Slow-Feature-Analysis-Based Batch Process Monitoring With Comprehensive Interpretation of Operation Condition Deviation and Dynamic Anomaly
- 12 July 2018
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industrial Electronics
- Vol. 66 (5) , 3773-3783
Abstract
In order to provide more sensitive monitoring results, the time dynamics and steady-state operating conditions should be separately monitored by distinguishing time information from the steady-state counterpart. However, it is a more challenging task for batch processes because they vary from phase to phase presenting multiple steady states and complex dynamic characteristics. To address the above issue, a concurrent monitoring strategy of multiphase steady states and process dynamics is developed for batch processes in this work. On one hand, multiple local models are constructed to identify a steady derivation from the normal operating condition for different phases. On the other hand, based on the recognition that the process dynamics can be considered to be irrelevant with the steady states, a global model is built to detect the dynamics anomalies by monitoring the time variations. Corresponding to alarms issued by different statistics, different operating statuses are indicated with meaningful physical interpretation and deep process understanding. To illustrate the feasibility and efficacy, the proposed algorithm is applied to a typical multiphase batch process, the injection molding process.Keywords
Funding Information
- NSFC-Zhejiang Joint Fund for the Integration of Industrialization and Informatization (U1709211)
- National Natural Science Foundation of China (61433005)
- State Key Laboratory of Industrial Control Technology
- Zhejiang University (ICT1802)
- State Key Laboratory of Industrial Control Technology
- Zhejiang University (ICT1800397)
This publication has 35 references indexed in Scilit:
- Multimode process monitoring with Bayesian inference‐based finite Gaussian mixture modelsAIChE Journal, 2008
- A Maximum-Likelihood Interpretation for Slow Feature AnalysisNeural Computation, 2007
- Slow feature analysis yields a rich repertoire of complex cell propertiesJournal of Vision, 2005
- A new fault diagnosis method using fault directions in Fisher discriminant analysisAIChE Journal, 2005
- Sub‐PCA modeling and on‐line monitoring strategy for batch processesAIChE Journal, 2004
- Multivariate dynamic data modeling for analysis and statistical process control of batch processes, start‐ups and grade transitionsJournal of Chemometrics, 2003
- On-line Batch Process Monitoring Using Different Unfolding Method and Independent Component Analysis.JOURNAL OF CHEMICAL ENGINEERING OF JAPAN, 2003
- Statistical monitoring of multistage, multiphase batch processesIEEE Control Systems, 2002
- Disturbance detection and isolation by dynamic principal component analysisChemometrics and Intelligent Laboratory Systems, 1995
- Monitoring batch processes using multiway principal component analysisAIChE Journal, 1994