Research On Analytical Method Of Hand Motion Eeg Signal In Motion Imagination

In recent years,the trend of population aging has been accelerating,and the number of elderly patients suffering from cerebrovascular diseases has gradually increased.The number of patients with hemiplegia caused by accidents has also increased year by year.How to treat these patients more effectively is the focus of solving these problems.The initial idea ofbrain-computer interface technology research is to provide convenience for patients who cannot communicate or squat.They can use BCI technology to use the brain to communicate directly with the outside world for rehabilitation or self-care.Therefore,the research of BCI technology is very urgent and important.In order to quantitatively describe the brain-computer interface system,this paper studies the analytical methods of hand-moving EEG signals in motor imaging,and mainly studies the EEG signals of different hand movements.Firstly,the experimental system scheme is developed to study the signal acquisition method.Secondly,the brain signal of the hand is opened and the motor image is taken as the acquisition signal type,and the original signal is subjected to wavelet improved threshold method to reduce noise.Then,multiple features are used.The fusion method is used to extract features,construct the feature model of the hand motion imaging EEG signal to identify the motion state,and use the t-SNE algorithm to reduce the dimension of the extracted multidimensional feature vector.Finally,the support vector machine is used for pattern recognition.The research includes the following aspects:(1)Based on the BCI system of motion imaging,the experimental system scheme was developed,and the EEG signal acquisition experiment was designed by itself to collect the EEG signals of the cerebral cortex.In order to facilitate subsequent analysis and processing,the wavelet-improved threshold method is used to denoise the collected EEG signals.The noise reduction effect is verified by increasing the signal-to-noise ratio and the mean square error value.(2)In order to fully extract the information in the hand motion imaging data,the EEG signal feature extraction method based on multi-feature fusion is used to extract wavelets of different frequency bands based on wavelet packet energy,autoregressive model and co-space model analysis algorithm.Packet energy,time series correlation coefficient and co-space feature vector,the feature vectors extracted by different methods are connected end to end,and the multi-dimensional feature vectors of different motion imaging EEG signals are constructed to ensure the integrity of data information.The eigenvectors with too large data and too many dimensions are subjected to dimensionality reduction processing,and the data ofhigh-dimensional space is projected to a low-dimensional space without changing the high-dimensional data structure.(3)The state of the subject's hand motion imaging EEG signal is classified by the support vector machine(SVM)pattern recognition method.The recognition accuracy of three single feature extraction methods and multi-feature fusion extraction methods are compared and analyzed,and the effectiveness of feature extraction methods in this paper is proved.