Suboptimal solution of a cheap control problem for linear systems with multiple state delays (Q2490400)

This paper is devoted to the study of a finite-horizon linear-quadratic optimal control problem with multiple pointwise and distributed state delays in the dynamics. The control cost in the performance index is small with respect to the state cost. More precisely, the author considers the following control system \[ \frac{dz(t)}{dt} = Az(t) + \sum_{i=1}^{N} H_{i}z(t-h_{i}) + \int_{-h}^{0} G(\tau)z(t+\tau)d\tau + Bu(t), \;t \in [0,T], \tag{1} \] where \(z(t)\in \mathbb{R}^{n}\) and \(u(t) \in \mathbb{R}^{r}\), \(n\geq r,\) are a state and a control, respectively; \(N > 0\) is a given integer; \(0 < h_{1}< \dots < h_{N} = h\) are given constants; \(A, H_i, \;i=1,\dots,N,\;G(\tau)\), and \(B\) are given time-invariant matrices of corresponding dimensions; the matrix \(B\) has the full rank \(r\); the matrix-valued function \(G(\tau)\) is piece-wise continuous for \(\tau \in [-h,0]\); and \(T>0\) is the duration of the control process. Two different types of initial conditions are considered. The first one is \((z(0),z(\tau)) = (\phi_{0},\phi(\tau)), \;\tau \in [-h,0)\) for given \(\phi_{0} \in \mathbb{R}^{n}\) and \(\phi \in L^{2}[-h,0].\) The second one is \(z(\tau)= \psi (\tau),\;\tau \in [-h,0]\) for a given \(\psi \in C[-h,0].\) For \(z(t) = (x(t),y(t)),\) where \(x(t)\) and \(y(t)\) are of the dimensions \(n-r\) and \(r\), respectively, the performance of the control process for system (1) is evaluated by the following cost functional to be minimized by a proper choice of the control \(u\); \[ J_{\varepsilon} (u) = \int_{0}^{T} [x'(t)D_{x}x(t) + y'(t)D_{y}y(t) + \varepsilon^{2}u'(t)Mu(t)] \;dt , \] where \(D_{x}\) is a symmetric positive semidefinite matrix, \(D_{y}\) and \(M\) are symmetric positive definite matrices and \(\varepsilon\) is a small parameter. The main purpose of the paper is to construct a suboptimal linear feedback control uniformly valid for all sufficiently small values of \(\varepsilon > 0.\)