Complete ensemble empirical mode decomposition with adaptive noise（CEEMDAN）can solve the incomplete decomposition of ensemble empirical mode decomposition（EEMD）in integration averaging process，but the residual noise and false components of CEEMDAN remain not to be solved. To overcome shortcomings of CEEMDAN，the weighted mean-optimized empirical mode decomposition with adaptive noise（WMEMDAN）is proposed. In this method，the intrinsic mode function（IMF）is extracted by the improved mean curve construction，and the optimal IMF is selected from the iterative screening results of different weights based on the minimum orthogonality. Therefore，WMEMDAN can improve the decomposition ability of CEEMDAN. At the same time，the decomposition results under different weights are screened to ensure that each order of IMF component is optimal，which can reduce the residual noise and false components. The analysis of simulation experiment signal shows that WMEM DAN has advantages in reducing false components and improving decomposition accuracy. The proposed method is applied to bearing fault diagnosis and gear fault diagnosis. The results show that the proposed method is effective and superior.