工程结构模态的连续型随机子空间分解识别方法

Stochastic subspace identification method based on continuous model for modal parameters of engineering structures

环境振动识别方法利用结构的输出信号识别结构的模态参数 ,主要的识别方法有时间序列分析法、ERA (eigensystemrealizationalgorithm)法和随机子空间法 ,这些方法均基于离散模型 .基于连续随机子空间模型 ,本文给出了一种识别大型工程结构模态参数的方法 .运用SVD(singularvaluedecomposition)分解将含噪声的输出信号空间分解为信号空间和噪声空间 ,然后直接估计结构的模态参数 .SVD分解保证了算法的鲁棒性 .最后讨论了一个 7层框架的理想建筑 ,仿真计算表明 ,该方法简单有效 ,能够使用在桥梁和建筑的健康监测和振动控制中 .

Ambient vibration method is identification of the modal structure parameters by the output data. Main identification methods are based on disperse space model, such as time serials analysis, ERA (eigensystem realization algorithm) method and Stochastic subspace method. Based on the continuous Stochastic subspace model, an identification approach was investigated to estimate structural modal under operating conditions. The output signal space was decomposed into signal space and noise space by SVD (singular value decomposition) method, then the modal parameters were estimated. The SVD method ensured the algorithm's robustness. Finally, the modal structure parameters of a 7-story steel frame building were discussed. The numerical simulation shows that the method is simple and effective, and it can be used in health monitoring and vibration controlling for bridges and architectures.