On linear robust \({\mathcal H}_\infty\) controllers for a class of nonlinear singular perturbed systems (Q1301400)

The nonlinear singularly perturbed system (NSPS) under consideration is described by: \[ \begin{aligned} \left[\begin{smallmatrix} \dot x_1\\ \varepsilon\dot x_2\end{smallmatrix}\right] & = A(x_1)\left[\begin{smallmatrix} x_1\\ x_2\end{smallmatrix}\right] + B_1(x_1)w+B_x(x_1)u\quad x_1(0)=0,\;x_2(0)=0\\ z & = C_1(x_1)\left[\begin{smallmatrix} x_1\\ x_2\end{smallmatrix}\right]+ D_{11}(x_1)w+D_{12}(x_1)u\\ y & = C_2(x_1)\left[\begin{smallmatrix} x_1\\ x_2\end{smallmatrix}\right]+ D_{21}(x_1)w+D_{22}(x_1)u\end{aligned}\tag{1} \] where \(x=(x_1,x_2)\in\mathbb{R}^n\times\mathbb{R}^m\) is the state, \(w\) is the disturbance, \(z\) is the regulated output, \(y\) is the measured output, \(\varepsilon\) is a small parameter, and \[ \begin{aligned} A(x_1) & = \left[\begin{smallmatrix} A_{11}(x_1) & A_{12}(x_1)\\ A_{21}(x_1) & A_{22}(x_1)\end{smallmatrix}\right]\\ B_i(x_1) & = \left[\begin{smallmatrix} B_{1i}(x_1)\\ B_{2i}(x_1)\end{smallmatrix}\right]\\ C_i(x_1) & = [C_{i1}(x_1)\quad C_{i2}(x_1)],\;i=1,2.\end{aligned}\tag{2} \] The linearized system of Eq. (1) is the following linear singularly perturbed system (LSPS): \[ \begin{aligned}\left[\begin{smallmatrix} \dot{\overline{x}}_1\\ \varepsilon\dot{\overline{x}}_2\end{smallmatrix}\right] & =A(0)\left[\begin{smallmatrix}\overline{x}_1\\\overline{x}_2\end{smallmatrix}\right] + B_1(0)w+B_2(0)u,\quad \overline{x}_1(0)=0,\;\overline{x}_2(0)=0\\ \overline{z} & = C_1(0)\left[\begin{smallmatrix} \overline{x}_1\\ \overline{x}_2\end{smallmatrix}\right]+ D_{11}(0)w+D_{12}(0)u\\ \overline{y} & = C_2(0)\left[\begin{smallmatrix} \overline{x}_1\\ \overline{x}_2\end{smallmatrix}\right]+ D_{21}(0)w+D_{22}(0)u.\end{aligned}\tag{3} \] A state feedback/output feedback controller \(\mu\) independent on \(\varepsilon\) is said to robustly stabilize Eq. (1) if it results in robust internal stability, i.e. there is \(\varepsilon_0\) such that for every \(0<\varepsilon\leq \varepsilon_0\) the system \[ \left[\begin{smallmatrix} \dot x_1\\ \varepsilon\dot x_2\end{smallmatrix}\right] =A(x_1)\left[\begin{smallmatrix} x_1\\ x_2\end{smallmatrix}\right]+B_2(x_1)\mu_{w=0} \] is asymptotically stable with a common domain of attraction. For \(\gamma > 0\), \(\mu\) is said to be a robust \({\mathcal H}_\infty\) controller of Eq. (1) if it robustly stabilizes Eq. (1) and moreover the resulting closed-loop system satisfies the following robust \(\gamma\)-gain condition: there are \(\ddot\gamma<\gamma\) and a neighborhood \({\mathcal U}\subset\mathbb{R}^n\times\mathbb{R}^m\) of zero such that whenever \(0<\varepsilon<\varepsilon_0\) the relation \[ \|z\|_2\leq \ddot\gamma\|w\|_2 \] holds true for every disturbance \(w\in L_2\) such that the corresponding state \((x_1(t),x_2(t))\) of Eq. (1) always remains in \(\mathcal U\). Here \(L_2\) is the space of all integrable functions \(w\) satisfying \[ \|w\|_2:=\left[\int^\infty_0\|w(t)\|^2dt\right]^{1/2}<\infty. \] It is shown in this paper that any state feedback/output feedback robust \({\mathcal H}_\infty\) controller of the linearized singular perturbed system yields a local solution of the nonlinear \({\mathcal H}_\infty\) control problem for the original nonlinear singularly perturbed system.