New Approach on the Automatic Tuning of Multivariable PI Controllers Using Relay Feedback

Abstract

A new algorithm for automatic tuning of multivariable PI controllers for multi-input multi-output processes is presented. The tuning procedure consists of four stages. In the first, the critical point of each possible loop is identified using the conventional relay feedback test. In this phase, a set of critical frequencies is formed. In the second stage, among the process critical frequency set, a design frequency is chosen. The choice of this design frequency is done with respect to the interaction problem in the multivariable system. In the third stage, the process steady state and frequency response matrices at the design frequency are estimated with good accuracy. In the fourth stage, the information about the process is used to tune the multivariable PI controller. For the multivariable PI controller, a new design technique is derived taking into account the decoupling condition at the low- and high-frequency range of the process. The proposed method is applied to various typical processes and experimented on using a laboratory-scaled model of a paper machine head-box. The experimental test has shown good results where significant performance improvement over the existing tuning methods is demonstrated.