Simulation of controls

Ordinary differential inclusions (34A60) Numerical methods in optimal control (49M99) Numerical optimization and variational techniques (65K10) Controllability (93B05) Control/observation systems governed by ordinary differential equations (93C15)

Review text

This paper is devoted to the study of the controlled differential inclusion \[ (1)\quad \dot x\in F(t,x,u),\quad x(t_ 0)=x_ 0,\quad x\in R^ n,\quad u\in P\subset R^ m \] with a compact and convex set P. It is assumed that the values of F are compact, convex and uniformly bounded sets, that F is upper-semicontinuous in (t,x), continuous in u and that for every integrable control u(\(\cdot)\) the set of all trajectories of (1) (for \(u=u(\cdot))\) is non-empty and all these trajectories lie in a fixed compact. The problem is the following one: during the process we measure at each instant t the state x(t) of the system (1), this observation is available with a small error as a signal \(y_ h(t)\) and we must construct an algorithm which restores the approximation \(u_ h(\cdot)\) of the control u(\(\cdot)\) defined on [0,t] generating the motion x(\(\cdot)\) on [0,t]. The algorithm must be regularizing in the following sense: if \(h\to 0\) then \(u_ h(\cdot)\) tends to the set \(U_ h\) of those controls which really generate x(\(\cdot).\) Two solution algorithms for the above problem are proposed and the corresponding controls \(u_ h(\cdot)\) are computed as piecewise-constant functions. Those algorithms are stable under information noise and are intended for real-time operations. The results are applicable to systems described by differential equations with discontinuous right-handsides when the solution is understood in Filippov's sense.

Recommendations

Cited in

(6)

This page was built for publication: Simulation of controls

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1105192)