A model-free approach to approximate the solution of time-varying optimal control based on Bezier control points technique

Aiming at the unsatisfactory effect of conventional control methods on time-varying control systems with unknown models, this paper proposes an on-line finite-time adaptive optimal control algorithm based on a data-driven model. More precisely, the realization of the algorithm relies on the on-policy adaptive dynamic programming and Bezier control point methods, where the Bezier curve is used to approximate the optimal control solution (i.e., the solution of the time-varying differential Riccati equation corresponding to the finite-time optimal control) and the on-policy adaptive dynamic programming is performed in the form of on-line iterative process to approximate successively the solution of the control strategy. At last, a simulation example is given to illustrate the effectiveness of the proposed algorithm.