Abstract
In this paper, a new online monitoring approach is proposed for handling the dynamical multimode problem in the industrial processes. The contributions are as follows: (1) extracting method of the common characteristics from different modes is proposed; (2) nonlinear dynamic monitoring method is proposed; and (3) a new model analysis method is proposed. There are both similarity and dissimilarity in the underlying correlations of different modes. After two different subspaces are separated, models of the common and specific subspaces are built respectively. Then the common subspace and specific subspace are analyzed, where the monitoring process is carried out in each subspace. When the mode switches, the specific monitoring model is changed. The corresponding confidence regions are constructed according to their models respectively. The effectiveness of the proposed method has been demonstrated via simulated examples. © 2012 Elsevier Inc. All rights reserved.
| Original language | English |
|---|---|
| Pages (from-to) | 97-105 |
| Number of pages | 8 |
| Journal | Information Sciences |
| Volume | 235 |
| DOIs | |
| Publication status | Published - 20 Jun 2013 |
Keywords
- Common subspace
- Kernel method
- Process monitoring
- Subspace separation