基于CSP的联合特征提取算法研究与优化(硕士)

以下是资料介绍,如需要完整的请充值下载. 本资料已审核过,确保内容和网页里介绍一致.  
无需注册登录,支付后按照提示操作即可获取该资料.
资料介绍:

基于CSP的联合特征提取算法研究与优化(硕士)(论文43000字)
摘要
脑机接口通过大脑产生的脑电信号(Electroencephalogram, EEG)实现人脑与计算机或其他设备之间通信和控制[1],为人类和环境之间提供了新的通信和控制渠道,拓宽了人类的控制能力,具有广泛的应用前景。然而,用于脑机接口(Brian-computer Interface, BCI)设备研究的脑电信号是一种弱的、非线性、非平稳并且随时间变化的信号。所以,提出一种有效的特征提取方法是改善识别精度的关键。
论文首先对运动想象脑电信号的现代分析方法进行综述,着重介绍多种脑电信号特征提取方法,其中,对经典算法的改进方法以及多种经典算法的组合方法进行了详细地阐述。
其次,在综述分析的基础上,论文提出了一种公共空间模式算法(Common Spatial Pattern,CSP)结合经验模式分解(Empirical Mode Decomposition, EMD)的特征提取方法。针对公共空间模式分解算法需要大量通道信号和缺乏频域信息的缺点进行改进,本章提出EMD-CSP算法。先将信号进行经验模式分解得到多个固有模态函数(Intrinsic Mode Functions,IMFs),选取符合μ节律和β节律的振荡模式组合成多通道信息簇,基于由公共空间模式滤波后的信号能量选取成分构造新的空间滤波器提取特征,得到特征向量集,经支持向量机分类后得到所有9位受试平均分类正确率为:92%,其中,最高的受试达到93.8%。改进的CSP滤波成分选择方法在保证分类正确率的基础上大大减少了实际使用的通道数量,为便携式脑机接口设计提供了简单可行的方法。
紧接着,论文设计思维任务实验,探究不同实验范式对特征分类结果的影响,选取较优的实验范式;
最后,进一步优化分类结果,一方面,为提高分类正确率,运用S变换进一步优化CSP滤波器,将公共空间模式算法分别与集总经验模式分解(Ensemble Empirical Mode Decomposition, EEMD)、双谱分析算法结合,构造联合特征。另一方面,为缩短分类过程耗时,从分类器参数优化、采用不同分类器、特征降维三个角度进行研究。论文结合实验数据和竞赛数据分析特征,从分类正确率和时间响应度出发,验证基于CSP的联合特征优化方案的可行性与有效性。结果表明:采用双谱-CSP的联合特征结合线性判别分类器(Linear Discrimination Analysis, LDA)的优化方案获得更优的分类正确率和更快的特征提取速度。

关键词:  运动想象脑电信号,联合特征优化,特征提取,公共空间模式分解
 
Abstract
Brian-computer Interface is not only a controlling pathway between human brain and a computer or other devices, but also an novel communication channel with various application based on electroencephalogram, linking human with outside environment, enlarging humans’ ability. However, the BCI signal source is an nonlinear and instable signal varied with the time changing. As a result, an effective feature extraction method is the key to improve the pattern classification accuracy. 
This thesis firstly summarizes the modern EEG anaylsis methods in motor imagery, and then emphasizes different feature extraction algroithms. Especially, many improved theories and combinatorial algorithms based on principle classic methods have been described in detail.
According the review, this thesis nextly prove an novel feature extraction method combined the Common Spatial Pattern method with the Empirical ModeDecomposition. Normal Common Spatial Pattern method is restricted to the abundant input channels and lacking frequency information. Firstly, The EMDmethod was proposed to decompose the EEG signal into a set of stationary time series called Intrinsic ModeFunctions (IMF). Secondly, these IMFs were analyzed with the band-power to detect the valuable IMFs withcharacteristics of sensorimotor rhythms (5-28Hz), and then the improved CSP filter based on the energy variance was attached to the featureextraction of screening IMFs. Finally, Once the feature vector was built, the classification of MI was performedusing Support Vector Machine(SVM). The results obtained show that the EMD-CSP allow the most reliablefeatures and that the accurate classification rate obtained is 92% which confirm the feasibility and availability ofthis method in portable further BCI systems.
The EEG experiment based on Motor Imagery is designed with two different task paradigms to select the better one. In order to achieve optimized feature sets, on the one hand, the advanced CSP filter is further improved with S transition, and this method is respectively combined with EEMD、Bispectrum anayalsis, achieving union features and increasing the final classification rates. On the other hand, parameter optimization in classifiers、choices between various classifiers and feature dimension reduction are used to shorten the time cost during classification process. The research results not only verify the feasibility and efficiency of these combined-feature sets based on CSP and other algorithms, but also show that when using the LDA classifier, the Bispectrum-CSP feature is the best optimization among the other ones with the highest classification accuracy and the shortest pattern recognition time.

Key words:Motor imagery EEG signal, Joint fweature optimization, Feature extraction, Common Spatial Pattern method
目录
摘要    I
用术语注释表    V
第一章绪论    1
1.1    脑电信号概述    1
1.1.1脑电信号概念和原理    1
1.1.2脑电信号的分类与特性    2
1.1.3脑电信号的研究现状    5
1.2脑机接口概述    6
1.2.1脑机接口的概念和原理    6
1.2.2脑机接口的分类与特性    7
1.2.3 脑机接口研究现状与意义    10
1.3 本文研究内容    11
第二章运动想象脑电信号现代分析方    13
2.1运动想象脑电信号特征提取算法    13
2.2 经典方法的改进算法    13
2.2.1离散小波变换的改进    13
2.2.2 经验模式分解的改进    14
2.2.3 公共空间模式分解的改进    17
2.3 多种方法的组合算法    20
2.3.1 经典方法之间的组合    20
2.3.2经典方法和统计学量的组合    21
2.3.3各类熵    22
2.3 本章总结    26
第三章 公共空间模式算法结合EMD的脑电信号特征提取    27
3.1 数据集描述    27
3.2 基于改进CSP算法的特征提取方法    28
3.2.1 经验模式分解    28
3.2.2 公共空间模式分解算法    29
3.2.3    改进的CSP滤波算法    30
3.2.4 频域能量分析    31
3.3 数据集处理过程与结果分析    31
3.3.1 数据预处理    31
3.3.2 EMD分解与频段筛选    32
3.3.3 改进的CSP滤波    34
3.3.4 结果与比较    35
3.4 本章小结    38
第四章左右手运动想象的实验设计与数据处理    39
4.1 脑电信号采集系统    39
4.1.2电极帽    39
4.1.3 SynAmps放大器    40
4.1.4 Curry7软件    41
4.1.5 EEGLAB平台    41
4.2 运动想象实验设计    42
4.3 实验数据处理    44
4.3.1 数据预处理    44
4.3.2 实验数据分析与比较    44
4.4 本章小结    47
第五章 基于改进CSP算法的特征优化    48
5.1 基于S变换的空间滤波器成分选择算法优化    48
5.1.1 S变换    48
5.1.2 基于S变换的公共空间滤波器成分选择算法    49
5.2 CSP结合EEMD的特征提取方法优化    55
5.2.1优化方案概述    55
5.2.2 优化结果与比较    56
5.3 双谱-CSP的特征提取方法优化    58
5.3.1 双谱特征概述    58
5.3.1 数据分析    60
5.4 特征降维    64
5.5 特征识别过程优化    66
5.5.1 SVM的内核参数优化    67
5.5.2 LDA线性判别分析    69
5.6 本章小结    71
第六章  总结与展望    73
6.1 全文总结    73
6.2 工作展望    74
参考文献    75