Development of MBPLS based control for serial operation processes
- First Online:
- 60 Downloads
A control scheme based on the multiblock PLS (MBPLS) model for multi-stage processes (or serially connected processes) is developed. MBPLS arranges a large number of variables into meaningful blocks for each stage of the large-scale system. Two control design strategies, course-to-course (CtC) and within-stage (WS) controls, are proposed for the re-optimization design in the whole multistage course. In CtC, MBPLS control and optimization are done by applying feedback from the finished output quality when one course for all stages is done. It utilizes the information from the current course to improve quality of the next one. In WS, the MBPLS-based re-optimization strategy is developed to explore the possible adjustments of the future inputs at the rest of the stages in order to fix up the disturbances just in time and to maintain the product specification when the current course is finished. The proposed technique is successfully applied to two simulated industrial problems, including a photolithography sequences and a reverse osmosis desalination process, and the advantages of the proposed method are demonstrated.
Key wordsIterative Learning Control Multistage Process Partial Least Squares
Unable to display preview. Download preview PDF.
- 1.J. C. Morud and S. Skogestad, The dynamic behavior of cascade processes with application to distillation columns, AIChE Annual Meeting, Miami Beach (1995).Google Scholar
- 3.S. Vaidyanathan, Stochastic control of sequential manufacturing process, Ph.D dissertation, Carnegie Mellon Univ., Pittsburgh, PA (1991).Google Scholar