The experimental results of 210.mat.

Large rotating machinery is an essential piece of equipment in modern industry, playing a critical role in industrial production. However, the complex working environment complicates the extraction of fault-related information. This paper proposes a fault diagnosis method based on the subtraction-average-based optimizer (SABO) and feature mode decomposition (FMD). To address the issue that FMD’s decomposition performance is highly sensitive to its parameter settings, this paper uses the minimum envelope entropy as the fitness function and employs the SABO algorithm to adaptively optimize FMD’s two key parameters: the mode number (n) and filter length (L). Additionally, for the intrinsic mode functions (IMFs) obtained from FMD decomposition, the maximum kurtosis value is used to filter IMFs containing fault information, and envelope spectrum analysis is applied to achieve fault diagnosis. When applied to experimental signals of rolling bearing faults, the results demonstrate that the proposed method can extract the amplitude of the fault characteristic frequency from the envelope spectrum and accurately diagnose the fault type. Compared with methods based on empirical mode decomposition (EMD) and fixed-parameter FMD, the proposed method provides a more prominent representation of the fault characteristic frequency and its harmonics in the envelope spectrum. Furthermore, the proposed method achieves a more prominent representation of the fault eigenfrequency in the envelope spectrum and a lower error rate. The proposed method demonstrates significant potential and value for rolling bearing fault diagnosis.

大型旋转机械是现代工业的关键装备，在工业生产中发挥着至关重要的作用。然而，复杂的工作环境使得故障相关信息的提取难度大幅提升。本文提出了一种基于减平均优化器（subtraction-average-based optimizer, SABO）与特征模式分解（feature mode decomposition, FMD）的故障诊断方法。针对FMD分解性能对参数设置高度敏感的问题，本文以最小包络熵作为适应度函数，采用SABO算法自适应优化FMD的两个关键参数：模态数（n）与滤波器长度（L）。此外，针对FMD分解得到的本征模态函数（intrinsic mode functions, IMFs），本文采用最大峭度值筛选出包含故障信息的IMF，并通过包络谱分析实现故障诊断。将该方法应用于滚动轴承故障实验信号后，结果表明所提方法可从包络谱中提取故障特征频率的幅值，并准确诊断故障类型。与基于经验模态分解（empirical mode decomposition, EMD）和固定参数FMD的方法相比，所提方法在包络谱中对故障特征频率及其谐波分量的表征更为突出。进一步而言，所提方法在包络谱中对故障固有频率的表征更为显著，且错误率更低。所提方法在滚动轴承故障诊断领域展现出显著的应用潜力与应用价值。