基于核函数PCA的非线性过程实时监控方法

为了克服基于主元分析的过程监控方法非线性处理能力弱的缺点和降低基于非线性主元分析的过程监控方法的计算复杂度，提出了将核函数PCA监控方法用于复杂工业过程实时监控系统的开发研究，并讨论了核函数参数选择对系统性能的影响。核函数PCA能有效地提取过程变量的非线性关系，而且计算复杂度低，便于在线实施。仿真结果表明该方法是一种有前途的复杂过程非线性实时监控技术。

On-Line Monitoring of Nonlinear Processes Based on Kernel Principal Component Analysis

On-line nonlinear monitoring of complex process is developed using kernel principal component analysis to overcome the drawbacks of principal component analysis(PCA) in dealing with nonlinear process and decrease the computation load of some process monitoring methods based on nonlinear PCA.Influence of kernel parameters on the process monitoring performance is also studied.Kernel PCA effectively captures the nonlinear relationship among the process variabres and is easy to implement on-line.Simulation results have shown that it is an promising on-line process monitoring technique for complex processes.