基于伪随机编码调制的视觉诱发电位脑机接口研究

发布时间：2018-04-28 10:54

本文选题：脑机接口 + 伪随机编码调制　；参考：《南昌大学》2017年硕士论文

【摘要】：脑机接口(Brain-computer interface,BCI)是通过大脑直接控制外部设备实现与外部环境的交流,打破了传统的大脑通过肌肉和外周神经与外界的交流。如今,随着脑机接口技术的迅速发展,科研工作者对于频率调制的视觉诱发电位脑机接口(frequency modulation visual evoked potential BCI,f-VEP BCI)进行了大量的研究,而对于伪随机编码调制的视觉诱发电位脑机接口(Pseudo-random code modulated VEP BCI,c-VEP BCI)的研究比较少,传统的c-VEP BCI是使用一种或者说一个编码及其时间移位来调制不同的刺激论标,限制了论标数的增加,从而限制了BCI系统的信息传输率。本文从信学处理的方法以及提高c-VEP BCI的刺激论标数出发,对c-VEP BCI进行了研究。基于一个伪随机序列调制的VEP脑机接口是通过典型相关分析(Canonical correlation analysis,CCA)来优化空域滤波器,使用模板匹配法(Template Matching Method,TMM)进行论标识别。而本文提出了通过信学分数分析方法(Signal Fraction Analysis,SFA)来优化空域滤波器,使用一类支持向量机(One Class Support Vector Machine,OCSVM)来分类识别论标,这两种优化空域滤波方法和两种分类识别论标方法可以互相组合,形成四种不同形式的方法,实验结果表明这四种方法都获得了较高的论标识别准确率。本文又提出了基于多个伪随机序列调制的VEP脑机接口,它是通过两个不同的Golay码和一个近完美序列来对论标进行分组调制,实现了具有48个论标的刺激器,大幅度提高了刺激论标数。通过典型相关分析进行优化空域滤波器,采用模板匹配法对论标进行分类识别,获得了很高的分类识别率。本系统采用了8名实验者的数据进行分析,识别准确率的结果达到了94.95%。
[Abstract]:Brain-Computer Interface (BCI) is a direct control of the brain to control the external equipment to achieve communication with the external environment, breaking the traditional brain through muscle and peripheral nerves to communicate with the outside world. Nowadays, with the rapid development of brain-computer interface technology, researchers have done a lot of research on frequency-modulated modulation visual evoked potential BCIf-VEP BCIs. However, there is little research on pseudo-random code modulated VEP BCIc-VEP BCIP BCIP BCII, which is a pseudorandom coded modulation interface. The traditional c-VEP BCI uses one code or one encoding and its time shift to modulate different stimuli. It limits the increase of scalar number, and thus limits the information transmission rate of BCI system. In this paper, c-VEP BCI is studied based on the methods of information processing and the enhancement of the stimulus scalar of c-VEP BCI. Based on a pseudorandom sequence modulated VEP brain-computer interface, the canonical correlation analysis is used to optimize the spatial filter, and the template Matching method is used to identify the spatial filter. In this paper, the signal Fraction Analysis (SFAs) method is proposed to optimize the spatial filter, and a class of support vector machines (SVM) is used to classify and identify the criteria. These two optimal spatial filtering methods and two classification recognition theory methods can be combined with each other to form four different forms of methods. The experimental results show that these four methods have higher accuracy of theoretical identification. In this paper, a VEP brain-computer interface based on multiple pseudorandom sequence modulation is proposed, which modulates the standard by two different Golay codes and a near-perfect sequence, and realizes 48 stimulators with the target of the theory. It greatly increases the number of stimuli. Based on the canonical correlation analysis, the spatial filter is optimized, and the template matching method is used to classify and identify the criteria, and a high classification recognition rate is obtained. The system uses the data of 8 experimenters to analyze, and the result of recognition accuracy reaches 94.95%.
【学位授予单位】：南昌大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：R318;TN911.7

【参考文献】