Aiming at the problem of brush wear of DC brush motor，a method of predicting and estimating brush state using particle filter algorithm is proposed. The dynamic model of DC motor is built，the brush wear process is simulated by changing the resistance value of armature winding，and the average current data of the brush wear process is simulated.According to the fitting results of the average current data，a basic model of brush state evolution is established. The unknown parameter b of the model is iteratively estimated by the particle filter algorithm of random resampling，and the value is stable near the truth value 0.002. The predicted model reflects the wear process of the brush more accurately，and can estimate the remaining service life of the brush，which is of great significance for the maintenance of the motor brush.