摘要
为了改善残疾人生活水平和促进医疗事业发展,提出一种基于神经网络的假肢无线控制系统设计方案。该系统以STM32为核心芯片,通过采集上肢肱二头肌、肱三头肌、指浅屈肌、指伸肌4块肌肉的肌电信号,使用BP神经网络与SOFM神经网络相结合对肌电信号进行模式识别,实时控制肌电假肢的完成伸臂、屈臂、腕内旋、腕外旋、握拳、张手6种动作行为。实验结果表明,该系统对6种动作的整体识别率可达97%,并且采用无线实时的控制方式,能够更方便地帮助部分肢体残疾患者完成这些基本的操作行为。
To improve the living standard of the disabled and promote the development of medical treatment cause,a design scheme of prosthesis wireless control system based on neural network is proposed. With STM32 as the core chip of the system,BP neural network and SOFM neural network are combined for pattern recognition of electromyography(EMG) signal which are collected from four muscles of biceps,triceps,superficial flexor and extensor. Six motions of arm stretching,arm bending,wrist pronating,wrist supinating,fist clenching and hand opening can be accomplished by means of real-time control of myoelectric prostheses. The experimental results show that the overall recognition rate of the system for the six motions can reach to 97%,and the wireless real-time control pattern can help disabled patients who lose part of their limbs perform these basic operations more conveniently.
出处
《现代电子技术》
北大核心
2018年第2期63-67,共5页
Modern Electronics Technique
基金
国家自然科学基金重大项目(61534003)
国家自然科学基金面上项目(81371663)
江苏高校品牌专业建设工程资助项目(PPZY2015B135)资助~~
关键词
神经网络
肌电信号
模式识别
STM32
无线控制
肌电假肢
neural network
EMG signal
pattern recognition
STM32
wireless control
myoelectric prosthesis
