A novel data compression method is developed for periodical data in power systems. Considering the unbalanced nature of information in cycles and between cycles， coupling of information caused by power system frequency fluctuation and non-integer-period sampling are eliminated based on the cubic spline interpolation re-sampling method to achieve large compression ratio. The error caused by re-sampling is analyzed. The data are compressed based on the lifting wavelet decomposition method. The real-life periodical data are used to test the proposed method's performance. The results indicate that much higher signal-to-noise ratio can be achieved than a compressing data method without the re-sampling process for the same compression ratio.