Fault detection for chemical process based on nonlinear dynamic global-local preserving projections

The performance of the traditional nonlinear fault detection method based on kernel mapping is greatly influenced by the type of kernel function and the tuning of kernel parameters. To solve this problem, a method named nonlinear dynamic global-local preserving projections(NDGLPP) is proposed for nonlinear process fault detection. Firstly, dynamic global-local preserving projection algorithm is used to reduce the dimension of data matrix. Since the second order polynomial mapping is established for the reduced dimension matrix to extract the relevant properties of nonlinear space. Then the two steps are iterated to obtain the higher-order nonlinear mapping. Finally, the proposed method is applied to the ethylene distillation process and Tennessee Eastman (TE) process simulation to verify the effectiveness and feasibility of the detection method.