Based on wear measurement of brush seals in existing tests， grey prediction model of wear amount of brush seal is established using the grey prediction theory. The service life of the brush seal is estimated， and the test verification is carried out. The results between prediction and test show that the grey model prediction is feasible with good prediction accuracy. Comparison of the wear amount of brush seals under three conditions for 500 min shows maximum relative residual error of 1.57% and maximum absolute residual error of 0.16 mm³.The result can be used as reference for determining operation time and repair and replacement demand of future brush seal devices.