The invention discloses a filter method and system of Beidou high-frequency deformation monitoring data. The method includes the steps of S100, using sample data to initialize the state equation and measuring equation of a measurement expansion Kalman filtering method; S200, using double gross error discrimination statistics to perform gross error type judgement on to-be-filter observation values,and performing gross error correction according to the gross error types; S300, performing noise filtering on the to-be-filter observation values after the gross error correction to obtain two-dimensional state vector filtered values; S400, calculating the filtered values of the observation values according to the two-dimensional state vector filtered values, and feeding back the filtered valuesof the observation values to S200 to serve as the filtered values of the observation values of the previous epoch; S500, repeating S200-S400 until the to-be-filtered observation values of all epochs are filtered. The method has the advantages that gross errors are effectively identified through the double gross error discrimination statistics, individual correction is performed according to the gross error types, effective information in the monitoring data is kept to an extremely large extent, and accordingly filter result reliability is guaranteed.