基于ELM和近似熵的脑电信号检测方法.pdfVIP

学兔兔 第33卷 第3期 仪 器 仪 表 学 报 V01．33 No．3 2012年3月 Chinese Joumal of Scientific Instrument Mar．2012 基于ELM和近似熵的脑电信号检测方法木 袁 琦，周卫东，李淑芳，蔡冬梅 (山东大学信息科学与工程学院 济南 250100) 摘 要：脑电癫痫波的自动检测与分类对癫痫病情的诊断具有重要意义。提出了一种基于极端学习机(extreme learning ma— chine，ELM)和近似熵的脑电信号检测方法。首先，计算脑电信号的近似熵作为非线性特征，并与利用小波变换技术提取的线 性特征波动指数相结合，组成特征向量，然后将特征向量送人单隐层前馈神经网络，采用ELM学习算法训练网络。实验表明， 与BP(backpropagation)和SVM(suppo~vector machine)算法相比，ELM在训练时间和识别精度两方面性能最佳，对用于实验的 脑电数据检测识别率达到98％以上。 关键词：癫痫脑电；近似熵；极端学习机；反向传播算法；支持向量机 中图分类号：TP391．4 R318．04 文献标识码：A 国家标准学科分类代码：510．40 520．20 Approach of EEG detection based on ELM and approximate entropy Yuan Qi，Zhou Weidong，Li Shufang，Cai Dongmei (School of Information Science and Engineering，Shandong University，Jinan 250100，China) Abstract：The automatic detection and classification of epileptic EEG are significant in diagnosis of epilepsy．This paper presents a new EEG detection approach based on extreme learning machine(ELM)and approximate entropy (ApEn)．Firstly，the value of the ApEn is calculated as the non-linear feature．Then wavelet transform is used to compute the fluctuation indices as linear features．Secondly，ELM algorithm is employed to train single-hidden layer feedforward neural network(SLFN)with the features．Finally，the SLFN is tested．Experiments demonstrate that compared with BP algorithm and support vector machine(SVM)，the performance of ELM is better in terms of train— ing time and classification accuracy，and a satisfied classification rate inore than 98％ for interictal and ictal EEGs is achieved． Key words：epileptic EEG；approximate entropy(ApEn)；extreme learning macfiine