Finite Element Model Updating of Levenberg-Marquardt Method Based on Trust Region
The finite element model updating method based on sensitivity often encounters difficulties due to the ill-conditioning or rank-deficiency issues of the sensitivity matrix.However,the traditional Levenberg-Marquardt(L-M)algorithm,while addressing these problems,has a high computational cost.To address this,a modified L-M algorithm based on the trust region method is proposed.The updating speed and accuracy differences between the two algorithms are compared through a simulation case of a T-tail wing.The results show that both algorithms can generally update the parameters to the target values,but the improved algorithm has a lower computational cost,demonstrating its advantage in updating efficiency.Using a steel T-tail wing model as an example,the first eight natural frequencies of the actual model and the finite element model are obtained through experiments and finite element analysis,respectively.The improved algorithm is used to update the T-tail wing model,and after three iterations,the absolute average error of the first eight frequencies is reduced from 13.02％to 1.67％,significantly enhancing the frequency accuracy.This demonstrates that the improved algorithm also holds substantial engineering application value.
model updatingfinite element modelmodified L-M methodnatural frequency
万方数据
维普
