The existence of periodic impacts in collected vibration signal is the representative symptom of rolling bearing localized defect. Due to the complicacy of the working condition, the fault-related impacts are usually submerged in other ingredients. This article proposes an adaptive Resonance-based Sparse Signal Decomposition (RSSD) for extracting the fault features of rolling bearings. Adaptive RSSD is constructed to fetch the impacts from collected vibration signal, by making RSSD decomposed signal kurtosis value maximum using Lion Swarm Algorithm (LSA). Multipoint Optimal Minimum Entropy Deconvolution Adjusted (MOMEDA) is further performed to enhance the amplitude and periodicity of impacts contained in RSSD decomposed signal, so that fault feature is highlighted. The superiority and availability of proposed strategy are validated by applying to single fault feature extraction of an experimental dataset and compound faults feature extraction of a locomotive rolling bearing.

收集到的振动信号中存在周期性冲击是滚动轴承局部缺陷的典型症状。由于工作条件的复杂性，与故障相关的影响通常被淹没在其他成分中。本文提出了一种自适应的基于共振的稀疏信号分解（RSSD），用于提取滚动轴承的故障特征。通过使用Lion Swarm算法（LSA）使RSSD分解信号的峰度值最大，可构造自适应RSSD来从收集的振动信号中提取影响。进一步执行多点最佳最小熵反卷积调整（MOMEDA），以增强RSSD分解信号中包含的影响的幅度和周期性，从而突出显示故障特征。