PCA)在数据压缩和故障检测与诊断时的不足。仿真结果验证了算法的有效性。
After dataare standardized by traditional Principal Component Analysis （PCA）, the processed data often show ＂Rotundity Distributing＂. Therefore, it is difficult to choose representative principal components for fault detection and diagnosis. The concept of Relative Principal Component （RPC ） was proposed, and a new algorithm was given based on Relative Principal Component Analysis （RPCA ）. Some new concepts such as Relative Transform, Rotundity Distributing and so on were proposed. The new algorithm could overcome some disadvantages of Principal Component Analysis （PCA） for fault detection and diagnosis when data was Rotundity Distributing. The Relative Principal Components selected by RPCA are more representative, and their significance of geometry is more notable. The simulation demonstrates the effectiveness of the algorithm proposed.
出处
《系统仿真学报》
EI
CAS
CSCD
北大核心
2007年第13期2889-2894,共6页
Journal of System Simulation
基金
国家自然科学基金(200660572051)
国家自然科学基金(200560434020)
上海市教委教育发展计划项目(05FZ04)
上海市教委项目(04FB08)
关键词
主元分析
相对主元
相对化变换
分布“均匀”
故障检测
principal component analysis
relative principal component
relative transform
rotundity distributing
fault detection
