Research On Recognition Of Epileptic EEG Based On Bionic Nervous System

Because epileptic seizures and the whole course of treatment and rehabilitation are concentrated in the human brain,Therefore,the study of epileptic EEG signals plays an important role in the research of the brain.The study of epileptic EEG not only improves the level of clinical diagnosis and treatment of epilepsy,but also helps people understand more about the brain.In this paper,the recognition of epileptic EEG signals is studied using least squares support vector machine(LS-SVM),K? model and deep learning.Firstly,the recognition of epileptic EEG signals in the four frequency bands of 0.5-3HZ,3-8HZ,8-13HZ,and 13-30HZ was studied using traditional EEG recognition methods.First of all,the signal is preprocessed by the mirror extension,and then each frequency band is extracted by FFT.Then the T test,F test and Kruskal-Waills test were used to test the characteristic values of each frequency band.Finally,based on the least squares support vector machine classifier,the effective features extracted by the three test methods are used in the recognition experiment.The experimental results show that,According to the three testing methods,the recognition effect of different frequency band features obtained by Kruskal-Waills test is the best;According to the epileptic EEG in different frequency bands,the 8-13HZ frequency band can be recognized as the best frequency band for the effective detection of epileptic brain.Then,the recognition of epileptic EEG based on K? model is studied.First,the appropriate segmentation preprocessing is carried out on the EEG,and then the K? model is used as a classifier to identify epileptic EEG.Two groups of experiments are identified:the sequence of feature extraction is identified,and the direct sequence is identified.The experimental results show that when the feature extraction method is identified,the recognition rate is the highest and the recognition rate is 91.67%,when the segmentation is divided into 50 segments,that is,the sequence length is 81.Based on direct sequence recognition,for EEG with different sequence lengths,the more channels the sample contains,the higher the recognition rate.Therefore,when the number of channels is 50,the recognition rate is the highest and all are above 96%.This experiment shows that the K? model can identify epileptic EEG well.And the segmentation of EEG signals is related to the rate of recognition,so it is very important to select a reasonable segmentation.Finally,recognition of epileptic EEG based on Deep Learning is studied,In this paper,a convolution neural network model in deep learning is adopted.Firstly,the EEG signal is segmented and preprocessed,and the convolutional neural network is used to directly recognize the epileptic EEG signal sequence.Experimental results show that convolutional neural network is the same as K? model,the recognition rate of convolutional neural network is higher with the increase of the number of channels under different sequence lengths,and the recognition rate is more than 99%when the number of channels is 50.But the recognition result is better than that based on K? model.Especially when the number of channels is 5 and the number of channels is 10,the recognition rate of convolutional neural network is significantly higher than that of K? model.