LMI robustness analysis for H2 performance of exactly discretized sampled-data models

Although it seems reasonable to conjecture that the closed-loop performance of a sampled-data system improves as processor speed increases, this is not the case. Specifically, although slow sampling degrades the H₂ performance, even slower sampling may be used without significant performance degradation. These results assumed, however, that the plant model is exact. Consequently, in this paper we design controllers based on the nominal plant, and study the effect of input, output, and dynamics uncertainty on the performance of H₂ optimal controllers designed for exactly discretized plants at various sampling rates. We then present a guaranteed cost inequality framework for analysing the robust stability and performance of a sampled-data system, and present an LMI method for solving the robust performance problem