Wei Yu, Xu Yuanbo, Hou Yinlong, Li Long
School of Automation, Xi'an University of Posts and Telecommunications, Xi'an 710121, China.
Entropy (Basel). 2023 Aug 6;25(8):1171. doi: 10.3390/e25081171.
Random impulsive noise is a special kind of noise, which has strong impact features and random disturbances with large amplitude, short duration, and long intervals. This type of noise often displays nonGaussianity, while common background noise obeys Gaussian distribution. Hence, random impulsive noise greatly differs from common background noise, which renders many commonly used approaches in bearing fault diagnosis inapplicable. In this work, we explore the challenge of bearing fault detection in the presence of random impulsive noise. To deal with this issue, an improved adaptive multipoint optimal minimum entropy deconvolution (IAMOMED) is introduced. In this IAMOMED, an envelope autocorrelation function is used to automatically estimate the cyclic impulse period instead of setting an approximate period range. Moreover, the target vector in the original MOMED is rearranged to enhance its practical applicability. Finally, particle swarm optimization is employed to determine the optimal filter length for selection purposes. According to these improvements, IAMOMED is more suitable for detecting bearing fault features in the case of random impulsive noise when compared to the original MOMED. The contrast experiments demonstrate that the proposed IAMOMED technique is capable of effectively identifying fault characteristics from the vibration signal with strong random impulsive noise and, in addition, it can accurately diagnose the fault types. Thus, the proposed method provides an alternative fault detection tool for rotating machinery in the presence of random impulsive noise.
随机脉冲噪声是一种特殊的噪声，具有强烈的冲击特性以及大振幅、短持续时间和长间隔的随机干扰。这种类型的噪声通常呈现非高斯性，而常见的背景噪声服从高斯分布。因此，随机脉冲噪声与常见背景噪声有很大不同，这使得许多常用的轴承故障诊断方法不再适用。在这项工作中，我们探讨了在存在随机脉冲噪声的情况下轴承故障检测的挑战。为了解决这个问题，引入了一种改进的自适应多点最优最小熵反卷积（IAMOMED）方法。在这种IAMOMED方法中，使用包络自相关函数来自动估计循环脉冲周期，而不是设置一个近似的周期范围。此外，对原始MOMED中的目标向量进行了重新排列，以增强其实际适用性。最后，采用粒子群优化算法来确定用于选择的最优滤波器长度。基于这些改进，与原始MOMED相比，IAMOMED在随机脉冲噪声情况下更适合检测轴承故障特征。对比实验表明，所提出的IAMOMED技术能够有效地从具有强随机脉冲噪声的振动信号中识别故障特征，此外，它还能准确诊断故障类型。因此，所提出的方法为存在随机脉冲噪声的旋转机械提供了一种替代的故障检测工具。
