| In robust control design, in order to incorporate classical control into robust control framework, designers have to use weighting functions or similarity scaling. The process of selecting weighting functions requires a considerable amount of expertise and numerous trial-and-error iterations. Thus, when different performance specifications are involved, the design process will become tedious and insufficient.;Hence, multi-objective robust control problems have been the center of attention lately. In these problems, different performance specifications may be addressed directly in the design without approximations or choices of weighting functions. Thus, trial-and-error iterations could be reduced and the design procedure will be simplified.;This dissertation concentrates on mixed ;However, the proposed method suffers from the common drawback with other robust control theories: high order controllers could result from low-order plants and cause numerical problems. Non-trivial process of model reduction becomes an essential step afterwards. A valuable alternative to model reduction is the Linear Matrix Inequalities (LMI) approach. This dissertation will introduce LMI applications in discrete-time robust control design, as well as the LMI approach (parameterization) of obtaining approximate solutions to mixed ;Finally, mixed robust performance problems will be introduced. Some observations and propositions upon further research work will also be discussed. |
