Comparison of decentralized pid controller tuning methods.

In controlling multivariable systems exist in most industrial processes, multivariable control methods are implemented to achieve desirable performances. Control strategy such as the decentralized control is usually preferred for multivariable system since less complexity is encountered as compared...

Full description

Saved in:
Bibliographic Details
Main Author: Choo, Wei Qiang.
Other Authors: Vinay Kumar Kariwala
Format: Final Year Project
Language:English
Published: 2009
Subjects:
Online Access:http://hdl.handle.net/10356/15667
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Nanyang Technological University
Language: English
Description
Summary:In controlling multivariable systems exist in most industrial processes, multivariable control methods are implemented to achieve desirable performances. Control strategy such as the decentralized control is usually preferred for multivariable system since less complexity is encountered as compared to fully centralized controller. The aim of this report is to present several analytical decentralized Proportional-Integral-Derivative (PID) controller tuning methods which are simple, effective and result in good closed-loop behavior of multivariable systems. The various methods considered are independent method such as the Effective Transfer Function (ETF) method, detuning method such as Biggest Log-modulus Tuning (BLT) method and sequential loop closing method such as Sequential SIMC method. Based on total variation (TV), integral of absolute error (IAE) and peak of closed-loop sensitivity function, the performance of the different tuning methods are evaluated and compared where setpoint tracking, disturbance rejections and robustness were considered respectively. Through the investigation of the results of controller performances for 2 x 2, 3 x 3 and 4 x 4 systems, recommendations are provided for selecting the optimal design method. The report also includes the algorithms implemented in MATLAB® which enable controllers design parameters to be determined efficiently, thus reducing computational time for future controller performance studies done on other systems.