基于支持向量机的P300脑电信号分类研究

【作者】 吕竟雷；

【导师】 田梦君；

【作者基本信息】 西北工业大学 ， 机械电子工程， 2005， 硕士

【摘要】 脑机接口是一种利用人脑生物电信号实现人脑与计算机或其他电子设备通讯和控制的系统,人脑生物电信号的识别是脑机接口的核心。目前,已有许多模式识别方法用于人脑生物电信号的识别,识别性能直接关系到脑机接口技术能否走出实验室。 本文针对P300脑电信号的特点,通过P300仿真信号和国际标准实验信号,对支持向量机分类识别P300脑电信号进行了研究。研究结果显示,支持向量机对P300仿真信号具有很高的识别率和识别速度:对于真实P300脑电信号,在标准识别方案的基础上,通过多次信号平均和参数优化,并采用整个训练集进行分类器的训练,识别率达到90.3%,满足了识别性能要求。值得一提的是,在分类识别过程中,仅仅采用了15个通道的采样数据和较少的预处理(总通道数:64),识别速度比较快。 研究结果表明,支持向量机对P300脑电信号有很强的分类能力,促进了脑机接口的发展。更多还原

【Abstract】 Brain Computer Interface (BCI) is the system that can realize the control and communication between human brain and computer or other electronic equipment by using bioelectric signal of human brain, the recognition of bioelectric signal of human brain is the core of the BCI. At present, a lot of pattern-recognition methods have been used in the recognition of the bioelectric signal of human brain, the performance of recognition determines directly whether the BCI technology could go out of the laboratory.Based on the characteristic of P300 EEG signal, this paper has studied the classification of the artificial P300 signal and international standard experiment signal using Support Vector Machines (SVM). The result shows that SVM has high recognition rate and speed to artificial P300 signal; To the true P300 EEG signal, on the basis of the standard scheme and the whole training set, through the average of several signals and the parameter optimization, the recognition rate is up to 90. 3% and the performance is good. What deserves to be mentioned is, in the course of classifying, only the sample data of 15 channels and less pretreatment are adopted (the overall channel number is 64), the speed of recognition is faster.The result of study indicates that SVM has the strong ability of classification to P300 EEG signal, and has promoted the development of BCI.更多还原