Near-optimal operation by self-optimizing control: from process control to marathon running and business systems

Abstract

The topic of this paper is how to implement optimal decisions in an uncertain world. A study of how this is done in real systems-from the nationwide optimization of the economy by the Central Bank to the optimal use of resources in a single cell-shows that a common approach is to use feedback strategies where selected controlled variables are kept at constant values. For example, in order to optimize the wealth of a country (overall objective), the Central Bank may to attempt to keep the inflation constant (selected controlled variable) by adjusting the interest rate (independent input variable). The underlying idea is that the system behavior is indirectly optimized by keeping selected controlled variables at given constant values (setpoints). In the paper this idea of "self-optimizing control" is explained and illustrated on a large number of examples.

Integration of process design and control: A review

“…Two-step approach Joseph (1987) Marlin andHrymak (1997) Two-step approach Joseph (1987) Marlin andHrymak (1997) Identification for optimization Srinivasan and Bonvin (2002) Bias update Forbes and Marline (1994) Constraint adaptation Chachuat , et al (2008) ISOPE Roberts (1979); Tatjewski, P., (2002); Brdys and Tatjewski (2005) Gradient correction Gao andEngell (2005) Marchetti, et al (2009) Self-optimizing Skogestad (2000b); Govatsmark and Skogestad (2005) Extreme seeking Ariyur and Krstic (2003); Guay and Zhang (2003) NCO tracking Francois, et al (2005); Srinivasan, et al(2008) Active constraints tracking Maarleveld and Rijnsdorp (1970 Later, researchers (e.g. Skogestad 2000aSkogestad , 2000bSkogestad , 2004bKariwala 2007) investigated the notion of self-optimizing control. The concept is shown in Fig.…”

Solution of Optimal Control Problems

“…Since the control structure (given by successive sequence of the arcs) is given by the nominal solution and if the sequence of arcs is not affected by an uncertainty, NCO can easily be checked at certain points and then they can be pushed towards their satisfaction. This approach is known as NCO-tracking or Self Optimising Control (SOC) and has been introduced by groups of Bonvin [25][26][27] and Skogestad [1,24].…”

Closed-loop control of a polymerisation plant using production performance indicators (PIs)

“…The idea of hierarchical control levels is related to the so-called self-optimising control that was presented by Skogestad (2000Skogestad ( , 2004. Generally speaking, for many systems (companies, chemical processes, biological systems, etc) we have available degrees of freedom (decisions), u, that we want to use in order to optimise the system operation.…”

“…If there were no changes in the process environment, there were constant measurable Figure 2. Implementation of the optimal operation of a process with separate layers for optimisation and control (Skogestad, 2004). influences, d, and there were no unmeasurable disturbances, then a suitably fixed setting, u, would solve the (optimal) control on the production level. As this is never true, an appropriate strategy to respond to the perturbations from the environment by readjusting the manipulative inputs (u) is needed.…”