一种简化的动态独立元分析方法

A simplified dynamic independent component analysis

在动态过程中,传统主元分析(PCA)法无法利用过程变量的自相关性对过程进行准确地分析,而且过程数据不一定满足高斯分布从而导致误报和漏报,这个问题可以采用动态独立元分析法(DICA)来解决.本文根据动态过程时间滞后的大小提出了一种简化的DICA方法,并用相应的二阶动态模型进行了仿真验证,结果表明与DICA相比,该方法同样有效且简便易行.

The traditional principal component analysis （PCA） can not analyze dynamic process with the selfcorrelation information of the process variable. Moreover, the process data may not follow Gaussian distribution so as to cause the improper detection of PCA. Dynamic independent component analysis （DICA） has the ability to solve the problems above. In this paper, a simplified DICA approach is proposed based on the time lag of the dynamic process, correspondingly, the second order autoregressive models are used to validate the possibility of the method. The simulation results show that it is an efficient and simple method compared with DICA.