基于神经网络的漂浮基空间机械臂的自学习模糊控制  

Self-learning Fuzzy Control of a Space Manipulator Based on a Neural Network

作　　者：洪昭斌[1] 陈力[1] 

机构地区：[1]福州大学机械工程与自动化学院,福州350002

出　　处：《机械科学与技术》2009年第7期976-980,共5页Mechanical Science and Technology for Aerospace Engineering

基　　金：国家自然科学基金项目(10672040)资助

摘　　要：讨论了载体位置、姿态均不受控制情况下,自由漂浮空间机械臂的高斯基神经网络自学习模糊控制问题。由于此类空间机械臂严格遵守动量守恒和动量矩守恒,其动力学方程表现出强烈的非线性性质。结合神经网络和模糊控制,即利用神经网络来实现模糊推理可使模糊控制具有自学习能力,在此基础上,设计了空间机械臂关节空间的高斯基神经网络自学习模糊控制方案。该控制方案的显著优点为,不需要测量、反馈漂浮基的位置、移动速度和移动加速度。In this paper, the self-learning fuzzy control of a free-floating space manipulator based on a neural network is studied. With the momentum conservation and the angular momentum conservation of the system, it is verified that the dynamic equations of the system can not be linearly dependent on the inertial parameters. When the fuzzy control- ler and neural network are combined, and the fuzzy inference is realized by the neural network, the fuzzy controUer can learn by itself. Based on the results, the self-learning controller of a space robot system in the joint space is de- signed. In particular, it doesn＇t require measuring the position, velocity or acceleration of the base. Numerical simu- lation is carried out, and the result confirms that the proposed controller is feasible and effective.

关 键 词：空间机械臂 模糊神经网络 关节空间 

分 类 号：TP241[自动化与计算机技术—检测技术与自动化装置]