Notes on \(l^ 1\)-optimal control (Q4028753)

[For the entire collection see Zbl 0747.00043.] These notes are an introduction to robust \(l^ 1\)-optimal control theory. In the design of control systems, there are two main sources of uncertainty, signal uncertainty and plant uncertainty. A design theory which takes both into consideration and attempts to minimize, over the class of uncertain signals, the maximum, over the class of uncertain plants, of a performance functional is called a robust design theory and the resulting control systems are called robust systems. The problem discussed in these notes begins with a linear discrete-time, time-invariant plant, subjected to bounded inputs (\(l^ \infty\) signals) and a performance functional which is the \(l^ \infty\)-induced norm of the system error. When the closed-loop system, consisting of plant and compensator, is linear and time-invariant this \(l^ \infty\)-induced norm is the \(l^ 1\) norm of the error impulse response. Minimization of this norm is discussed in detail. Next, plant uncertainty is considered and some preliminary results are presented on the problem of maintaining system performance over all signal and plant uncertainties. This is the robust performance problem and is the fundamental problem in control system design. The results obtained are similar to the \(\mu\)-synthesis results of Doyle for problems in which the performance functional is the \(H_ \infty\) norm.