Abstract:
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.