基于EEG的运动想象分类与识别算法及其在脑—机接口中的应用

发布时间：2018-01-12 03:14

本文关键词：基于EEG的运动想象分类与识别算法及其在脑—机接口中的应用　出处：《安徽大学》2012年硕士论文　论文类型：学位论文

【摘要】：基于脑电的脑—机接口(Brain-Computer Interface, BC1)技术作为一种新型人机交互手段,近年来已经成为康复工程以及生物医学工程等领域的研究热点。脑一机接口是在人脑和外界环境之间建立直接通信,而不依赖外周神经以及肌肉这种正常的输出通道。对脑电数据的正确分类是决定脑—机接口性能的关键因素,因此研究基于脑电的分类识别算法具有重要的现实意义。本文以基于运动想象的脑—机接口作为研究对象,对基于运动想象的脑电信号的特征提取方法和分类识别算法进行了系统的研究,并实现了以基于滑动窗的ICA(Sliding Window ICA, SWICA)为核心算法的在线BCI系统。论文的主要内容如下： 1.设计了基于运动想象的BCI实验范式,采集了较丰富的左右手运动想象脑电数据；结合已有的标准EEG数据,建立了用于本文离线BCl分析的实测脑电信号数据库。 2.研究了基于后验概率的支持向量机(Posteriori Probability Support Vector Machine, PPSVM)和隐马尔科夫模型(Hidden Markov Models, HMM)等模式识别方法,并结合脑电模式分类问题,进行了不同分类方法的性能比较。并用基于后验概率的支持向量机检测出了运动想象过程中的“休息”状态,为实现运动想象在线控制系统创造了良好的条件。 3.研究并实现了基于能量、共同空间模式(Common Spatical Pattern, CSP)和SWICA算法的EEG特征提取方法。重点研究了基于SWICA的信号包络检测新方法,并将该算法应用于脑电mu节律的动态特性分析和运动想象分类,得到了较高的运动想象分类识别率。 4.设计并实现了基于SWICA算法的在线BCI系统,实验证明,该系统可以实时在线的识别出左右手运动想象,识别率最高可达92.7%。
[Abstract]:EEG based brain computer interface (Brain-Computer Interface BC1) technology as a new means of human-computer interaction, in recent years has become a hotspot in the research fields of rehabilitation engineering and biomedical engineering. A brain computer interface is a direct communication between human and environment, and not rely on peripheral nerves and muscles of the normal output channel. The correct classification of EEG data is a key factor determining the BCI performance, so the research on classification and recognition algorithm based on EEG has important practical significance.
Based on motor imagery based on brain computer interface as the research object, the feature extraction method of EEG of motor imagery classification and recognition algorithm based on systematic research, and in order to realize the sliding window based on ICA (Sliding Window ICA, SWICA) online BCI system as the core of the algorithm. The main contents of this paper the following:
1., we designed a BCI experimental paradigm based on motor imagery, and collected abundant left and right hand motor imagery EEG data. Combined with the existing standard EEG data, we set up a database of measured EEG for off-line BCl analysis.
2. support vector machine is studied based on the posterior probability (Posteriori Probability Support Vector Machine, PPSVM) and hidden Markov model (Hidden Markov Models, HMM) and other methods of pattern recognition, and combined with EEG pattern classification problems, compare the performance of different classification methods. The support vector machine and detection based on posterior probability the movement of imagination in the process of "rest" state, to create good conditions for online motor imagery control system.
3. research and Implementation Based on energy, common spatial pattern (Common Spatical, Pattern, CSP) EEG feature extraction and SWICA algorithm. The method focuses on the new method of signal detection based on SWICA, and the classification of imagination dynamic characteristics analysis and motion of the algorithm is applied to mu rhythm of EEG, get high motor imagery classification and recognition rate.
4., we design and implement an online BCI system based on SWICA algorithm. Experiments show that the system can identify the left and right hand motor imagery online and online, and the recognition rate is up to 92.7%..

【学位授予单位】：安徽大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TP391.41;TP334.7

【引证文献】