Research and Improvement Methods for EMD on the Basis of Statistical Models

Empirical Mode Decomposition (EMD) is a powerful adaptive method for decomposing signals into intrinsic mode functions (IMFs), enabling the separation of slow and fast oscillations. However, EMD faces challenges such as mode mixing and spurious modes, particularly when dealing with intermittent oscillatory behavior and noisy data. This article explores the theoretical and practical aspects of EMD and its variants, including Ensemble EMD (EEMD), Complementary EEMD (CEEMD), and Complete EEMD with Adaptive Noise (CEEMDAN). It delves into the impact of signal parameters like phase, signal-to-noise ratio (SNR), sampling frequency, and amplitude ratios on EMD performance. The study proposes an improved algorithm, DEC, which addresses the limitations of EMD by ensuring mathematically complete decomposition results. Through experimental simulations and theoretical derivations, the article provides a comprehensive analysis of EMD's behavior under various conditions, offering insights into its decomposition capabilities and suggesting pathways for enhancement. The findings highlight the importance of considering noise and sampling frequency in EMD applications, paving the way for more accurate and reliable signal processing techniques.