基于形函数方法快速识别结构动态荷载的试验验证

Experimental validation of a fast dynamic load identification method based on load shape function

在动态荷载识别中常常由于矩阵的病态性影响识别的精度,利用有限元理论中的形函数逼近荷载曲线,将识别离散的荷载历程转化为计算有限的形函数权重,从而显著改善反卷积法识别荷载中存在的采样时间长或采样频率高时数值求解困难的问题;并能改善反问题的病态性,提高对噪声的鲁棒性。一个连续梁的数值算例比较验证了该方法在5%的高斯噪声影响下能精确地识别未知荷载。悬臂梁试验中,通过实测的结构动态响应,在移动时间窗内利用荷载形函数方法可以实现激励的在线识别。

A dynamic load identification method was introduced and complemented,in which the force was approximated by using the shape functions in the finite element theory.The computational work was reduced a lot by calculating the weights of shape functions instead of indentifying the discrete load time history and the performance of the deconvolution method in the condition of long sampling time period or high sampling frequency was improved.Moreover,the ill-conditioning of the inverse problem was corrected,and the robustness to noise was strengthened.The numerical example of a continuous beam verifies that the load can be identified precisely by this method under the influence of 5% Gaussian noise pollution in the signals.In a cantilever beam experiment,via the measured structural responses,the method enables the online load identification which is performed repeatedly in a moving time window.