Bifurcation and Normal Form of Nonlinear Control Systems, Part II

From MaRDI portal





This is a continuation of a Part I [ibid. 36, No. 1, 193-212 (1998); see the review above]. The paper is concerned with smooth, SISO affine control systems in \(\mathbb{R}^n\), depending on a real parameter \(\mu\), of the form \[ \dot x= f(x,\mu)+ g(x,\mu) u,\tag{\(*\)} \] with \(f(0,0)= 0\), \(g(0,0)\neq 0\), subject to a \(\mu\)-dependent static state feedback. As in the first part of the paper, it is assumed that the linear approximation \(A= {\partial f(0,0)\over\partial x}\), \(B= g(0,0)\) of the system \((*)\) has the Kalman matrix \([B,AB,\dots, A^{-1}B]\) of rank \(n-1\). Analogously, three questions are studied in this paper: bifurcation diagrams of the equilibria \[ E= \{(x,\mu)\mid(\exists u\in\mathbb{R})(f(x,\mu)+ g(x,\mu) u= 0\}, \] controllability of the linear approximation to the system \((*)\) at the points belonging to \(E\), and the local feedback stabilizability of the system \((*)\). Again, all the results presented in the paper are based on the quadratic normal forms of the control system \((*)\) (Theorem 2.2): \[ \dot z=\lambda z+ \sum^{n-1}_{i=1} \gamma_{x_ix_i} x^2_i+ \gamma_{zx_1} xz_1+ \gamma_{x_1\mu} x_1\mu+ \gamma_{z\mu} z\mu+ O^3, \] \[ \dot x= A_2x+ B_2u+ f_2(x)+ O^3, \] \[ \dot z= \mu+\sum^{n-1}_{i=1} \gamma_{x_ix_i} x^2_i+ \gamma_{zx_1}zx_1+ \gamma_{x_1\mu} x_1\mu+ \gamma_{zz} z^2+ O^3, \] \[ \dot x= A_2x+ B_2u+ f_2(x)+ O^3, \] \[ \dot z= \sum^{n-1}_{i=1} \gamma_{x_ix_i} x^2_i+ \gamma_{zx_1} zx_1+ \gamma_{x_1\mu} x_1\mu+ \gamma_{z\mu} z\mu+ \gamma_{zz} z^2+ \gamma_{\mu\mu} \mu^2+ O^3, \] \[ \dot x= A_2x+ B_2u+ f_2(x)+ O^3. \] where \(z\in\mathbb{R}\), \(x\in\mathbb{R}^{n-1}\), \((A,B)\) are in the Brunovsky canonical form, \(f_2(x)= (f_{21}(x),\dots,\) \(f_{2n-3}(x), 0,0)\), along with \(f_{2i}(x)= \sum^{n-1}_{j= i+2} a_{in-j} x^2_j\), and \(O^3\) denotes terms of order greater than or equal to 3 in \(x\), \(z\), \(\mu\), \(u\). The constants \(\gamma_{x_ix_i}\), \(\gamma_{zx_1}\), \(\gamma_{x_1\mu}\), \(\gamma_{z\mu}\), \(\gamma_{\mu\mu}\), \(\gamma_{zz}\), \(a_{ir}\) have been defined explicitly, and proved to be quadratic invariants of the static state feedback (Theorem 2.4). Relying on these normal forms and invariants the bifurcation diagrams of equilibria have been classified (Theorems 3.1, 4.1, 5.1) as follows: a smooth surface \((\lambda\neq 0)\), a paraboloid (\(\lambda=0\), \(\gamma_{zz}\gamma_{x_1x_1}- 1/4\gamma^2_{xz_1}> 0\)), a saddle (\(\lambda=0\), \(\gamma_{zz}\gamma_{x_1x_1}- 1/4\gamma^2_{zx_1}< 0\)), an isolated point (\(\lambda=0\), \(Q\) sign-definite), and a cone (\(\lambda= 0\), \(Q\) of full rank, sign-indefinite). The matrix \[ Q= \left[\begin{matrix}\gamma_{zz} & 1/2\gamma_{zx_1} & 1/2\gamma_{z\mu}\\ 1/2\gamma_{zx_1} &\gamma_{x_1x_1} & 1/2\gamma_{x_1\mu}\\ 1/2\gamma_{z\mu} & 1/2\gamma_{x_1\mu} & \gamma_{\mu\mu}\end{matrix}\right] \] contains quadratic invariants. Furthermore, the theorems mentioned above include necessary and sufficient conditions for controllability of the linear approximation to the system \((*)\) at the points of \(E\). For a curve in \(E\) consisting of points at which the system \((*)\) has uncontrollable linear approximation, and under the assumption that \(\gamma_{zz} \gamma_{x_1x_1}- 1/4\gamma^2_{zx_1}\neq 0\), \(\gamma^2_{x_1x_1}+ \gamma^2_{zx_1}\neq 0\), a stabilizability result has been proved (Theorem 6.1).




Cited in
(33)








This page was built for publication: Bifurcation and Normal Form of Nonlinear Control Systems, Part II

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4388911)