Block Decoupling Control of Linear Multivariable Systems

博士 === 國立交通大學 === 電機與控制工程系 === 88 === In this thesis, we consider the unity-feedback configuration for the linear multivariable systems. We discuss the conditions for existence of block decoupling controllers and design the parameterization of all block decoupling controllers and propose block decou...

Full description

Bibliographic Details
Main Authors: Chang-Ming Wu, 吳章銘
Other Authors: Ching-An Lin
Format: Others
Language:en_US
Published: 2000
Online Access:http://ndltd.ncl.edu.tw/handle/84339298083341480240
Description
Summary:博士 === 國立交通大學 === 電機與控制工程系 === 88 === In this thesis, we consider the unity-feedback configuration for the linear multivariable systems. We discuss the conditions for existence of block decoupling controllers and design the parameterization of all block decoupling controllers and propose block decoupling precompensators. Under the assumption of the plant without unstable pole-zero coincidence, the literature proposed that the unity-feedback system exists a block decoupling controller. Thus, we propose how to judge the existence of block decoupling controllers under the plant with unstable pole-zero coincidences. If the plant has simple unstable pole-zero coincidences, it is easy to verify that the plant exists a block decoupling controller. In our dissertation, we design the parameterization of all block decoupling controllers under the assumption of the plant without unstable pole-zero coincidence. Another method to design block decoupling controllers is to first design a block decoupling precompensator, then design the feedback controllers for each of the block channel. The cascade connection of the block decoupling precompensator and the plant is block diagonal, and there is no unstable pole-zero cancellation between the precompensator and the plant. Since the block decoupling precompensator separate the plant into several smaller dimension subplants, it will reduce the complexity to design large dimension stabilizable controllers. We use the parameterization of block decoupling controllers to design optimal weighted sensitivity function problems and use the block decoupling precompensator to analyze the cost of block decoupling.