针对脑电控制假手的运动想象识别及其本体感觉反馈研究

发布时间：2019-04-08 18:03

【摘要】：人手作为人类最灵巧的肢体器官之一，在人类认识世界、改造世界过程中扮演着重要的角色。人们一旦因为交通事故、工伤等意外情况而造成手的缺失，则会给其生活带来极大的不便，生理和心理上也会承受巨大的痛苦。仿生假手正是在这种情况下应运而生，理想的假手设计应满足下面两个要求，一方面要有高效可靠的控制方式，能够准确的获取使用者的操作意图，，并准确控制假手的动作，另一方面要能够向人体反馈假手的位置状态、运动状态及周围环境信息等，让使用者通过假手间接获得感觉。本文结合课题“针对脑电控制假手的运动想象识别及其本体感觉反馈研究”的要求，根据运动想象脑电信号的神经机理，对运动想象脑电信号的采集、预处理、特征提取和模式分类等做了探讨，并将其与仿生假手的控制方法相结合，提出了脑电控制假手的系统组成及控制策略。更深入地，在研究了本体感觉反馈现象之后，提出将振动诱发的运动幻觉作为本体感觉反馈应用到脑电控制的假手上，通过其向使用者反馈假手的运动感觉，实现仿生假手更加自然的运动感觉反馈。本文主要完成了下列工作，并取得了部分创新成果：（1）对脑电控制假手的感觉反馈进行了深入的探讨，介绍了目前已应用于假手的感觉反馈方式。结合本体感觉和肌肉神经感知机制，阐述了经高频振动诱发的运动幻觉产生机理，并进行了振动状态下的本体感觉反馈实验，对运动幻觉的特性做了进一步的分析。提出利用腕部运动幻觉作为假手本体感觉反馈应用到脑电控制假手系统中。在此基础上，对基于脑电信号控制的假手及其本体感觉反馈系统方案做了详细的介绍。（2）针对使用传统EMD方法对脑电信号进行消噪时会导致某些IMF所包含的有用信号与噪声一起被过滤掉情况，提出了一种基于EMD小波阈值消噪的方法，克服了传统EMD去噪无法保留高频成分中有用信息的缺陷。同时，针对脑电信号中眼电伪迹干扰的问题，提出了一种基于二阶非平稳源的盲源分离算法来消除脑电信号中混杂的眼电伪迹，信号的预处理效果明显。（3）提出了采用模糊熵算法对脑电信号进行特征提取，模糊熵算法选择指数函数作为模糊函数来度量两个向量的相似性，避免了近似熵及样本熵使用二值函数方法缺乏连续性、对阈值的取值敏感、容易导致熵值突变的问题。结合运动想象脑电信号的ERD/ERS现象，提出利用C3、C4通道脑电信号的分段模糊熵差值作为特征向量，最后通过实验对分类效果做了评价与对比。（4）在脑电信号模式分类方面，首先介绍了线性分类方法，接着阐述了支持向量机的基本原理，讨论和分析了针对支持向量机中惩罚因子C及核函数中的参数变量最优选取的问题，阐述了基于GA优化的支持向量机算法原理，最后通过实验对算法效果进行了对比分析与讨论。
[Abstract]:As one of the most dexterous human body organs, human hands play an important role in understanding the world and transforming the world. Once people lose their hands because of accident such as traffic accident, work injury and so on, they will bring great inconvenience to their life, and they will suffer a great deal of pain both physically and psychologically. The ideal design of artificial hand should meet the following two requirements. On the one hand, there should be an efficient and reliable control mode, which can accurately obtain the user's operating intention, and accurately control the movement of the artificial hand. On the other hand, we should be able to feedback the position, motion and surrounding information of the prosthetic hand to the human body, so that the user can obtain the feeling indirectly through the artificial hand. According to the requirement of the subject "recognition of Motion Imagination of artificial hand controlled by Electroencephalogram and Research of proprioceptive feedback", according to the neural mechanism of Motor Imagination EEG, the acquisition and pretreatment of Motor Imagination EEG signal are carried out in this paper. Feature extraction and pattern classification are discussed, and combined with the control method of bionic artificial hand, the system composition and control strategy of EEG controlled artificial hand are put forward. More deeply, after studying the phenomenon of proprioceptive feedback, this paper proposes to apply the vibration-induced motor hallucination as proprioceptive feedback to the artificial hand controlled by EEG, through which the motor sensation of the prosthetic hand can be fed back to the user. Realize the bionic prosthetic hand more natural motion sensation feedback. The main contents of this paper are as follows: (1) the sensory feedback of the artificial hand controlled by EEG is deeply discussed and the sensory feedback method which has been applied to the prosthetic hand is introduced in this paper. (1) the research results are as follows: (1) the sensory feedback of the artificial hand controlled by EEG is deeply discussed. In this paper, the mechanism of motor hallucination induced by high frequency vibration is expounded based on the mechanism of proprioceptive sensation and muscular nerve perception, and the experiments of proprioceptive feedback under vibration state are carried out, and the characteristics of motor hallucination are further analyzed. This paper presents the application of wrist motor hallucination as the proprioceptive feedback of prosthetic hand to electroencephalogram (EEG) control of prosthetic hand system. On this basis, the scheme of artificial hand and its proprioceptive sensory feedback system based on EEG control is introduced in detail. (2) in view of the fact that some useful signals contained in IMF are filtered out together with noise when the traditional EMD method is used to Denoise EEG signals, a denoising method based on EMD wavelet threshold is proposed. It overcomes the defect that traditional EMD denoising can not retain useful information in high frequency components. At the same time, a blind source separation algorithm based on second-order non-stationary source is proposed to eliminate the mixed eye artifacts in EEG signals, and the preprocessing effect of the signals is obvious. This paper proposes a blind source separation algorithm based on second-order non-stationary sources to eliminate the interference of eye artifacts in EEG signals. (3) the fuzzy entropy algorithm is proposed to extract the features of EEG signals. The fuzzy entropy algorithm selects the exponential function as the fuzzy function to measure the similarity between the two vectors. The approximate entropy and sample entropy are not continuous by using the binary function method, which is sensitive to the threshold value and can easily lead to the sudden change of entropy value. Combined with the ERD/ERS phenomenon of motor imagination EEG signal, the fuzzy entropy difference of C _ 3 and C _ 4 channel EEG signals is used as the characteristic vector. Finally, the classification effect is evaluated and compared by experiments. (4) in the field of EEG pattern classification, the linear classification method is introduced firstly, and then the basic principle of support vector machine (SVM) is expounded. In this paper, the optimal selection of penalty factor C and parameter variables in kernel function of support vector machine (SVM) is discussed and analyzed. The principle of support vector machine (SVM) algorithm based on GA optimization is expounded. Finally, the experimental results of the algorithm are compared and discussed.
【学位授予单位】：杭州电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：R318.17;TP391.4

【参考文献】