Abstract: In order to handle the multiple operating modes problem and the widely existing non-Gaussian information of industrial processes, a new online process monitoring method was proposed, which was based on external analysis and independent component analysis and principal component analysis (ICA-PCA).First, process variables were divided into two parts: external variables and main variables, and partial least square (PLS) was used to remove the influence of external variables on main variables.Then a two-step information extraction method ICA-PCA was used to extract the process information completely.Finally, three statistics were built to monitor the process.An online algorithm was also developed.Simulations of a linear system and CSTR process showed the feasibility and effectiveness of the new method.

Key words:

多工况, 非高斯, 外部分析, ICA-PCA, PLS

摘要：

针对复杂工业过程中的多工况和非高斯信息问题，提出一种基于外部分析的ICA-PCA（independent component analysis and principal component analysis）在线统计监测新方法。首先把过程变量分为外部变量和主要变量，通过偏最小二乘（PLS）回归方法分离外部变量对主要变量的影响，然后利用ICA-PCA两步信息提取策略，完整地提取过程的信息，最后用3个统计量对过程进行监测，建立了一种具有非高斯特性的多工况过程在线监测算法。通过对一个数值例子和连续搅拌槽（CSTR）过程的仿真研究，说明提出的方法是可行、有效的。

关键词:

多工况, 非高斯, 外部分析, ICA-PCA, PLS