人体上肢表面肌电信号采集与处理的研究

发布时间：2018-04-30 14:01

本文选题：50Hz陷波器 + 小波多分辨分析　；参考：《东北大学》2012年硕士论文

【摘要】：人体表面肌电信号是一种微弱的、复杂的生物医学信号,是由肌肉收缩而伴随产生的,而肌肉的收缩是由人体神经所控制,所以,对表面肌电信号的分析可以反映出神经控制肌肉运动变化趋势,进而在临床诊断、康复医疗等领域进行研究。本文基于“985工程”项目“康复机器人关键技术的研究与应用”,针对表面肌电信号的采集系统的搭建、表面肌电信号的预处理以及表面肌电信号的特征提取和分类问题,做了深入的研究,成功的搭建了表面肌电信号的采集系统；解决了表面肌电信号噪声去噪问题；成功的设计了表面肌电信号模式分类器,并得到了较高的动作识别结果。本文设计应用的表面肌电信号采集和处理系统是一个十通道多参数的生理信号采集设备(FlexComp Infiniti),配合Biograph Infiniti软件对采集到的表面肌电信号进行处理。通过多次的信号采集实验,证明此系统能够很好地采集和处理表面肌电信号。在表面肌电信号的采集过程中,常常伴随着诸多的干扰信号,其中除了主要的50Hz工频干扰外,也夹杂着其它噪声信号,需要对信号进行去噪处理。 Biograph Infiniti软件自带与MATLAB软件的接口程序。本文应用MATLAB仿真软件设计了IIR50Hz陷波器和FIR50Hz陷波器,在尽量不影响有用信号中50Hz频率信号的情况下,对信号中夹杂的50Hz干扰进行去除。经过实验验证,证明FIR50Hz滤波器更能有效的去除表面肌电信号中的50Hz工频干扰。本文在去除表面肌电信号的其它噪声方面,首先运用小波多分辨分析对信号进行阈值去噪。并将小波去噪和数字滤波器的基本理论相结合,构造出了基于小波变换的数字滤波器。经过实验验证,证明基于小波变换的数字滤波器能够更好的去除表面肌电信号的其它噪声。此外,本文在对表面肌电信号去噪研究后,还对经过去噪的信号进行了处理。通过小波变换对信号进行特征提取,并且应用BP神经网络对上肢四种运动进行模式分类,得到了识别率较高的实验结果。通过本文的研究,提出了一套可以对肌电信号进行采集和分析的算法和方案。在未来的工作中,将继续对各种特征处理的方法进行研究,找出更适合肌电信号的方法,同时也尝试不同分类器对分类性能的影响,提高系统性能,并将其应用于康复机器人中。
[Abstract]:The surface EMG signal is a weak, complex biomedical signal that is accompanied by muscle contraction, which is controlled by the human body's nerves, so, The analysis of surface EMG signals can reflect the trend of neuro-controlled muscle movement and then be studied in the field of clinical diagnosis and rehabilitation. Based on "Project 985", "Research and application of key technology of rehabilitation robot", this paper aims at the construction of surface electromyography acquisition system, pretreatment of surface electromyography signal and feature extraction and classification of surface electromyography signal. In this paper, the acquisition system of surface EMG signal is set up successfully; the problem of noise denoising is solved; the mode classifier of SEMG signal is designed successfully, and the result of motion recognition is obtained. The surface EMG signal acquisition and processing system designed and applied in this paper is a 10-channel multi-parameter physiological signal acquisition equipment, FlexComp Infinitig, which is used to process the collected surface EMG signal with Biograph Infiniti software. Through many signal acquisition experiments, it is proved that the system can collect and process surface EMG signal well. In the process of surface EMG signal acquisition, many interference signals are often accompanied, except for the main 50Hz power frequency interference, there are also other noise signals, which need to be de-noised. Biograph Infiniti software comes with MATLAB software interface program. In this paper, the IIR50Hz notch and FIR50Hz notch are designed by using MATLAB simulation software. The 50Hz interference in the signal is removed without affecting the 50Hz frequency signal in the useful signal as much as possible. The experimental results show that the FIR50Hz filter is more effective to remove the 50Hz power frequency interference from the surface EMG signal. In this paper, in order to remove other noises of surface EMG signal, wavelet Multiresolution analysis is first used to de-noise the signal. A digital filter based on wavelet transform is constructed by combining wavelet denoising with the basic theory of digital filter. The experimental results show that the digital filter based on wavelet transform can better remove other noises of surface EMG signal. In addition, after the research of surface EMG signal denoising, the de-noised signal is processed. The wavelet transform is used to extract the feature of the signal and the BP neural network is used to classify the four movements of the upper limb. The experimental results with high recognition rate are obtained. Through the research in this paper, a set of algorithms and schemes can be used to collect and analyze EMG signals. In the future work, we will continue to study various feature processing methods to find out more suitable methods for EMG signals. At the same time, we will also try to improve the performance of the system by the influence of different classifiers on classification performance. It is applied to rehabilitation robot.
【学位授予单位】：东北大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：R318.04;TN911.7

【参考文献】