Research On Egg Signal Analysis And Pattern Recognition Based On Deep Learing

EEG signals are often used in brain-computer interface research,human emotion recognition,fatigue driving,epilepsy,and sleep quality monitoring.However,the key to solving these problems is to quickly and effectively extract and classify EEG signals,so as to improve the recognition accuracy of EEG signals on these issues.In order to improve the accuracy of EEG signal recognition,the work and innovations in EEG signal r-ecognition are as follows:Firstly,in the EEG signal feature extraction algorithm,based on the histogram and amplitude co-occurrence matrix of EEG signals,the texture feature extraction algorithm of EEG signals is proposed.The algorithm is managed on the data set given by Kaggle and challenged in BCI..The experimental results show that the EEG signal texture features designed in this paper are added to the first scheme of the competition,which improves the recognition accuracy of the task from 87.56%to 89.54%,which proves the EEG signal texture feature extraction algorithm proposed in this paper.It is effective for EEG signal recognition tasks.Secondly,the channel attention module is proposed and designed,and connected to the input layer of the Convolutional Neural Network(CNN)to automatically learn the contribution of EEG signals of different channels to the recognition task.At the same time,based on this,this paper combines the residual unit and the channel attention model,and proposes a convolutional neural network structure that is very effective for EEG signal recognition.Finally,the experiment is carried out in the EEG dataset of the Stanford Research Project.The experimental results show that the proposed model improves the recognition accuracy of the task fr-om 82.58%to 85.68%of ResNetl3.Thirdly,the convolutional neural network architecture is proposed and designed for the problem that the convolutional neural network can not fully utilize the EEG signal time series information.The network sends the characteristics of EEG signals extracted by the convolutional neural network into the circulating neural network,making full use of the time series information of the EEG signals.At the same time,based on the eigenvectors outputted by the cyclic neural network at each moment,this paper designs a attention concentration model,which allows the model to automatically learn the contribution of the eigenvectors outputted by the cyclic neural network at each moment to the classification task.Finally,in order to verify whether the method is effective,this paper also carried out experiments on the EEG dataset of the Stanford research project.The experimental results show that the algorithm improves the recognition accuracy to 91.05%based on the second innovation.