AR模型功率谱改进的频域模态自动识别方法

Automated Modal Identification Method in Frequency Domain Improved by AR Model Power Spectrum

针对频域分解法在模态自动识别上的不足,提出了一种基于AR模型功率谱改进的频域模态自动识别方法——AR-AFDD法。首先,通过AR模型功率谱估计确定所有可能是物理模态的频率点,并对相应功率谱矩阵进行奇异值分解,然后,利用时间振型相关性对所选取的峰值进行判别,实现物理模态与噪声模态的自动区分,最后,以自功率谱的最优拟合完成阻尼比估计;采用试验模型和实际桥梁对AR-AFDD法进行验证。研究表明:AR-AFDD法能够快速有效地实现结构模态参数的自动识别,识别结果能够区分物理模态与噪声模态,在实际结构的健康监测数据分析中具有良好的可靠性。

Aiming at the shortcomings of frequency domain decomposition method in automatic modal identification,an improved automated modal identification——AR-AFDD method in frequency domain based on AR model power spectrum was proposed.Firstly,all possible frequency points of physical mode were determined through AR model power spectrum estimation,and the corresponding power spectrum matrix was decomposed by singular value.Secondly,the selected peak value was distinguished by the correlation of time vibration mode to realize the automatic distinction between physical mode and noise mode.Finally,the damping ratio was estimated by the optimal fitting of the self-power spectrum.The AR-AFDD method was verified by the experimental model and the actual bridge respectively.The study shows that the AR-AFDD method can quickly and effectively realize the automatic identification of structural modal parameters,and the identification results can distinguish the physical mode from the noise mode,which has good reliability in the analysis of the actual structural health monitoring data.