摘要
In this paper, a reconstruction-based method is proposed to monitor nonlinear processes and isolate their fault types. The fault direction and enhanced statistics are introduced to improve the monitoring results and effectiveness of fault isolation. The main contributions are given as follows: 1) For fault detection, two statistics, which are decomposed according to the load directions, are defined in the complementary subspaces. 2) The fault direction is used to obtain the reconstruction model. 3) Based on the reconstruction model, the fault type can be isolated. An electro-fused magnesia furnace is introduced to test the proposed method, and the results of this test validate its effectiveness.
- 出版日期2017-12
- 单位东北大学