Abstract: Aiming at the problem of considerable radar data random error, the autoregressive integrated moving average (ARIMA) model is analyzed by time-series analysis method. The ARIMA model and Kalman filter are designed based on the foregoing 4 000 data points to smooth the radar data, which is obtained by measuring the International Space Station. The foregoing 4 000 data points and other points are processed by the filter respectively, and random error reduces to 13.7% and 20.1% of the original data separately. The results show that the random error of radar measurements can be reduced effectively by the new method, improving the data quality.