ABSTRACT
In this chapter, the control of linear multi variable systems is considered. First, the design of a MIMO control system is reduced to several SISO d& sign problems. The relative gain array (RGA) method aims at helping to choose suitable pairs of control and controlled variables. If the interactions between the variables axe strong, the system may not be satisfactorily controlled by SISO controllers only. In this case the interactions can be actively reduced by decouplers and the control of the decoupled system can then be designed using SISO methods. Decoupling is considered in the second section, and a simple multivariable PI controller (MPI) based on decoupling on both low and high frequencies is presented. The third approach considered in this section is a 'true' multivariable control approach. The design of a multivariable generalized predictive controller (MGPC) is considered, which solves the MIMO control design problem by minimizing a quadratic cost function. Simulation examples conclude this chapter.
