基于奇异值差分谱理论的大型转子轴心轨迹提纯

针对大型轴承实验台的转子轴心轨迹的提纯问题，提出采用奇异值差分谱提纯轴心轨迹的方法。奇异值差分谱可以直观表示有用分量和噪声分量的差异，由差分谱首个峰值可以确定有用分量个数。将原始振动信号构造Hankel矩阵，对其进行奇异值分解(Singular value decomposition，SVD），利用奇异值差分谱来选取特征奇异值，通过SVD重构得到特征信号，利用特征信号得到提纯的轴心轨迹。比较了SVD和谐波小波包算法的处理效果，结果表明，由奇异值差分谱提纯的轴心轨迹更清晰。
Abstract：Aiming at purification of rotor axis’s orbit in a large bearing testing platform,an approach was proposed based on the difference spectrum theory of singular value.The difference spectrum could directly describe the difference between the beneficial component and noise.The number of the useful components could be determined by the first peak of the difference spectrum.The Hankel matrix was constructed from original vibration signal and was decomposed by the singular value decomposition (SVD).The feature singular values were selected using the difference spectrum and feature components were obtained by the SVD reconstruction,and then the purified rotor axis’s orbit could be obtained with the feature components.The processing effect of SVD and harmonic wavelet packet algorithm was compared,and the results show that the axis orbit purified by the difference spectrum of singular value was much clearer.