基于CSP-PSO-SVM的运动想象EEG信号特征提取与分类算法

来源期刊:中南大学学报(自然科学版)2020年第6期

论文作者:刘宝 唐雨琦 蔡梦迪 薄迎春 张月

文章页码:1551 - 1565

关键词:运动想象;共空间模式;支持向量机;粒子群优化

Key words:motion imagining; common spatial pattern; support vector machine; particle swarm optimization

摘    要:为了解决EEG信号特征提取困难及识别率低等问题,提出一种基于CSP-PSO-SVM的脑电(EEG)信号特征提取与分类算法。该算法首先通过小波包变换实现EEG信号的预处理,提取出EEG信号中的特定频段信号,然后通过构建“一对一”共空间滤波器对EEG信号进行特征提取,最后通过粒子群优化的支持向量机算法实现EEG信号分类识别,并选用2008BCI竞赛2A数据集进行算法分类效果校验。研究结果表明:改进型CSP-PSO-SVM算法的分类准确率最高可达到93.07%,且其平均准确率比其他算法的高。其特征能很好地反映EEG信号的特点,可明显提高分类识别的准确率,可为脑机接口的发展与应用提供参考。

Abstract: In order to solve the problems of EEG signal feature extraction difficulty and low recognition rate, an EEG signal feature extraction and classification algorithm based on CSP-PSO-SVM was proposed. Firstly, the preprocessing and specific frequency band signal extraction of EEG signal was realized through wavelet packet transformation. Then, the feature of EEG signal was extracted by constructing a "one-to-one" common space filter. Finally, EEG signal classification and recognition were realized by the support vector machine algorithm optimized by particle swarm. The 2008BCI competition 2A data set was selected for algorithm classification effect verification. The results show that the improved CSP-PSO-SVM algorithm can achieve a classification accuracy of up to 93.07%, and the average accuracy is higher than that from other algorithms. The characteristics of EEG signals are well reflected by the proposed algorithm, the accuracy of classification and recognition is significantly improved, and it can provide a reference for the development and application of brain-computer interfaces.

有色金属在线官网  |   会议  |   在线投稿  |   购买纸书  |   科技图书馆

中南大学出版社 技术支持 版权声明   电话:0731-88830515 88830516   传真:0731-88710482   Email:administrator@cnnmol.com

互联网出版许可证:(署)网出证(京)字第342号   京ICP备17050991号-6      京公网安备11010802042557号