基于遗传算法的降阶模型修正方法研究

以实验模态数据修正降阶模型为研究对象，由于所测量的特征对阶数通常小于模型阶数，造成了特征方程为超越方程。为了求解超越方程以确定质量与刚度修正矩阵，提出随机给定超越方程中的多余变量值，采用遗传算法来寻优。为了使修正矩阵的范数较小，对多余变量值进行了约束处理。为了达到修正目的，通过不断改变多余变量值，并采用基于实代码遗传算法来确定优化结果，最后给出了一个计算实例。该算法的优点是可以得到多组最优阶，精度较高；缺点是计算量较大，不适合于动态模型修正。

Research on Modification of Model Reduction Based on Genetic Algorithms

In this paper,that using experimental mode data to modify calculating model is taken as the research object.Considering the eigen-equations belong to redundance equations because the measured eigenpairs are less than the orders of model,in order to solving the ones and get the modification matrices of mass and stiffness,the excrescent variables are given randomly.At one time,the excrescent variables are also limited to a given range so as to acquire a small norm of modified mass and stiffness matrices.Meanwhile,the real coded genetic algorithm is used to gain the modified mass and stiffness matrices.Finally,an example is given to demonstrate the effectiveness of the presented algorithm.