Denoising Method for GyroWheel Signal Based on Iterative CEEMDAN-IT

Aiming at the non-stationary current signal in the GyroWheel (GW) calibration test, this paper proposes a denoising method based on iterative complete ensemble EMD interval threshold (CEEMDNAN-IT). Firstly, the signal is decomposed into intrinsic mode functions (imfs) through CEEMDAN, and the problem of mode mixing in the decomposition is solved. Secondly, imfs are divided into noise mode, mixed mode and information mode by continuous mean square error (CMSE) and l
2
-norm. For the mixed mode, the interval threshold calculated by the hurst exponent is adopted to denoise each imf. Then the denoising process is iteratively performed by randomly altering the order of the 1st imf to reduce the sensitivity of the EMD to abnormal point in signals. Finally, the output signal of the GW prototype is processed to remove noise effectively and improve the signal quality.