论文部分内容阅读
提取出脑电信号中微弱征兆信息,可以更好地了解脑电信号的特征,但由于各类外界信号的相互混叠,信号呈现出非线性、非平稳性,因此脑电信号的提取是个难题。为此本研究提出了优于小波分解的经验模式分解(EMD)算法对脑电信号进行分解,提取主要IMF分量的特征值,随后采取代价敏感支持向量机(CSVM)进行分类,并对参数进行寻优。在对癫痫患者脑电信号研究的实验中,分类准确率均达到90%以上,验证了本方法的可行性。
Extracting the faint sign information from EEG signals can better understand the characteristics of EEG signals. However, because of the aliasing of various external signals, the signals appear nonlinear and non-stationary, so the extraction of EEG signals is a difficult problem . Therefore, the EMD algorithm superior to wavelet decomposition is proposed in this study to decompose the EEG signals and extract the eigenvalues of the main IMF components. Then, this algorithm uses the cost-sensitive support vector machine (CSVM) to classify the eigenvalues, Optimistic. In the experiments of epilepsy patients with EEG signals, the classification accuracy rate reached more than 90%, which verifies the feasibility of this method.