The analysis of the influencing factors of the wavelet entropy is not simple, which involves the cognitive analysis based on the wavelet entropy of EEG, so some selection and calculation principles of the decomposition level and P_n should be determined. A new wavelet entropy method based on the EEG rhythm characteristics is proposed in this paper, which takes all subspaces of each rhythm as a unit, then calculates the wavelet entropy based on the power of the rhythm. The relationships between the EEG and the wavelet entropy are established by the method. The method is verified in detail by the BCI IV EEG data. Furthermore the relationships between the EEG wavelet entropy and the EEG characteristic are further explored. The method provides an important way for further physiological analysis of the EEG, and the relationships between EEG-rhythm characteristics, EEG wavelet entropy and cognitive features are established by the method.