An Integral Invariance Principle for Differential Inclusions with Applications in Adaptive Control (Q4388943)

The following extension of the invariance principle by \textit{C. I. Byrnes} and \textit{C. F. Martin} [IEEE Trans. Autom. Control 40, No. 6, 983-994 (1995; Zbl 0831.93058)] for differential inclusion is obtained. Theorem: Let \(x:[0,\infty)\to \mathbb {R}^n\) be a maximal solution of Cauchy problem \[ \dot{x}(t)\in X(x(t)),\qquad x(0)=x^0, \] such that \(x(t)\in U\subset G\) for all \(t\geq 0\). Here \(X(\cdot)\) is a compact-valued multifunction. Assume that \(l:G\to R\) is a lower semi-continuous scalar function which is nonnegative on \(U\subset G\). If \(l(x(t))\in L^1([0,\infty))\), then the trajectory \(x(t)\) approaches the largest weakly-invariant set in \(\Sigma = \{z\in \text{cl} U\cap G: l(z)\leq 0 \}\), as \(t\to\infty\). As implications of this general result, stability of discontinuous feedback control is demonstrated for an adaptive system, for an adaptive servomechanism and for a dynamically perturbed system characterized by a differential equation of first or second order.