Bayesian method for multimode non-Gaussian process monitoring

Non-Gaussian processes monitoring has recently caught much attentions in this area, with several methods successfully developed, such as non-parameter estimation, independent component analysis (ICA), support vector data description (SVDD), and etc. However, most of current research works are under the assumption that the process is operated in a single mode. This paper proposed a novel method for monitoring multimode non-Gaussian processes, which is based on Bayesian inference. To improve the comprehension of the process for the operation engineer, a corresponding mode localization approach is also given. A case study on the Tennessee Eastman (TE) benchmark process shows the feasibility and efficiency of the proposed method.