Qiu Jiandong, Zhang Qiang, Tang Minan, Lin Dingqiang, Liu Jiaxuan, Xu Shusheng
School of Mechanical Engineering, Lanzhou Jiaotong University, Lanzhou 730070, China.
School of Automation and Electrical Engineering, Lanzhou Jiaotong University, Lanzhou 730070, China.
Sensors (Basel). 2024 Jul 17;24(14):4638. doi: 10.3390/s24144638.
In light of the issue that the vibration signal from an axle-box bearing collected during the operation of an electric multiple unit (EMU) is seriously polluted by background noise, which leads to difficulty in identifying fault characteristic frequency, this paper proposes a resonance-based sparse signal decomposition (RSSD) and variational mode decomposition (VMD) method based on sparrow search algorithm (SSA) optimization to extract the fault characteristic frequency of the bearing. Firstly, the RSSD method is utilized to decompose the signal based on the obtained optimal combination of quality factors, resulting in the optimal low-resonance component with periodic fault information. Then, the VMD method is performed on this low-resonance component. The parameter combinations for both methods are optimized utilizing the SSA method. Subsequently, envelope demodulation is applied to the intrinsic mode function (IMF) with maximum kurtosis, and fault diagnosis is achieved by comparing it with the theoretical fault characteristic frequency. Finally, experimental validation and comparison are conducted by utilizing simulated signals and example signals. The results demonstrate that the proposed method extracts more obvious periodic fault impact components. It effectively filters out the interference of complex noise and reduces the blindness of setting weights on parameters due to human experience, indicating excellent adaptability and robustness.
针对动车组运行过程中采集到的轴箱轴承振动信号受到背景噪声严重污染，导致故障特征频率难以识别的问题，提出一种基于麻雀搜索算法（SSA）优化的基于共振的稀疏信号分解（RSSD）与变分模态分解（VMD）方法来提取轴承的故障特征频率。首先，利用RSSD方法基于获得的品质因数最优组合对信号进行分解，得到具有周期性故障信息的最优低共振分量。然后，对该低共振分量进行VMD方法处理。利用SSA方法对两种方法的参数组合进行优化。随后，对峭度最大的本征模态函数（IMF）进行包络解调，并与理论故障特征频率进行比较实现故障诊断。最后，利用仿真信号和实例信号进行实验验证与对比。结果表明，所提方法提取出的周期性故障冲击分量更为明显，有效滤除了复杂噪声的干扰，减少了因人为经验对参数设置权重的盲目性，具有良好的适应性和鲁棒性。
