基于Kriging模型与MOGA算法的有限元模型修正

Finite element model updating based on Kriging model and MOGA

精确的有限元模型可以准确预测真实结构的动力学响应,采用一种融合Kriging模型与多目标遗传算法(MOGA)的模型修正方法,针对GARTEUR飞机进行模型修正。首先采用Spearman相关性分析方法,引入显著性水平系数对飞机模型的初始参数进行筛选;然后将筛选后的参数作为设计变量,利用最佳填充空间(OSF)设计方法获得初始样本点,构建Kriging响应面模型,将响应面计算结果与实验结果的差作为目标函数;最后利用MOGA对目标函数进行优化,搜索Pareto最优解,并且对候选点添加验证点来检验其精度。结果表明:修正后的GARTEUR飞机模型具有良好的频率复现和预测能力,满足工程精度要求,采用融合Kriging模型与MOGA的模型修正方法具备有效性和可靠性。

The accurate finite element model can precisely predict the structural dynamics response.A model upda-ting method integrating the Kriging model and multi objective genetic algorithm(MOGA)is adopted for the model correction of GARTEUR aircraft.Firstly,the initial parameters of the aircraft model are screened using Spearman correlation analysis and the introduction of a significance level coefficient.And then,the screened parameters are used as design variables to obtain the initial sample points using the optimal space-filling(OSF)design method,construct the Kriging response surface model,and use the difference between the response surface calculation re‐sults and experimental results as the objective function.Finally,the MOGA is used to optimize the objective func‐tion,search for the Pareto optimal solution,and add validation points to the candidate points to check its accuracy.The results show that the updated GARTEUR aircraft model has good frequency reproduction and prediction capa‐bility,can meet the engineering accuracy requirements.The model updating method using Kriging model and MO‐GA is effective and reliable.