Stability Conditions, Parametrizations, and Computer-Aided Decoupling Controller Design\\ for Linear Multivariable Systems

• Main Authors: Tung-Fu Hsieh, 謝東富
• Other Authors: Prof. Ching-An Lin
• Format: Others
• Language: en_US
• Published: 1993
• Online Access: http://ndltd.ncl.edu.tw/handle/87225410456852287121

博士 === 國立交通大學 === 電子研究所 === 81 === We establish in this dissertation some simplified conditions for the closed-loop stability of the linear multivariable unity- feedback system and for the system to remain stable under sensor or actuator failures. We also propose the parametrizations of all stabilizing controllers and the descriptions of all achievable I/O maps. Such controller parametrizations and I/O map descriptions are applied to characterize the set of all decoupling controllers, and lead to simple computational algorithms for the construction ofdecoupling controller, stable decoupling controller, and decoupling controllers that retain the closed-loop stability under sensor failures. By the algebraic property of our analysis, most results in this dissertation can apply to continuous-time systems as well as discrete-time systems. Based on our characterization of decoupling controllers and the corresponding computational algorithm, we develop an optimization-based decoupling control design procedure which can also apply to SISO system design without any modifications. This design procedure is systematic in that the design is improved in each iteration based on a well-defined performance index. It is also practical in that many engineering-level design specifications such as rise time, maximum overshoot, plant input limit, and robust stability can be easily incorporated into the optimization program. By our formulation, there is no equality constraints which are in general hard to achieve in optimization problems. This design procedure has been implemented as an interactive CAD package for use under MATLAB. Two illustrative design examples are also proposed to verify the effectiveness of this design approach.