Cacaded H-infinity Design for Doubly Coprime Factorization Disturbance Observer and Controller Applied to Non-minimum Phase MIMO Systems

碩士 === 國立交通大學 === 機械工程系所 === 103 === In this thesis, one provides an approach which can be used in non-minimum phase and MIMO systems based on cascade design method. In mechanical control, we can eliminate the disturbance at low-frequency and resist the uncertainty at high-frequency by using distu...

Full description

Bibliographic Details
Main Authors: Yen, Chia-Te, 嚴家德
Other Authors: Lee, An-Chen
Format: Others
Language:zh-TW
Published: 2015
Online Access:http://ndltd.ncl.edu.tw/handle/26taua
Description
Summary:碩士 === 國立交通大學 === 機械工程系所 === 103 === In this thesis, one provides an approach which can be used in non-minimum phase and MIMO systems based on cascade design method. In mechanical control, we can eliminate the disturbance at low-frequency and resist the uncertainty at high-frequency by using disturbance observer. The classical DOB has some constrains when it is used because it only can be applied to minimum phase. Therefore, doubly coprime disturbance observer (DCFDOB) was proposed a few years ago, it can be used in non-minimum phase systems. The approach which is proposed in this thesis, it transforms the inner-loop disturbance observer and outer-loop controller problems into two mixed-sensitivity problems, called improving cascade design method. This method can help find a DOB with maximum bandwidth, it has more efficiency of eliminating disturbance and good robustness. Also, the outer controller has allowable response under robustly stable. By using DCFDOB to replace the classical DOB, it can be used more widely in mechanical control, e.g. on non-minimum phase and MIMO systems. Besides, there are some methods which can be applied to non-minimum phase systems, arranged in this thesis. To test the design results, using MATLAB / Simulink to simulate the design parameter and to run the experiments after getting a set of DOB and controller. As a result, the DOB and controller by this approach eliminates disturbance more effectively and keeps tracking performance under robust stability.