P300输入系统刺激范式的优化研究

发布时间：2018-10-09 10:26

【摘要】：脑机接口(Brain-Computer Interface,简称BCI),指的是在大脑和计算机及其他外部设备之间进行直接通讯的一种交互技术。在BCI领域中,EEG信号中的P300成分是学者们研究的比较多的一种信号,因其训练要求低的特性使得实现人工智能化的系统很方便。但是基于P300的拼写系统仍然存在很多不足,在拼写速率、拥堵效应、用户接受程度等很多方面都各有差异。为了更好的体现P300拼写性能,本文也做不了不少努力。本文的主要工作和成果如下:1.该论文主要研究的是基于P300的BCI字符输入系统,主要从刺激范式、刺激界面两大方面展开工作。本文在学习了现有的比较经典的几种刺激范式后,提出了一种新的边闪烁边检测的刺激范式,然后在刺激界面上(多种类型混杂、矩阵大小两方面)做了研究,最后在BCI2000平台上实现了该P300字符输入系统。2.与传统的行列式字符输入系统相比,本系统具有以下几个特点:只需要较短时间的有监督训练之后,即可结束使用系统之前单调沉闷的训练过程,进入动态输出模式,使用户的体验感受大有改善;并且随着每个轮次闪烁的进行,目标刺激的定位不断缩小,拼写的精度也在不断提高。3.刺激范式的设计:利用训练数据,对受试者的脑电信号进行分类,得到受试者的P300信号模型,将此信号模型应用于新的刺激范式,可以计算出拼写界面中各字符的分类器响应值,通过动态方法判断目标刺激,从而提高了拼写器的识别速度。从实验数据可以看出,该改进过的P300字符输入系统不仅提高了系统的拼写速率,也提高了拼写的准确率。
[Abstract]:Brain-Computer Interface (Brain-Computer Interface,) is an interactive technology that communicates directly between brain, computer and other external devices. In the field of BCI, the P300 component of BCI signal is a kind of signal studied by many scholars. Because of its low training requirement, it is convenient to realize the system of artificial intelligence. But the spelling system based on P300 still has many shortcomings, such as spelling rate, congestion effect, user acceptance and so on. In order to better reflect the spelling performance of the P300, this article is not a lot of efforts. The main work and results of this paper are as follows: 1. This paper mainly studies the BCI character input system based on P300, which mainly works from two aspects: stimulus paradigm and stimulation interface. In this paper, after studying some classical stimuli, we propose a new stimulus paradigm for edge flicker detection, and then we do some research on the stimulation interface (multiple types of mixing, matrix size). Finally, the P300 character input system. 2. 2 is implemented on BCI2000 platform. Compared with the traditional determinant character input system, the system has the following characteristics: after a short period of supervised training, the monotonous training process before using the system can be ended and the dynamic output mode can be entered. The user's experience is greatly improved, and as each cycle flashes, the location of the target stimulus shrinks and the spelling accuracy improves. Design of stimulus Paradigm: using training data to classify the EEG signals of the subjects, the P300 signal model of the subjects is obtained, and the signal model is applied to the new stimulus paradigm. The classifier response value of each character in the spelling interface can be calculated, and the target stimulus can be judged by dynamic method, thus the recognition speed of the spelling device can be improved. From the experimental data, it can be seen that the improved P300 character input system not only improves the spelling rate of the system, but also improves the spelling accuracy.
【学位授予单位】：浙江工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：R318;TN911.7

【相似文献】