Multi-criteria optimization for filter design of L1 adaptive control (Q2247903)

This paper presents heuristic but systematic approaches toward optimal filter design for the \(\mathcal L_1\) adaptive controller based on the search of the space of design parameters. At first, an overview on \(\mathcal L_1\) adaptive controllers is given. The considered systems are described by ODEs, linear in the state, with a one-dimensional control signal and unknown n-dimensional parameters. Similar to these problems, the performance and the time-delay margin are to optimize. These problems are described as optimization problems. They are multi-criteria filter design problems. Unfortunately, the problems are not convex in the \(\mathcal L_1\)-norm. Unconstrained optimal filter design and constrained multi-criteria optimal filter design are studied. Heuristic design methods are proposed. Two examples illustrate the steps of the design procedure. An outlook of extending the presented methods to more complex systems is given.