## Delay-dependent robust stabilization of uncertain systems with multiple state delays.(English)Zbl 0973.93043

Consider uncertain time-delay systems described by the following state equations $\dot x(t)= [A_0+\Delta A_0(x, t)] x(t)+ [B+\Delta B(x,t)] u(t)+ \sum^l_{i=1} [A_i+\Delta A_i(x, t)] x(t- d_i(t)),\tag{1}$
$0< d_i(t)\leq \tau_i,\quad i= 1,\dots, l,$
$\Delta A_i(x, t)= D_iF_i(x, t) E_i,\quad \Delta B(x, t)= D_0 F_0(x, t) E_0,$ where $$F_j(x,t)$$ are unknown real time-varying matrices with Lebesgue measurable elements bounded by $F^T_j(x, t) F_j(x, t)\leq I\quad\text{for all }t.$ Theorem: Consider the uncertain delay system (1) with $$u(t)\equiv 0$$. Given scalars $$\tau_i$$ and $$\tau$$ satisfying $$0< \tau_i\leq \tau$$, then for any time delays $$0< d_i(t)\leq \tau_i$$, this system is robustly stable if there exists matrices $$X>0$$, $$P_{ij}> 0$$, and scalars $$\varepsilon_i> 0$$, $$d_j> 0$$, $$\rho_{ij}> 0$$ satisfying the following $S_1(\tau_i):= \left[\begin{matrix} S_{10} & H_1 & H_2\\ H^T_1 &-J_1 & 0\\ H^T_2 & 0 & -J_2\end{matrix}\right]< 0,\quad \left[\begin{matrix} X & XA^T_j & XE^T_j\\ A_jX & P_{ij}- \rho_{ij}D_i D^T_j & 0\\ E_j X & 0 & \rho_{ij} I\end{matrix}\right]> 0,$ where $S_{10} = \sum^l_{i=0} A_iX+ X \sum^l_{i=0} A^T_i+ \sum^l_{i=0} \alpha_i D_i D^T_i+ \sum^l_{i=1} \tau_i\varepsilon_i D_i D^T_i+ \sum^l_{i=1} \tau_i A_i W_i A^T_i+ \sum^l_{i=1} (l+ 1) \tau_i X,$
$W_i = \sum^l_{j=0} P_{ij},\quad H_1= [XE^T_0, XE^T_1,\dots, XE^T_l],\quad J_1= \text{diag}(\alpha_0 I,\alpha_1I,\dots, \alpha_lI),$
$H_2 = [\tau_1 A_1 W_1 E^T_1, \tau_2 A_2 W_2 E^T_2,\dots, \tau_l A_l W_l E^T_l],$
$J_2 = \text{diag}(\tau_1(\varepsilon_1 I- E_1 W_1 E^T_1),\;\tau_2(\varepsilon_2 I- E_2 W_2 E^T_2),\dots, \tau_l(\varepsilon_l I- E_l W_l E^T_l)).$ Moreover, a stabilizing control law has the form $$u(t)= YX^{-1} x(t)$$ for a suitable $$Y$$.

### MSC:

 93D21 Adaptive or robust stabilization 93C23 Control/observation systems governed by functional-differential equations 93D09 Robust stability
Full Text: