Abstract: Taking one working face of Xishan coalfield in Shanxi province as the experiment area, this paper processes and evaluates the GNSS (global navigation satellite system) monitoring data about mining ground subsidence using two Kalman filter methods (usual Kalman filter and total Kalman filter), on the basis of analyzing the data characteristics of GNSS monitoring data, and compares the distribution status of root mean square error (RMSE) of the two Kalman filter methods. The research results show that the tendency is similar between the filter results of the two methods and the observation values, although a few abnormal points exist. The difference between the filter results and the observation values becomes higher when the subsidence larger, and the difference for the total Kalman filter is much lower than the usual Kalman filter, especially for the monitoring points in the large subsidence area. The RMSE values increase with the rise of the subsidence for the monitoring points, and the increasing velocity for the total Kalman filter is much lower than the usual Kalman filter. The maximum RMSE value for the total Kalman filter is lower than 0.1 m, but it is approximate to 0.4 m for the usual Kalman filter.

Key words: usual Kalman filter, total Kalman filter, mining ground subsidence, GNSS monitoring, RMSE, Xishan coalfield