结合部虚拟材料参数的逆向优化识别方法

摘要

考虑结合部的有限元模型是准确分析、预测结构动态响应与振动控制的基础.以哑铃状结构实验模型为研究对象, 采用虚拟材料层法对结合部进行等效参数化建模.综合模态实验数据和有限元优化仿真技术, 提出了一种简单实用的结合部虚拟材料参数逆向优化识别方法.为关联有限元模型和实验数据, 特引入基于仿真和实验固有频率之间差值最小化的目标函数.采用灵敏度法对整体结构模型做局部有限元反解, 利用多目标遗传算法优化求解结合部虚拟材料参数.通过对比实验模型、正向模型和逆向模型得到的固有频率, 验证了所提方法的正确性和可行性, 并进一步探讨了不同预紧力矩对结合部虚拟材料参数的影响.%A consideration of the joint involved into the finite element (FE) model is the foundation of accurate analysis and prediction of structural dynamic response and vibration control.Taking a dumbbell structure as research object, the parametric modeling of the joint is given using the virtual material layer method.Based on the modal test data together with the FE optimization and simulation technologies, a simple and practical method to inversely optimize and identify the virtual material parameters of joints is proposed.An objective function based on minimizing the discrepancy between the simulation and experimental natural frequencies is introduced to establish the linkage between the FE model and the test data.By using the sensitivity method, the local FE inverse solution of the whole structural model is carried out, and then the multi-objective genetic algorithm is used to optimize and solve the virtual material parameters of the joint.Comparisons between the natural frequencies from the test model, forward model and inverse model are made to validate the correctness and feasibility of the proposed method.Besides, further study is carried out on the influence of different pre-tightening torques on the virtual material parameters of joints.