一种EEG信号盲分离和分类的神经网络方法

A NEURAL NETWORK METHOD OF BLIND SEPARATION AND CLASSIFICATION OF EEG SIGNALS

提出一种采用多神经网络处理脑电 (EEG)信号的方法。首先 ,对混有噪声的脑电信号给出一种盲分离的自适应神经算法。通过寻求采样时间序列线性组合的kurtosis系数的局部极值 ,得出该算法的模型和步骤。在盲分离的基础上 ,对分离出的估计信号进一步利用Kohonen网络进行分类。将该算法用于 30 0个EEG样本处理 ,并给出处理结果。

A multiple neural network method of processing EEG signals is proposed. To begin with, a self-adaptive neural algorithm for blind separation of noisy EEG signals was given. By seeking the local extrema of the kurtosis coefficients of a linear combination of the sampled time series, the model and the process of this algorithm were obtained. Based on the blind separation, a further classification of the estimated signals was carried out by using the Kohonen net. Use this algorithm for the processing of 300 EEG samples, the results of the processing was given.