A linear matrix inequality approach to synthesizing low-order suboptimal mixed \(\ell_1/{\mathcal H}_p\) controllers (Q1571076)

The authors propose a two stage design process based on a linear matrix inequality approach which enables parallel optimization of \(\ell_1\) and \(H_2\) (or \(H_\infty\)) criteria for discrete-time linear systems. The main idea is to use some upper bounds for the relevant norms and to modify the problem in such a way that this leads to feasible solutions. This modification leads to the design of suboptimal controllers which contain a single parameter. The proper value of the parameter can be obtained via a line search over the unit interval. The authors show that in the case of perfect state measurement the controllers ensuring optimal performance are static while in the case of output feedback the controller is of the same order as the plant. The results are illustrated by a simple example.