采用不同方法识别结构模态参数的比较

结构振动信号中混杂的噪声会影响模态参数的识别精度,并会引起虚假模态。通过一五自由度结构系统的数值仿真研究表明:直接识别法的精度低且应用范围有限,应用频域平均法可提高模态参数的识别精度,并且有利于识别出被噪声淹没的模态;修正的LSCE小波阈值滤波法有明显的去噪效果,其提高模态参数识别精度可达到或超过3次平均的效果且识别速度快。通过比较3种方法的结构模态识别结果,得到了修正的LSCE小波阈值滤波法节约试验成本,提高了识别速度和精度,具有明显的可适用性和推广性。

Comparative study on the identification of structural modal parameter using different methods

Noise in the structural vibration signal affects the identification accuracy of the modal parameter and generates false mode. The direct identification method is with low accuracy and limited in application scope, while the frequency domain average method could improve the identification accuracy of the modal parameter and be good at identifying the modes drown in noise. The improved LSCE wavelet threshold filtering method,is with remarkable de noise effect, which can reach or even surpass the effect of three times averaging in identification accuracy, and with fast identification. By comparing three structural modal identification methods, it shows that the improved LSCE wavelet threshold filtering method can save test costs and improve the recognition speed and accuracy. The improved LSCE wavelet threshold filtering method shows obvious applicability and generalization.