Supervisory multiobjective optimization of a class of unit processes: power unit case study

摘要

There are many industrial unit processes that require several control loops to operate and that are commanded through a single unit demand signal along predefined wide-range unit production patterns. Normally, these processes are required to satisfy multiple operating objectives simultaneously including: increase production, save raw materials, decrease energy consumption, increase quality of product, etc., all of them with different priorities. Often, control systems for these processes consist of a generic two-level hierarchical control scheme with a direct regulatory control layer and a supervisory optimization layer. This paper presents a reference governor in the supervisory layer for such class of control systems to achieve multiobjective optimal wide-range operation through set-point scheduling. The reference governor consists of a set of mappings, from the unit demand to the set-points of the control loops in the control layer, which are designed by solving a multiple objective optimization problem. This approach provides a way to specify an operating policy for the unit process in terms of an arbitrary number of objective functions and their preferences. This in turn constitutes the means to optimally accommodate, in a multiobjective sense, a great diversity of operating scenarios. The scheme is applied to a power unit as a case study.