Data-driven robust optimal control design for uncertain cascaded systems using value iteration

Abstract

In this paper, a new non-model-based control design is proposed to solve the H∞ control problem for linear continuous-time systems. Our first contribution is to develop a robust control design by combining the zero-sum differential game theory with the gain assignment technique together. Compared with traditional game theory-based approaches, the obtained result allows us to assign arbitrarily the input-to-output L2 gain for a class of continuous-time linear cascaded systems. Moreover, the presence of dynamic uncertainty is tackled using the small-gain theory. Our second contribution is to give a new non-model-based robust adaptive dynamic programming (RADP) algorithm. In sharp contrast to the existing methods, the obtained algorithm is based on continuous-time value iteration (VI), and an initial stabilizing control policy is no longer required. Finally, an example of a power system is adopted to illustrate the effectiveness of the obtained algorithm.