桥梁断面颤振导数识别的加权整体最小二乘法

IDENTIFICATION FOR FLUTTER DERIVATIVES OF BRIDGE DECK CROSS SECTION BY WEIGHTING ENSEMBLE LEAST-SQUARE METHOD

在已有研究的基础上 ,基于桥梁节段模型自由衰减振动记录 ,提出了一种识别桥梁断面颤振导数的新方法———加权整体最小二乘法 (WELS)。该方法将同一风速下多次自由振动记录作为一个整体 ,在整体残差平方和最小的意义上同时对该风速下多次竖向和扭转运动时程曲线进行非线性最小二乘拟合 ,获取多条自由振动曲线共同的阻尼和刚度信息 ,从而提高了颤振导数识别的精度。为了避免弱信号被强信号“淹没” ,并使各时程信号的识别误差相当 ,识别过程中对各试验记录引入了加权因子 ,使各试验记录在整体残差分析中具有相同权重。此外 ,该方法对多次试验记录同时进行拟合 ,在一定程度上减弱了个别信号中有色噪声对迭代收敛性的影响 ,增加了非线性最小二乘迭代过程的稳定性。为考查加权整体最小二乘识别方法的有效性 。

Based on the decaying free motion records of bridge dec k section model,a new method,namely Weighting Ensemble Least Square (WELS) method, is presented to identify flutter derivatives of bridge deck cross section.In thi s method,the multi-times free vibration records tested at a same wind velocity a re regarded as an ensemble.Many experimental records of the ensemble are simulta neously fitted to identify the common mode parameters of those experimental reco rds by nonlinear least-square method,which can increase the accuracy for inderct ification of flutter derivatives.Weighting factors are proposed for all experime ntal records in the identifying process to make each experimental record with th e same weight in sense of total residual error.In addition,the ensemble of data records can reduce the effect of the noise of few records on the convergence of least-square iteration and increase the stability in the process of least square iteration.Finally,wind tunnel tests of two bridge decks sections models were ca rried out to identify their flutter derivatives,which can indicate the effective ness of weighting ensemble least-square method.