Error Reduction in Data-Based Control Synthesis:Tests and Comparisons of Various Design Techniques

• Main Authors: Yi-Jia Chan, 陳奕嘉
• Other Authors: Jenq-Tzong Hermann Chan
• Format: Others
• Language: zh-TW
• Published: 2006
• Online Access: http://ndltd.ncl.edu.tw/handle/29427850791241730314

碩士 === 國立成功大學 === 航空太空工程學系碩博士班 === 94 === The traditional linear quadratic (LQ) design faces two problems : the system state must be measured at alltime and the system model need to be estimated with inevitable estimate error. The data-based linear quadratic (DBLQ) design has solved these problem. However during the process of data-based controller synthesis (DBCS), if the plant experimental data has been corrupted with noise signal, the well being of the synthesized controller will be in trouble.Therefore, it is important that we filter out the noise signal before the data is need for the data-based controller design. The purpose of this paper is to get one set of experimental data, filtering it through a generalized auto-regression sequence annihilator (GARSA), and then processing the filted data through DBCS. The merit of the GARSA-based noise filter is measured by comparing the resulting design with the designs which uses un-filted test data.