Research Of Non-linear System Theory On EEG Analysis

EEG is a kind of complicated non-periodic bioelectricity signal and its research is still an important field for modern science. Recent years, using non-linear dynamics analysis to develop EEG has become a mainstream. However, its research exists many problems as EEG is very complicated and particular.On the base of investigation on the EEG research at home and abroad, this paper develops non-linear dynamics analysis method, the main work and conclusion are as follows:1) Demonstrates the possibility of space reconstruct and the feasibility of calculating correlation dimension, solving the problems of selected embedding dimension and delay time as well as data length. The paper has greatly improved the G-P arithmetic, so it can run much faster. And, the resampling technique is used to reduce the quantity of data but maintain the information. All this is a base for the application of clinic.2) The theory of bursts of 40Hz EEG is used to get rid of the possible disturbs of EMG. This method makes the non-linear EEG analysis more accurate and reasonable.3) The correlation dimension, Lyapunov exponent and Kolmogorov entropy are studied and their computer programs are implemented.4) The paper analyses the 30 objects under different conditions, validating the correctness of methods provided in this paper. The results indicate that the correlation dimension is larger when the brain is under the conditions of calculating, remembering and associating with eyes closed than passive eyes closed; the correlation dimension is larger in left brain than right under the condition of reasoning with eyes open.