非线性最优化方法在振动系统模态参数识别问题中的应用

Application of Nonlinear Optimization on Identification of Model Parameters of Vibration System

非线性最小二乘问题常用解法是高斯牛顿法,但用于结构动力学的模态参数识别问题时常遇到不少困难。本文采用增加阻尼以克服病态,一维寻优以加速收敛,利用其它结构动力学方法以尽可能减少初始迭代向量的误差等,使该方法得以实用。并通过二个工程结构实例对上述问题做了说明。

The general method for solving nonlinear leastsquare problem is Gauss-Newton's method. But there are some difficulties in using this method for identification of model parameters of vibration system. In this paper, due to applying damping factor to overcome sickly equation,by using one dimensional optimization to accelarate convergence, and adapting other dynamics method to determine initial iterative recter with less error, we meet with actual success of this method. Two engineering examples are presented for explaining above discussion.