A unified approach to deal with multivariable constrained process control problems

GARELLI, F.; MANTZ, R.; DE BATTISTA, H.

Advances in Mechanical Engineering Research

Nova Science Publishers

Lugar: New York; Año: 2011; p. 255 - 273

In every real control loop exist physical limits, security bounds or system dynamicbehaviors that constrain the reachable closed-loop performance. In particular, physicaland/or technological limitations of actuators give rise to plant input constraints, whilstsafety operation regions or non-minimum phase characteristics generally affect theevolution of the controlled variables or system outputs. In a multivariable or MIMO(Multiple-InputMultiple-Output)process, the effects of these constraints are worsenedbecause of the presence of directions associated to input/output vectors and, even moreimportant, the crossed coupling or interactions between the system variables.This chapter deals with some relevant practical problems of multivariable processcontrol, which are consequence of crossed interactions, control directionality and in-put or output constraints. To this end, recently proposed control strategies are unifiedin a generalized framework to deal with both input and output constraints. The re-sulting control strategy, which is herein called sliding mode reference conditioning(SMRC) technique, makes use of variable structure control (VSC) and sliding modes(SM) related concepts in order to prevent the process from violating its restrictionswhile minimizing their undesired effects. Sliding regimes are just transiently em-ployed by SMRC over bounding surfaces corresponding to the unavoidable processconstraints, and they are confined to an auxiliary reference conditioning loop in thelow-power side of the system. This way of operation permits accomplishing distinc-tive properties and design methodologies with respect to the conventional applicationsof VSC and SM.The chapter is outlined as follows. Firstly, the basic ideas behind SMRC are intro-duced for systems involving biproper transfer function descriptions, such as propor-tional or proportional-integral industrial controllers. Then, the methodology analysisand design is extended to deal with strictly-proper systems. This general analysis isperformed independently of the type of constraint. From then on, some applications orcase studies of practical interest in multivariable control are discussed and results arepresented: (i) the preservation of control directionality and full dynamic decouplingeven in the presence of actuator saturation, (ii) the limitation of remaining couplingin partial decoupled non-minimum phase processes, and (iii) a way of delimiting thecrossed interactions in industrial decentralized or multi-loop control of multivariableprocesses.