摘要详情

ID / 提交时间

111 / 2025-05-15 13:59:27

标题

LEASGgram: an adaptive detection method of locomotive bearing repetitive transient impulses with automatic frequency band segmentation

关键字

Fault detection, Frequency band segmentation, Strong interference, Transient impulses.

主题及专题

全体主题

状态

全文待审

作者

Binghuan Cai / Beijing University of Chemical Technology

Gang Tang / Beijing University of Chemical Technology

摘要

Repetitive transient impulses detection methods are often hindered under the severe operating conditions of locomotive real-world engineering applications. One of the primary challenges in existing approaches is the use of fixed spectral segmentation strategies, which fail to account for the unique characteristics of the spectrum, often leading to the loss of critical fault signatures. Additionally, current state-of-the-art indicators are highly sensitive to strong background noise and depend heavily on prior knowledge of fault-related frequencies. To address these limitations, this paper proposes a novel fault detection method called LEASGgram, specifically designed to identify repetitive transient impulses under severe disturbance conditions. The method introduces an adaptive frequency band segmentation technique based on multi-scale spectral analysis, enabling automatic partitioning of the frequency domain according to its intrinsic distribution features. Furthermore, a new evaluation index is developed from the perspective of the log envelope autocorrelation spectrum, which effectively captures both the sparsity and cyclostationarity properties of the signal. This allows for accurate localization of the resonance bands excited by repetitive fault-induced transients. Compared with conventional signal processing techniques, the proposed LEASGgram demonstrates superior performance in extracting informative frequency bands with enhanced fault relevance and reduced noise interference, even under harsh operational conditions.