基于遗传算法优化神经网络的表面肌电信号识别  

Identification for Surface Electromyography Signal of Optimized Neural Network Based on Genetic Algorithm

作　　者：耿丽清[1] 袁国顺 

机构地区：[1]天津职业技术师范大学自动化与电气工程学院天津市信息传感与智能控制重点实验室,天津300222

出　　处：《工业控制计算机》2013年第2期96-97,共2页Industrial Control Computer

摘　　要：表面肌电信号是人体运动时肌肉、神经活动发出的生物电信号在体表的表现,目前已成为对多自由度假肢理想控制的信号源,BP神经网络在肌电信号源模式识别上被应用广泛,但存在如学习收敛速度慢、不能保证收敛到全局最优点等缺点,鉴于以上缺点设计了基于遗传算法的BP神经网络对肌电信号进行模式识别,较好的改善BP神经网络缺点,提高了识别的准确率。Surface Electromyography （SEMG）is a bio-electric signal emitted by the human motion muscle and nerve activity.It is the performance of the body surface,which has become a multi-degree of freedom prosthetic ideal control signal source. BP neural network is widely used in the pattern recognition of SEMG signal source,but the existence of such learning convergence rate is slow and can not be guaranteed to converge to the global optimum.In view of the above shortcomings. Designed in this paper a new method SEMG signal pattern recognition method based on genetic algorithm BP neural network,it is better to improve BP neural network shortcomings and improve recognition accuracy.

关 键 词：表面肌电信号 BP神经网络 遗传算法 

分 类 号：TN911.7[电子电信—通信与信息系统] TP183[电子电信—信息与通信工程]