Thanks to high temporal resolution and non-invasiveness，electroencephalogram（EEG） signals are widely used for fatigue and brain load analysis of aerospace mission operators. Since the EEG signals hold multi-channels and the information within each channel is not exactly the same， this paper proposes an EEG characteristic evaluation method based on the minimum redundancy maximum relevance（mRMR） algorithm. The performance of EEG characteristics can be evaluated by setting the target variables， calculating the amount of mutual information between the EEG characteristics and the target variable within each channel， as well as the redundancy of the characteristics within the channel. Further， the EEG data of controllers under different brain loads are acquired. A series of EEG characteristics are evaluated. The results are compared with those of the existing studies， and the separability of the characteristics under different classification methods are also analyzed. The effectiveness of the proposed evaluation method is verified. Compared with the existing studies， this method avoids the subjectivity of the grey correlation analysis method in determining weight parameters and grey correlation degree，as well as the discrepancy between different classifier evaluation methods. Compared to the existing characteristic selection algorithms， this method considers the redundancy of information within the channel and outputs more accurate results. Compared with the methods based on statistics， this method can quantitatively evaluate the performance of the characteristics in order to compare different indicators. Finally， the application of this method in terms of fatigue level analysis and emotion recognition is described.