二自由度飞行姿态模拟器的模糊强化学习控制  

Fuzzy learning controller design of 2-DOF flight attitude simulator

作　　者：任立伟 班晓军[1] 吴奋 黄显林[1] REN Li-wei;BAN Xiao-jun;WU Fen;HUANG Xian-lin(Center for Control Theory and Guidance Technology,Harbin Institute of Technology,Harbin 150001,China;Department of Mechanical and Aerospace Engineering,North Carolina State University,Raleigh 27695-7910,USA)

机构地区：[1]哈尔滨工业大学控制理论与制导技术研究中心,哈尔滨150001 [2]Department of Mechanical and Aerospace Engineering,North Carolina State University,Raleigh 27695-7910,USA

出　　处：《电机与控制学报》2019年第11期127-134,共8页Electric Machines and Control

基　　金：国家自然科学基金(61304006,61273095)

摘　　要：针对二自由度飞行姿态模拟器的姿态稳定问题,依据强化学习中的策略迭代算法设计姿态稳定控制器。将策略迭代学习算法与多项式T-S模糊系统相结合,对控制器参数进行学习调整,实现对二自由度飞行姿态模拟器姿态稳定控制性能的优化。通过多项式T-S模糊模型对执行器的策略函数以及评价器的值函数进行逼近,建立基于多项式T-S模糊模型的执行器-评价器结构,经过策略迭代过程,学习得到最优控制器参数,使得值函数最小。通过仿真验证,证明了基于多项式T-S模糊模型的执行器—评价器结构的策略迭代算法在飞行器姿态稳定控制方面的有效性。Aiming at the attitude stabilization problem of two-degrees-of-freedom flight attitude simulator,an attitude stabilization controller was designed based on the policy iteration algorithm in the reinforcement learning.The policyiteration learning algorithm and the polynomial T-S fuzzy systems were combined together,conducting parameters′adjustment of the controller,and achievingthe optimization of the attitude stability control performance of the two-degrees-of-freedom flight attitude simulator.By approximating the policy function of the actor and the value function of the critic with the polynomial T-S fuzzy models,the actor-critic structure based on the polynomial T-S fuzzy models was established.Through the policy iteration process,the optimal parameters of the controller were learned to minimize the value function.The simulation results show that the policy iteration algorithm based on polynomial T-S fuzzy models is effective in controlling aircraft attitude stabilization.

关 键 词：飞行器控制 姿态稳定 强化学习 策略迭代算法 多项式T-S模糊系统 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]