Self-optimizing control of complex processes

Abstract

The selection of appropriate controlled variables (CVs) is important during the design of control systems for complex processes. In this project, a systematic method for CV selection using the concept of self-optimizing control is developed. In particular, a method for selecting linear combinations of measurements as CVs, as compared to the traditional approach of selecting a subset of available measurements as CVs, has been derived. In addition, branch and bound (BAB) methods for efficient selection of CVs from the large number of available measurements have been developed. The BAB method has also been extended to select the pairings of the selected CVs with manipulated variables for decentralized control. The practical application of the theoretical results has been demonstrated using case studies of forced circulation evaporator, liquefied natural gas (LNG) plant and solid oxide fuel cells. The derived results will be useful for researchers as well as practitioners in efficiently designing control systems for industrial processes. This work has also resulted in publication of 5 papers in international journals and 8 papers in conference proceedings.