模型参数化方法的一种改进算法

针对线性结构的有限元模型,并基于时域内的响应测量信息,建立了直接识别单元水平结构物理参数的改进算法;通过引入浮点运算指标Flop,定量分析了改进算法与原算法的计算效率.数值仿真了一大型空间网壳结构的参数识别,并对比分析了改进算法与原算法在相同计算条件下的识别效率,结果表明:在保证无噪声条件下识别精度100%时,改进算法的计算效率提高了约30倍;且改进算法可以有效地识别未知输入条件下的结构参数.由此得出结论,改进算法不仅能够大幅度提高参数识别的计算效率,而且易于扩展到未知输入情况下的参数识别,因而特别适合于大型结构的参数识别.

A modified algorithm of model parameterization method

A modified algorithm is proposed for the model parameterization method based on the sensitivity of response force vectors.Against the finite element model of linear structures and based on the tested responses in time domain,the modified algorithm was established to identify structural physical parameters on the element level.Through introducing a floating-point calculation index of Flop,the computing efficiencies of both the modified algorithm and the original method were quantitatively analyzed.The parametric identification of a large spatial shell composed of spatial beam elements was numerically simulated,and the identification efficiencies of both the modified algorithm and the original method in the same computing environment were compared.The results show that the computing efficiency of the modified algorithm is increased by about 30 times while the identification accuracy of 100% is guaranteed in the case of no noise,and the structural parameters in the condition of unknown inputs may be effectively identified using the modified algorithm.A conclusion is given that the modified algorithm is not only able to improve the computing efficiency of parameter identification,but also easy to extend to the parameter identification in the case of unknown inputs,therefore it is well suitable for the parameter identification of large structures.