RAT矩函数在模态振形数据去噪中的应用

Denoising Application of RAT Moment Function to Modal Shape Data

提出将RAT矩函数应用于三维结构模态振型数据去噪的方法,使用RAT矩函数对结构的振型数据进行矩变换,从振型数据中提取出包含所有振型信息的特征矩集,将幅值较小的矩值作为噪声从特征矩集中剔除,再利用特征矩集重构振型图像,即能实现对振型数据去噪。采用一个仿真的机匣结构,以正态分布的随机噪声模拟真实情况下的噪声,在不同噪声强度下利用RAT矩对振型数据去噪,来验证RAT矩函数的去噪能力。结果表明:RAT矩函数具有良好的抗噪性,能大幅度减小噪声对模态振型的影响。

RAT moment function was proposed to denoise the modal shape data of a three-dimensional structure.The RAT moment function was adopted to extract feature moment sets,which contained the vibration information of the structure.Then small moments were removed as noise from the feature moment set.Using the feature moment set to reconstruct the mode shape image,the mode data could be denoised.Taking a finite element model as an example,the noise under actual conditions was simulated with randomly distributed noise.Under different noise intensities,RAT moment was applied to denoise mode shape data,which verified the denoising capability of RAT moments.The results show that the RAT moments are of good noise resistance and able to reduce noise substantially.