On the choice of methods for stabilizing systems containing a linear delay (Q6132612)

The authors study asymptotic stabilization of the family of delay-differential equations governed by \[ \dot{x}(t) = A x(t) + B x (t - \sigma) + D x(\mu t) + R \dot{x}(\mu t) + Mu(t) \] where \(\sigma > 0\), \(0 < \mu < 1\), \(x(t) \in \mathbb{R}^n\) is the state variable, \(u(t) \in \mathbb{R}^m\) is the input and \(A\), \(B\), \(D\), \(R\), \(M\) are constant real matrices of appropriate dimensions. They first consider the case where \(B = 0\); using the fact that \(\dot{x}(t) = A x(t) + D x(\mu t)\) is asymptotically stable when the continuous-time system \(\dot{x}(t) = A x(t)\) and the discrete-time system \(Az(t+1) = D z(t)\) are asymptotically stable, they show that under some simple controllability conditions, a linear control \(u(t) = -K_1 x(t) - K_2 x(\mu t) - K_3 \dot{x}(t)\) can asymptotically stabilize the system. The matrix gains \(K_i\) are chosen by solving a sequence Riccati equations, but alternative methods such as pole assignment could also be used. Then, they consider the case where \(B \neq 0\) but \(R = 0\) and achieve asympotic stabilization by controlling a reduced-order system. Finally, they leverage the result to handle the general case and present a numerical example of the method in the general case (with non-zero matrices \(A\), \(B\), \(D\), and \(R\)).