Abstract: With respect to the uncertainty process in goaf residual subsidence, the Ensemble Kalman Filter (EnKF) was introduced, the coal mine deformation was treated as a dynamic stochastic system and a new prediction model named Ensemble Kalman Filter model was proposed.Then the ensemble Kalman filter predicted value was compared with the original measured data.The numerical example shows that the ensemble Kalman filter model can effectively deal with the measured data polluted by noise.It proves that the prediction effect of Ensemble Kalman Filter is good, and Ensemble Kalman Filter offers a new way to predict the goaf residual subsidence.

Key words: Old goaf, Residual subsidence, Ensemble Kalman Filter (EnKF), Prediction