文摘
This paper addresses the problem of designing stabilizing globallinearization controllers subjectto bounded inputs. It is well-known that saturation of the inputvariables usually leads to aserious performance degradation of even instability of processes ifundesired attractors arepresent. Instability occurs in open-loop unstable plants once thecontroller hits the constraints.If the plant is open-loop stable, saturation may also lead toinstability in the form of limit cycles.This paper provides conditions under which the existence of stableglobal linearization controllersfor SISO and MIMO plants is ensured. These conditions will set upthe basis to derive a tuningtechnique that, preserving stability, renders a quite acceptableperformance in terms of closed-loop response. Two problem categories, and solution strategies,will be considered: (1) If theset of constrained inputs Ub is given, afeasible region in the state space is calculated and alow-gain static controller computed to preserve stability.Performance is then improved by addingand saturating and external controller. (2) If the feasible set inthe state space and the desireddynamic behavior are given, a feasible set, Ub,and thus a global linearization controller, iscomputed. Performance is improved through an external and possiblysaturated controller. Themethodology we propose represents a simple way of designing nonlinearstate feedback controllersfor systems affected by severe nonlinearities and undesired attractorssuch as chemical reactorsand it is particulary attractive for open-loop unstable plants. Itapplies to SISO as well as toMIMO systems and is illustrated through two simulation examples thatinvolve continuous-stirred tank reactors. The first of them consists of the control ofa non-isothermal reactor atthe unstable state, while the second consists of the control of leveland concentration.