Optimal and suboptimal feedback controls for a class of nonlinear systems (Q1913459)

The authors investigate a nonlinear regulator problem for the nonlinear control system \[ \dot x = A(x) x + B(x) u, \quad x(0) = x_0 \in \mathbb R^n \] where \((A(x), B(x))\) is a stabilizable pair for each \(x\). They consider the quadratic performance index \[ J_\infty = \int^\infty_0 (x^T Qx + u^T Ru)\,dt \] and its optimal feedback solution from [\textit{S. P. Banks} and \textit{K. J. Mhana}, IMA J. Math. Control Inf. 9, No. 2, 179--196 (1992; Zbl 0773.49018)] which is obtained by solving a state dependent algebraic Riccati equation. After discussing the stability properties of the resulting closed loop system the authors turn to the numerical calculation of a suboptimal feedback control. The algorithm proposed solves the corresponding Riccati equation at discrete times along the resulting trajectory and hence yields a discrete feedback law by using the calculated feedback for the next time step. Although no formal proof for the stability of the resulting trajectory is available, several numerical examples illustrate the effectiveness of the algorithm. In the second part, the system is perturbed by additive bounded noise \(B(x) w(x)\) where \(w: \mathbb R^n \to W\) is assumed to be continuous and bounded. Here the (analytical) optimal feedback law from the first part is extended by an additive controller which is constructed from the optimal value function of the unperturbed problem and the bounds on \(w\). For this feedback law it is shown that the resulting closed loop system is uniformly bounded and uniformly ultimately bounded. As an alternative strategy, a Lyapunov function based controller with the same properties is discussed.