论文部分内容阅读
独立元分析(ICA)在线性过程监控中得到了成功的应用,但是实际工业过程大部分是非线性的。在利用核ICA(KICA)建立过程非线性模型的基础上,根据核技巧,给出了一种高维空间分离矩阵的排序和独立元个数的选择方法,并将监控指标扩展到高维空间,从而提出一种基于KICA的非线性过程监控方法,解决了ICA对非线性过程监控效果不理想的缺点。以田纳西-伊斯曼过程(TE过程)为例,对比了KICA与ICA的监控效果,结果证明了该方法的优越性。
Independent element analysis (ICA) has been successfully applied in linear process monitoring, but most of the actual industrial processes are non-linear. Based on the kernel ICA (KICA), a method based on kernel technique is proposed to select a sort of high dimensional separation matrix and the number of independent elements, and the monitoring index is extended to high dimensional space , A KICA-based nonlinear process monitoring method is proposed, which solves the defect that ICA is not ideal for nonlinear process monitoring. Taking the Tennessee-Eastman process (TE process) as an example, the monitoring results of KICA and ICA are compared. The result proves the superiority of this method.