用正则化Lanczos迭代法进行模型修正  

REGULARIZED LANCZOS METHOD FOR MODEL UPDATING

作　　者：吴颉尔[1,2] 戴华[1] 

机构地区：[1]南京航空航天大学理学院,南京210016 [2]江苏科技大学数理学院,镇江212003

出　　处：《振动与冲击》2008年第10期65-69,共5页Journal of Vibration and Shock

基　　金：国家自然科学基金资助项目(编号:10271055)

摘　　要：用灵敏度方法进行有限元模型修正,常遇到病态和亏秩的线性方程组,当模态数据含有测量误差时,最小范数解常常没有物理意义。解决这类问题的有效方法是正则化方法。讨论用正则化Lanczos方法进行有限元修正,其中正则化参数用L曲线确定。数值模拟结果表明,该方法能够较好地进行模型修正。In model updating,linear equations derived from the sensitivity approach are often ill-conditioned and with rank deficiency.When the modal measurements are contaminated by noise,the minimum norm solution rarely leads to physically meaningful updated parameters.The regularization is an efficient method for solving these problems.Here,the regularized Lanczos method is used to update a FEM model.The L-curve criterion is used to determine the regularization parameter.Numerical examples show that the regularized Lanczos method can better be used for model updating.

关 键 词：Lanczos迭代法 正则化 模型修正 有限元模型 灵敏度方法 

分 类 号：TJ765.365[兵器科学与技术—武器系统与运用工程]