Vibration signal denoising method based on CEEMDAN and its application in brake disc unbalance detection

The research objective is the application of the modal decomposition methods in dynamic balancing machine detection. This paper analyzes the problem of feature extraction of dynamic unbalanced signals by using complementary EEMD with adaptive noise (CEEMDAN). CEEMDAN was introduced to decompose the original unbalanced signal. The orthogonal coefficient method, wavelet soft threshold method, and autocorrelation function (ACF) was introduced to remove the irrelevant intrinsic mode functions (IMFs). Moreover, we compare and analyze these methods through simulation and experiments. Simulation results demonstrate that the CEEMDAN method greatly reduces the reconstruction error. The feasibility of the proposed method has been verified experimentally, the real machine signal is sampled and the actual imbalance information is extracted for comparison, which proves the superiority of CEEMDAN over EMD and EEMD in practical applications.