基于S变换谱阈值去噪的冲击特征提取方法

发布时间：2018-06-15 18:13

本文选题：故障诊断 + 冲击特征　；参考：《振动与冲击》2014年21期

【摘要】：提纯旋转机械设备故障振动信号中的冲击特征,可以有效地实现相关故障的诊断。利用S变换适合于处理冲击特征信号的特点,提出基于S变换谱阈值去噪的冲击特征提取方法。先将信号进行S变换,得到其时频谱。考虑到此S变换谱为一复数矩阵,故而根据谱系数的模值大小进行阈值去噪。去噪过程中分别采用了基本的硬阈值函数和软阈值函数。对于最优阈值的估计,以所提出的改进风险函数为评价标准,利用步长迭代算法在零到系数最大模值的区间内获取。最后将去噪后的时频谱进行S逆变换,重构得到时域冲击特征。仿真信号与滚动轴承故障振动信号的处理结果表明,利用所获取的最优阈值,S变换谱阈值去噪方法能够从噪声混合信号中提取出冲击特征,从而实现相关故障的诊断。
[Abstract]:To purify the shock characteristics of the fault vibration signal of the rotating machinery, the diagnosis of the related fault can be realized effectively. Using the S transform to deal with the characteristics of the shock characteristic signal, the method of extracting the impact feature based on the S transform spectrum threshold denoising is proposed. The signal is transformed by S to get the time spectrum. Considering the S transform spectrum is one complex In the process of de-noising, the basic hard threshold function and the soft threshold function are used in the denoising process. The estimation of the optimal threshold is taken with the proposed improved risk function as the evaluation criterion and the step length iteration algorithm is used in the interval of the maximum modulus value of zero to coefficient. The time spectrum after noise is replaced by S inversion, and the time domain impact characteristics are reconstructed. The results of simulation signal and rolling bearing fault vibration signal processing show that the S transform spectrum threshold de-noising method can extract the impact feature from the noise mixed signal and realize the related fault diagnosis.
【作者单位】：浙江大学现代制造工程研究所;
【基金】：国家自然科学基金项目(51275453)
【分类号】：TN911.4

【参考文献】