Study On The Method For Feature Extraction Of EEG Signals Based On Wavelet Packet Decomposition And Approximate Entropy And Its Application To Brain-computer Interfaces

Brain-computer interface technology is a direct communication and control channel between the brain and external hardware and software equipment,which allows patients to communicate with the outside world using brain signals or control external devices without the help of language and physical activity,For patients to open up a channel with the outside world,is a new way of human-computer interaction.Brain-brain interface technology not only to those who are damaged but the brain consciousness of normal paralyzed patients to see the hope of interaction with the outside world,but also the basis for future development of artificial intelligence.Feature extraction is one of the most important aspects in brain computer interface system,which affects the classification performance of the whole system.Traditional feature extraction methods have analyzed EEG signals from time domain,frequency domain,time-frequency combination,and airspace.(EEG)is a very complex and very weak random signal,but also easily by the EMG signal,eye electrical signal(EOG)and other external factors of interference,will be the correct rate of classification has been influences.In order to improve the accuracy of classification of brain-computer interface,this paper uses the wavelet packet decomposition and approximate entropy to extract the EEG signals.This method uses the wavelet packet to decompose the whole frequency band of the EEG signal.The classification entropy is extracted by the approximate entropy function.Then the feature vector is dimensioned by sparse representation.Finally,the power difference method is used to classify the wavelet packet.And select different channels in order to find effective channels,so that these channels can accurately reflect a variety of motion imagination tasks or limb movement characteristics,and reduce noise and irrelevant channel interference to reduce the complexity of the algorithm and reduce the burden of communication systems,Improve the operation efficiency of brain and brain interface system.The experimental results show that the method achieves a good classification effect when using two different sets of channels under the condition of classification using 1 second data.This method has improved the accuracy of using different channel classification compared with wavelet packet decomposition and spatial filtering method and traditional common airspace model method.In addition,the shorter the data length,the higher the classification recognition rate,indicating that the method is more suitable for shorter data,is conducive to improving the brain and computer interface information transmission speed.