摘要
建立了5自由度上肢康复机器人的BP神经网络控制模型。在此模型基础上,通过对正常人肌电信号的训练学习,修正了网络权值,得到了较为理想的控制模型。最后,通过病人的肌电信号,得到了良好的输出结果。仿真实例表明,BP神经网络方法比传统方法收敛快,学习精度高,且具有较好的网络泛化能力,可以用于5自由度上肢康复机器人的智能控制。
The back propagation (BP) neural network model of 5-degree of freedom (IX)F) upper limb rehabilitant robot was built. On the basis of the model, the weight of tile network was adjusted through training and studying the electromyography (EMG) signal of normal people and the ideal control model was obtained. At last, the well output effect was gotten by the EMG signal of patients. The simulation instances show that the method of BP neural network converges faster than the conventional method and its accuracy is higher. The capacity of network generalization is better. This method can be used in the intellectual control of 5-DOF upper limb rehabilitant robot,
出处
《石油大学学报(自然科学版)》
EI
CSCD
北大核心
2005年第5期87-90,94,共5页
Journal of the University of Petroleum,China(Edition of Natural Science)
基金
国家高科技研究发展计划资助项目(863-2003AA404060)
关键词
康复机器人
肌电信号
BP神经网络
网络泛化能力
仿真实例
rehabilitant robot
electromyo-graphy sigqaal
back propagation neural network
capacity of network generalization
simulation instance