首页|Fault diagnosis based on online dynamic integration model and transfer entropy
Fault diagnosis based on online dynamic integration model and transfer entropy
扫码查看
点击上方二维码区域，可以放大扫码查看
原文链接
NSTL
Elsevier
? 2022 The Author(s)In this paper, the monitoring method based on slow feature analysis and the root cause diagnosis based on transfer entropy are proposed. Different from the traditional methods, this paper performing online modelling to improve the adaptability to different data features and the dynamics of the model. A Dynamic Dissimilarity Index (DDI) is proposed to construct a dynamic online sub-block partitioning model. For process monitoring of multidimensional quality variables, a weighted fusion method is proposed to integrate multiple models so that the monitoring conclusions include the quality changes of the process. The fault variables are identified for quality-related faults, and the traditional method is improved by proposing a transfer entropy method with time lag parameters added to optimize the fault causality diagram. Finally, the study of real chemical processes shows that the proposed method gives more accurate and richer monitoring conclusions and can find the root cause more quickly and efficiently.
Dynamic integration modelModified transfer entropyQuality-relevantSlow feature analysisSub-block partitioning
Yang Y.、Kang W.、Liu X.
展开 >
College of Information Science and Engineering Northeastern University
NSTL
Elsevier
