| Electroencephalography(EEG)is considered as a standard method of recording human brain activity in the form of electrical pulses as a function of time,which can directly reflect the process of human brain activity.It has been widely used in brain science research,diagnosis of brain diseases,and design of brain-computer interfaces.In addition,EEG data has shown great potential as a diagnostic and monitoring tool for a variety of clinical applications.However,the EEG itself is relatively weak,and there are various reasons such as human physiological activities,acquisition equipment and environment.The collected EEG signals will inevitably be polluted by various artifacts,which will make the subsequent analysis of EEG signals difficult.Therefore,the study on the removal of EEG artifacts becomes very important.In recent years,the research on the artifact removal of multi-channel EEG signals has made relatively effective progress,but with the development of portable EEG collection equipment,sometimes only a few or even a single channel EEG signals can be collected,Therefore,the research on artifact removal of single channel EEG signals has practical significance.At present,most research on the removal of artifacts from single channel EEG signals often only focuses on the situation where one artifact is contaminated.The study of artifact removal methods for EEG signals contaminated by multiple artifacts is currently an urgent theoretical and practical issue to be addressed.This article focuses on the research of multiple artifact removal methods for single channel EEG signals,mainly including the following two aspects:Firstly,this thesis focuses on the artifact removal of single channel EEG signal,and analyzes and optimizes parameters of variational modal decomposition(VMD)method.For the two important parameters of modal decomposition number and quadratic penalty factor,the selection strategy of invalid center frequency based on an evaluation index and selection method based on minimum permutation entropy are proposed respectively,the parameter selection time is shortened.Based on the VMD method for parameter optimization,a combined VMD algorithm and Second-order blind identification(SOBI)algorithm were proposed to remove multiple artifacts of single-channel EEG signals respectively and simultaneously.The removal effect was verified by simulation experiments of different cases,and compared with the existing method EEMD-SOBI.In this thesis,simulation experiments are carried out for variational mode decomposition method of parameter optimization,joint variational mode decomposition method and second-order blind identification method respectively,to achieve the effect of removing multiple artifacts of single channel EEG signal,which provides a reference for biomedical signal processing research. |
