As modern industrial processes become more complex with the development of technology, more advanced monitoring methodologies are required to keep high efficiency and safe operation. In order to improve monitoring performance, it is important to identify the operational phase of a process by evaluating all measurable variables comprehensively. Multiple operational characteristic that occurs in most industrial processes like chemical processes by their physical and chemical properties, variations in market requirement and product specification, catalyst degradation, etc. has become an important factor to be considered in process monitoring.
Since most multimode monitoring approaches with the mode-identification are only based on the similarity of process mean, the identification of transitions and faults is possible only after the cluster of process data should be formed or neighbour modes could be analyzed. Furthermore, mode-identification based approach seems to be inappropriate to monitoring of process with lots of modes not only because it requires a large amount of off-line data but also because its capability of adapting to unexpected states may not be acceptable.
Song Kwang Rim, a researcher at the Faculty of Automation Engineering, has proposed an online transition-identification based monitoring procedure for industrial process with multiple operational modes using process variability.
Firstly, he detected the change point in the dynamical behavior of multimode process by the log determinant of covariance matrix.
Second, based upon the analysis of the change behavior of process variability during transition and fault state, he developed new statistics to identify the transition.
Finally, he adopted an adaptive monitoring strategy to monitor multimode processes by the single model using the transition-identification.
Then, he conducted three case studies through TE benchmark to verify the usefulness and effectiveness of the proposed approach.
The result showed that the proposed approach has high ability to identify transitions and faults and to cope with the occurrence of new modes.
The details of this are found in his paper “Monitoring Industrial Processes with Multiple Operation Modes: a Transition-Identi¦cation Approach Based on Process Variability” in “Industrial & Engineering Chemistry Research” (SCI).
© 2021 KumChaek University of Technology