Simulation study of multivariable system

Abstract

This dissertation presents sparse control system design method which is based on the concept of energy transmission ratio. The sparse control system structure can help to reduce the coupling among loops and result in a simpler system. By introducing the concepts of REGA (Relative Effective Gain Array) and ETF (Equivalent Transfer Function), the controllers can be easily designed using the existing PI/PID tuning rules. These concepts are also feasible in the decentralized control system. There are altogether ten multivariable system process models studied to evaluate the effectiveness and simplicity of the proposed REGA based decentralized method and sparse control method. These ten process models process different static and dynamic characteristics and vary from 2-by-2 systems up to 4-by-4 ones. By comparison, the conclusion can be made that the sparse control method is very simple and straightforward which can be easily understood and implemented by field control engineers. In addition, the simulation study results also indicate that the REGA based methods used in the decentralized control structure is able to provide quite satisfactory control performances as well when the loop interactions are not so severe.