Instantaneous impact signal analysis investigation and feature extraction have been broadly investigated by many researchers. How to determine an effective time-frequency distribution analysis method is urgent needed. In this paper, Hilbert spectrum frequency resolution is investigated to show signal typical features according to performance estimation model. Different frequency units for a Hilbert spectrum expression will be investigated on the effect of feature analysis in detail. An estimation model will be constructed to determine the best frequency resolution for signal analysis. By using this estimation, frequency resolution can be determined. This model can improve the performance of Hilbert spectrum to show signal features and accuracy for signal analysis and pattern recognition. To testify the effectiveness of this estimation model, numerical simulation will be used as an example to analyze the accuracy of this model. At the same time, different bearing working condition’s pattern recognition based on Hilbert spectrum will be used to testify the effectiveness of this model.