Abstract
Monitoring of chemical processes is becoming increasingly difticult as a result of growing complexity and larger scale of operation. In this paper, Kohonen self-organizing map (SOM) is used to monitor the operation of a lab-scale distillation column and to identify process states. SOM projects high-dimensional data to a lower two dimensional grid map while preserving the metric relations of the original data. The results from this paper show that using this property of SOM, process monitoring can be performed effectively through observing time series trajectory of process operations on SOM while fulfilling the objective of state identification at the same time. Occurrence of a fault will result in the deviation from the normal operating trajectory. Root cause identification can also be performed through simple visualization of component planes.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call