Uncertainty analysis of decomposition level choice in wavelet threshold de-noising (Q657509)

Summary: The complexities of various noises, which are quantified by wavelet energy entropy (WEE) and differential coefficient of WEE (D(WEE)), are first analyzed and their uncertainties then estimated and described using confidence intervals. Then, quantitative criterion for judging the WEE and D(WEE) difference between noisy series and noise is put forward, based on which the decomposition level (DL) choice method in wavelet threshold de-noising proposed in 2010 by the authors [ibid. 12, No. 6, 1499--1513 (2010; Zbl 1229.62125] is improved. Finally, analytical results from examples verified the performance of the improved method, and also demonstrated its much wider applicability; moreover, the DL chosen using it is more reliable because of the fact that uncertainty is taken into consideration.