<p>A novel weighted multiscale combined difference morphological filter (WMCDMF) is developed to extract impulsive fault components and achieve incipient bearing fault diagnosis. First, to overcome the limitation of the single-scale morphological filter, the multiscale combination difference morphological filter is presented to extract fault information from different scales. Then, the particle swarm optimization is exploited to obtain the optimal weighted coefficients, and the fault feature ratio is adopted as the evaluation criterion. Finally, the fault features of the filtered signal are extracted to realize incipient bearing fault diagnosis. Bearing data are implemented to evaluate the effectiveness of the designed WMCDMF scheme. Experimental results indicate that the WMCDMF scheme can detect incipient bearing fault components. Furthermore, the advantage of the WMCDMF scheme has been verified in comparison with the representative morphological filter schemes.</p>