On the discrete-time normal form

From MaRDI portal





The paper studies analytic, discrete-time control systems of the form \[ x(k+1)= F(x(k), u(k)),\tag{\(*\)} \] \(x\in \mathbb{R}^n\), \(u\in\mathbb{R}^p\), equipped with a \(p\)-dimensional analytic output map \(y(k)= \varphi(x(k))\) such that the relative degree \(r= (r_1,r_2,\dots, r_p)\) of the system is well defined around the equilibrium point \((0,0)\) of \((*)\). The existence of a state space diffeomorphism and a static state feedback is examined under which the control system can be transformed locally to the following normal form \[ \xi_{i1}(k+1)= \xi_{i2}(k),\dots, \xi_{ir_i- 1}(k+1)= \xi_{ir_i}(k),\;\xi_{ir_i}(k+1)= \psi_i(\xi(k), \eta(k),u(k)), \] \[ \varphi_i(k)= \xi_{i1}(k),\quad i= 1,2,\dots, p\quad\text{and}\quad \eta(k+ 1)= \gamma(\xi(k), \eta(k)). \] The independence of the unobservable dynamics \(\gamma\) on the control \(u\) is characteristic. Necessary and sufficient conditions for the existence of this normal form have been derived in a geometry (Theorem 1) or an algebraic (Theorem 2) form. Both forms are equivalent, and require integrability of a certain distribution or a codistribution associated with the control system \((*)\).











This page was built for publication: On the discrete-time normal form

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4506581)