基于统计分析的有限元模型修正  

Finite element model updating based on statistics analysis

作　　者：秦玉灵[1] 孔宪仁[1] 罗文波[2] 

机构地区：[1]哈尔滨工业大学卫星技术研究所,哈尔滨150001 [2]北京空间飞行器总体设计部,北京100094

出　　处：《航天器环境工程》2011年第3期239-245,共7页Spacecraft Environment Engineering

基　　金："微小型航天器系统技术"长江学者创新团队发展计划(项目编号:IRT0520)

摘　　要：响应面模型代替有限元模型进行修正便于与粒子群算法等优化算法结合,有利于提高修正效率和精度。文章用某雷达卫星的简化有限元模型计算基于正交设计的各水平参数下模态频率,用方差分析确定待修正参数并构造二次响应面模型,用最小二乘法确定多项式系数,以响应面计算结果与实测结果的差值构造适应度函数并用之引导含混沌搜索机制的改进粒子群算法对待修正参数的偏移量进行寻优,修正后参数代入原有限元模型得到修正模型。修正后模型不仅能以高精度复现测试频段内的模态频率,而且能以一定精度预测测试频段外频率,证实了修正方法的有效性。It is convenient to combine the response surface model with the particle swarm optimization（PSO） and other optimization algorithms to replace the finite element model（FEM） for improving the updating efficiency and precision.Modal frequencies of a concise radar satellite were calculated by the FEM based on several sets of parameters obtained from the orthogonal design,the non-updated parameters were determined by the variance analysis,the quadratic polynomial was constructed,with its coefficients being calculated by the least squares method（LSM）,and the fitness function was constructed by the differences between the results of the response surface model and the FEM,which was used to guide the particle swarm with chaos search mechanism to search for the optimal solution of the non-updated parameters to constitute the updated model.The modal frequencies of the updated model can both reproduce the modal frequencies of the benchmark model in the test range with high precision and predict those out of the test range with acceptable precision,which shows the validity of the model updating method.

关 键 词：响应面模型 有限元模型 混沌搜索机制 粒子群算法 正交设计 方差分析 

分 类 号：V414.1[航空宇航科学与技术—航空宇航推进理论与工程]