关键词: 复杂信号去噪, CEEMDAN, 非平稳性度量 (NS), S-G 滤波, 小波阈值去噪

Abstract:

Noise reduction in complex environments is of paramount importance for the accurate extraction and analysis of signals. Prevailing research predominantly centers on the application of single or combined methods in specific scenarios, but encounters challenges in effectively addressing the nonlinearity and non-stationarity of complex signals. Based on the measurement of non-stationarity, this study proposes a two-stage noise reduction strategy using CEEMDAN. The NS index is used to quantify the non-stationarity of modal components, achieving precise separation of high-frequency noise from low-frequency signals. A novel logarithmic threshold function is adopted to remove high-frequency noise, and the SG filtering method is combined to smooth low-frequency signals, significantly improving the noise reduction effect and signal reconstruction accuracy. The results indicate that the new method demonstrates outstanding modal discrimination and noise reduction performance under different signal-to-noise ratios and noise types.

Key words: complex signal denoising, CEEMDAN, non-stationarity measure (NS), S-G filtering, wavelet threshold denoising