Abstract:The Mars pole motion belongs to the Martian orientation parameter. By studying the period term of the Mars pole motion time series, people can have a deeper understanding of the characteristics and laws of the Mars liquid core and mantle structure, as well as the Martian atmosphere and tidal movement. In this paper, we use NP.ang and NP.ds Mars pole motion sequences provided by NASA(National Aeronautics and Space Administration) for time-varying analysis and prediction. The time-variant analysis part uses fast fourier transform(FFT) for processing data, and results show that, the periods of NP.ang and NP.ds are basically the same, and strongly correlate with Martian years. In addition, the autoregressive integrated moving average(ARIMA) model is used to predict NP.ds within 20 days. The experiment results show that it is feasible to use the least square extrapolation model to forecast the NP.ds data in the medium and long term, and the fitting of the least square extrapolation model with more accurate period term can significantly improve the prediction accuracy of NP.ds. These results can provide reference for the prediction of Mars pole motion.