摘要
针对脑机接口(BC I)研究中存在脑电信号(EEG)识别率低的问题,提出一种基于遗传算法(GA)和概率神经网络(PNN)的GA-PNN识别方法.用该方法对EEG提取时频特征,构成模式识别的初始特征.以训练样本识别正确率为适应度函数,采用GA对初始特征进行组合优化.基于优选后的特征,用PNN对测试样本进行分类.该方法使EEG识别正确率达到92.49%,与2003年BC I国际竞赛最好的处理结果(88.7%)相比,提高近4%,为BC I中EEG的识别提供了有效的手段.
Aimed at the problem that the recognition rate of electroencephalogram（EEG） is low in braincomputer interfaces（BCIs）, a GA-PNN recognition method based on genetic algorithm（GA） and probabilistic neural network（PNN） was presented. EEG features from time domain and frequency domain are extracted. These features form the initial features for pattern recognition. Then,a GA is used to combine and optimize initial features. The fitness function of the GA is the recognition rate of training samples. Finally, a PNN is used to classify testing samples based on these optimized features. The recognition rate can obtain 92.49%, which improves about 4% compared with the best result （88.7 % ） of 2003 BCI competition. The method provides an effective way to EEG recognition in BCIs.
出处
《上海交通大学学报》
EI
CAS
CSCD
北大核心
2005年第10期1689-1692,共4页
Journal of Shanghai Jiaotong University
关键词
脑机接口
脑电信号
遗传算法
概率神经网络
组合优化
brain-computer interface （BCI）
electroencephalogram （EEG）
genetic algorithm （GA）
probabilistic neural networks（PNN）
combinatorial optimization
