Design of Decentralized Controller for Interactive Processes through Relative Frequency Array

Abstract

Tuning of controllers for multivariable process is a difficult task because of interaction involved among the variables. In this paper, a simple tuning strategy is used for design of a multi-loop PI controller to achieve desired frequency response for interactive multivariable processes. To handle the loop interaction, a pairing of controlled variable and manipulted variable is determined based on the integrity obtained from effective relative gain (ERG) and Niederlinski’s Index (NI).Then the Equivalent transfer function of model is approximated by the use of relative frequency array (RFA), Relative Gain Array (RGA). Finally a Decentralized PI controller is designed for suggested pair of variables to achieve desired gain, phase margins. The performance is verified on the industrial multivariable processe models to show the effectiveness of the proposed method. The results clearly reveal that, it gives better performance for set point changes and disturbance rejection. Simulation results are included to validate the robustness of the presented algorithm.