基于实测数据的新型大跨度空间结构模型修正研究

A STUDY OF MODEL CORRECTION OF THE LARGE SPAN SPATIAL STRUCTURE BASED ON MEASURED DATA

空间网格结构通常是由相同或相似的子结构按照某种规则在三维空间范围内排列而成,由于组成结构的杆件相似、繁多且密集,致使结构具有较高的超静定次数,且频谱分布十分密集,这给空间结构的模态识别和模型修正带来了挑战。基于实测数据,首先运用ANSYS有限元软件建立精确的数值分析模型用于模型修正,同时通过实测手段采集到环境荷载激励下的结构温度数据和振动数据,继而采用频域分解法(FDD法)、自然激励技术联合特征系统实现算法(NEx T+ERA)识别出结构的模态频率和振型,最后通过混合优化算法修正结构节点刚度、杆件刚度和结构质量等参数来实现对结构有限元模型的修正。修正结果显示:该修正方法可大大缩减有限元模型模态频率与结构真实模态频率间的误差,达到了修正目的。

A spatial grid structure usually consists of the same or similar substructures in a regular arrangement in three dimensional space and has highly degree of statical indeterminacy and intensive spectrum distribution due to dense bars,which brings a challenge to modal identification and model updating of large span spatial structure. In the article,the accurate analysis model was firstly built,and measured temperature and vibration data were obtained and then modal frequencies and shapes were identified by methods of FDD and NEx T ＋ ERA. In the end,by correcting parameters such as joint stiffness,bar stiffness and structural mass etc,the finite element model was updated through the hybrid optimization algorithm. The results showed that method could greatly reduce the deviation between modal frequencies and shapes,with which the goal was achieved.