In view of the fact that the selection of noise amplitude and population average number parameters in complementary ensemble empirical mode decomposition(CEEMD) depends on personal experience

the traditional empirical mode decomposition(EMD) produces mode aliasing and the computation of population average empirical mode decomposition(EEMD) is too large

an adaptive resampling parameter optimization CEEMD decomposition method is proposed. In this method

the original signal is resampled by cubic spline interpolation to increase the sampling frequency

and a pair of positive and negative white noise is added. The noise amplitude is set as 0.01 SD

and the overall average number is set as 2. The optimal resampling frequency is determined by the change of maximum correlation coefficient between the component and the original signal after CEEMD decomposition.After the optimal resampling frequency is selected

the decomposition effect of the improved method is significantly improved. Through the simulation and experimental signal verification

the research significantly improves the decomposition performance of CEEMD

and is applied to the local fault analysis of planetary gear

the results show that it can extract accurate features and has obvious effect.