表面肌电采集与在线识别系统

Collection and Online Identification System of sEMG

摘要:

设计了一套简易且分辨率高的表面肌电采集与在线识别系统；系统硬件部分包括信号两级放大、带通滤波、精密整流、16位AD转换芯片ADS1120、AVR单片机等部分;软件部分基于JAVA编程,具有实时滤波、显示并存储肌电信号、在线识别手部动作等功能；系统放大增益倍数为100～2 500可调,根据不同被试同一动作的肌电信息,微调放大倍数以减少个体差异；当放大倍数为1 000倍时,识别精度达0.3 μV；此外还设计了训练范式,根据被试的训练数据提取在线识别算法的参数,以提高识别准确率；实验结果表明:该系统具有较好的稳定性,能够准确识别四类手部动作,平均识别率达84.37%。

Abstract:

The system is designed to real-time acquire and online identify high-resolution sEMG signals conveniently. The hardware section consists of a two-stage amplifier, a band pass filter, a precision rectifier, a 16-bit analog to digital converter ADS1120 and an AVR microcontroller. The software section is programmed by Java. It realizes the functions of real-time filtering, displaying and storing sEMG signals, online indentifying the hand motion. The gain of the amplifier is adjustable in the range of 100~2 500.The resolution of the system reaches 0.3 μV when the gain is set to 1 000. What’s more, a paradigm is designed to train the subjects before online identification to improve the classification rate. The experimental results showed that the system has good stability, which can identify the four types of hand motion accurately, and the average recognition rate is 84.37%.