基于GMPSO的有限元模型修正方法验证  

VALIDATION OF FINITE ELEMENT MODEL UPDATING METHODOLOGY BASED ON GMPSO

作　　者：夏志远 李爱群[3,4] 李建慧[5] 陈鑫[1,2] XIA Zhi-yuan;LI Ai-qun;LI Jian-hui;CHEN Xin(School of Civil Engineering,Suzhou University of Science and Technology,Suzhou,Jiangsu 215011,China;Jiangsu Province Key Laboratory of Structure Engineering,Suzhou University of Science and Technology,Suzhou,Jiangsu 215011,China;School of Civil Engineering,Southeast University,Nanjing,Jiangsu 210096,China;Beijing Advanced Innovation Center for Future Urban Design,Beijing University of Civil Engineering and Architecture,Beijing 100044,China;School of Civil Engineering,Nanjing Forestry University,Nanjing,Jiangsu 210037,China)

机构地区：[1]苏州科技大学土木工程学院,江苏苏州215011 [2]苏州科技大学江苏省结构工程重点实验室,江苏苏州215011 [3]东南大学土木工程学院,江苏南京210096 [4]北京建筑大学北京未来城市设计高精尖创新中心,北京100044 [5]南京林业大学土木工程学院,江苏南京210037

出　　处：《工程力学》2019年第10期66-74,共9页Engineering Mechanics

基　　金：国家自然科学基金项目(51278104);江苏省自然科学基金面上项目(BK20161581);江苏省高等院校自然基金项目(12KJB560003);江苏省结构工程重点实验室资助课题项目(ZD1803);苏州科技大学自然科学基金青年项目(XKQ2018008);苏州科技大学人才引进科研资助项目;江苏省“青蓝工程”

摘　　要：为提高有限元模型修正方法效率,保证修正精度,提出基于高斯白噪声扰动的粒子群优化(GMPSO)有限元模型修正方法。介绍标准粒子群优化(PSO)方法和改进后的GMPSO方法,基于测试函数比对两种方法的全局寻优能力和寻优效率;提出高效的基于GMPSO有限元模型修正方法,阐述方法流程并明确各参数与实际物理量的对应关系;基于GMPSO有限元模型修正方法对高维有损伤简支梁模型(变量维度为10)实施修正,并与基于遗传算法(GA)的模型修正结果进行比对;基于GMPSO有限元模型修正方法对某在役桥梁结构实施修正(变量维度为13),验证所提方法可行性。结果表明:经局部改进的GMPSO方法较原PSO方法的优化能力显著提升;高维损伤简支梁模型修正结果显示,基于GMPSO模型修正方法可获得较好的修正结果,修正效率较基于GA的模型修正方法有显著提升;在役桥梁结构有限元模型修正结果显示,基于GMPSO模型修正方法可有效降低主梁计算频率和试验频率的误差,所提方法可适用于较工程复杂结构模型修正问题。A FE Model updating approach based on Gaussian white noise mutation particle swarm optimization(GMPSO)is proposed to improve the efficiency of the model updating method and keep its accuracy.The particle swarm optimization(PSO)and the improved method GMPSO were introduced and the global optimization searching capacity was compared between the two methods.The FE model updating method based on GMPSO was then proposed and the one-to-one match between the parameters in the method and the real physical variables was discussed together with the method process.The new approach was applied to the model updating of a damaged simply supported beam(DSSB)with high-dimension(with ten variables)and the updating results was compared with that of model updating method based on Genetic Algorithm(GA).Meanwhile,the method is also applied to an in-serviced bridge structure(with thirteen variables)to validate the feasibility of the proposed method.The results show that:the global optimization searching capacity of the improved GMPSO is significantly higher than that of PSO;the relationships between the parameters and physical variables are clear in the model updating method based on GMPSO and this is suitable for modularized program operation and facilitate the application of commercial software.The updating process of the DSSB with high-dimension using the proposed methodology reveals accurate results and its efficiency is significantly improved,compared with the process of model updating based on GA.The updating of the in-serviced bridge using the proposed methodology narrows the differences between the main girder vibration frequencies of model and those of test.The feasibility of the application of the proposed model updating methodology to the updating issue of high-dimensional and complex engineering structures is verified.

关 键 词：有限元模型 模型修正 粒子群优化方法 高斯白噪声扰动 优化效率 验证研究 

分 类 号：O224[理学—运筹学与控制论]