He Xiaoliang, Zhao Feng, Song Nianyun, Liu Zepeng, Cao Libing
School of Mechanical Engineering, Southeast University, Nanjing 211189, China.
College of Design and Engineering, National University of Singapore, Singapore 117576, Singapore.
Sensors (Basel). 2025 Jul 16;25(14):4421. doi: 10.3390/s25144421.
To address the challenges of weak fault features and strong non-stationarity in early-stage vibration signals, this study proposes a novel fault diagnosis method combining enhanced variational mode decomposition (VMD) with a structurally improved GoogLeNet. Specifically, an improved wild horse optimizer (IWHO) with tent chaotic mapping is employed to automatically optimize critical VMD parameters, including the number of modes and the penalty factor , enabling precise decomposition of non-stationary signals to extract weak fault features. The vibration signal is decomposed, and the top five intrinsic mode functions (IMFs) are selected based on the kurtosis criterion. Time-frequency features are then extracted from these IMFs and input into a modified GoogLeNet classifier. The GoogLeNet structure is improved by replacing standard × convolution kernels with cascaded 1 × and × 1 kernels, and by substituting the ReLU activation function with a parameterized TReLU function to enhance adaptability and convergence. Experimental results on two public rolling bearing datasets demonstrate that the proposed method effectively handles non-stationary signals, achieving 99.17% accuracy across four fault types and maintaining over 95.80% accuracy under noisy conditions.
为解决早期振动信号中故障特征微弱和非平稳性强的挑战，本研究提出了一种将增强型变分模态分解（VMD）与结构改进的GoogLeNet相结合的新型故障诊断方法。具体而言，采用带有帐篷混沌映射的改进型野马优化器（IWHO）自动优化关键的VMD参数，包括模态数和惩罚因子，从而能够精确分解非平稳信号以提取微弱故障特征。对振动信号进行分解，并基于峭度准则选择前五个本征模态函数（IMF）。然后从这些IMF中提取时频特征，并将其输入到改进的GoogLeNet分类器中。通过用级联的1×和×1内核替换标准的×卷积内核，并将ReLU激活函数替换为参数化的TReLU函数来改进GoogLeNet结构，以增强适应性和收敛性。在两个公共滚动轴承数据集上的实验结果表明，该方法能够有效处理非平稳信号，在四种故障类型上的准确率达到99.17%，在噪声条件下的准确率保持在95.80%以上。
