期刊
IEEE TRANSACTIONS ON CYBERNETICS
卷 45, 期 1, 页码 65-76出版社
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/TCYB.2014.2319577
关键词
H-infinity control design; Hamilton-Jacobi-Isaacs equation; neural network; off-policy learning; reinforcement learning
类别
资金
- National Basic Research Program of China under 973 Program [2012CB720003]
- National Natural Science Foundation of China [61121003]
- General Research Fund project from Science and Technology on Aircraft Control Laboratory of Beihang University [9140C480301130C48001]
- NPRP from Qatar National Research Fund (Qatar Foundation) [NPRP 4-1162-1-181]
The H-infinity control design problem is considered for nonlinear systems with unknown internal system model. It is known that the nonlinear H-infinity control problem can be transformed into solving the so-called Hamilton-Jacobi-Isaacs (HJI) equation, which is a nonlinear partial differential equation that is generally impossible to be solved analytically. Even worse, model-based approaches cannot be used for approximately solving HJI equation, when the accurate system model is unavailable or costly to obtain in practice. To overcome these difficulties, an off-policy reinforcement leaning (RL) method is introduced to learn the solution of HJI equation from real system data instead of mathematical system model, and its convergence is proved. In the off-policy RL method, the system data can be generated with arbitrary policies rather than the evaluating policy, which is extremely important and promising for practical systems. For implementation purpose, a neural network (NN)-based actor-critic structure is employed and a least-square NN weight update algorithm is derived based on the method of weighted residuals. Finally, the developed NN-based off-policy RL method is tested on a linear F16 aircraft plant, and further applied to a rotational/translational actuator system.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据