کنترل بهینه نزدیک به حد افق برای یک کلاس غیرخطی غیر زمانبندی غیرافین براساس برنامه ریزی پویا سازگار

In this paper, a novel adaptive dynamic programming (ADP) algorithm is developed to solve the nearly optimal finite-horizon control problem for a class of deterministic nonaffine nonlinear time-delay systems. The idea is to use ADP technique to obtain the nearly optimal control which makes the optimal performance index function close to the greatest lower bound of all performance index functions within finite time. The proposed algorithm contains two cases with respective different initial iterations. In the first case, there exists control policy which makes arbitrary state of the system reach to zero in one time step. In the second case, there exists a control sequence which makes the system reach to zero in multiple time steps. The state updating is used to determine the optimal state. Convergence analysis of the performance index function is given. Furthermore, the relationship between the iteration steps and the length of the control sequence is presented. Two neural networks are used to approximate the performance index function and compute the optimal control policy for facilitating the implementation of ADP iteration algorithm. At last, two examples are used to demonstrate the effectiveness of the proposed ADP iteration algorithm.