An adaptive ultra-narrow band filtering method based on flexible sliding band segmentation

Abstract

Searching for the optimal frequency band is the key step in state feature extraction. However, the actually selected optimal frequency band is often inaccurate or the in-band noise is obvious, which greatly affects the accuracy of feature extraction. Therefore, a novel flexible filtering method is proposed in this paper to realize adaptive period impulse feature extraction, which is called adaptive sliding Ramanujan decomposition (ASRD). Firstly, ASRD method realizes the adaptive segmentation of ultra-narrow band by flexible sliding band segmentation, which not only improves the noise robustness, but also avoids the destruction of the state feature band structure. Then, a reweighted fusion index (RFI) is constructed with excellent period impulse sensitivity, interference component robustness and monotonicity, so as to evaluate the state features of ultra-narrow band sub-modes, adaptively select effective sub-modes and reconstruct period impulses. Finally, the RFI is used to determine optimal decomposition level and select the optimal elastic filter components (EFC), so as to realize the adaptive extraction of period impulses. The analysis results of simulation and experimental signal can verify the effectiveness and superiority of ASRD method.