摘要
全量补偿法的提出为部分输入未知条件下的结构参数识别以及荷载反演提供了一个很好的思路,但由于该算法在进行参数估计时没有考虑已知输入与未知输入的可信度差别,因此参数收敛过程中会产生振荡现象,收敛速度相对较慢。在此基础上,充分利用部分输入可确知而部分输入未知的激励特性,构造了一个基于加权最小二乘准则的改进算法。与原算法相比,改进算法不仅在理论上更加完备,而且其收敛特性也有质的改善。在同等的参数识别精度条件下,其所需的迭代次数仅为原算法的十分之一。
The total compensation method (TCM) provides an effective approach to the solution of the composite inversion problem with incomplete input information. But the algorithm neglects the difference of certainty between known and unknown input when estimating the parameters, and therefore causes instability and slow parameter convergence. In this paper, an updated total compensation method (UTCM) based on weighted least square criterion is proposed, which makes full use of the incomplete input information. The new algorithm is improved significantly in both theory and convergence. The efficiency and accuracy of the new algorithm is investigated numerically and the results are compared with those obtained using TCM. It is shown that the computing time needed for UTCM is 10 times less than that for TCM.
出处
《工程力学》
EI
CSCD
北大核心
2005年第1期28-32,共5页
Engineering Mechanics
基金
国家自然科学基金项目(59978015)
关键词
结构工程
复合反演
全量补偿
改进算法
收敛速度
Convergence of numerical methods
Inverse problems
Least squares approximations
Parameter estimation
Structural loads