基于人工神经网络的藏式结构有限元模型修正

Finite element model updating of tibetan structure based on artificial neural network

针对典型藏式传统建筑年代久远、建造过程无设计图纸、存在太多不确定性,通过有限元数值分析所得结果与实验结果不能较好吻合问题,建立所测结构的有限元模型,并通过人工神经网络方法,以实测结构模态参数为目标对典型藏式结构的有限元模型中部分不确定因素—梁及雀替的等效变截面梁高、材料密度及弹性模量进行修正,得到更接近真实状态的有限元模型。该模型对该典型藏式结构的损伤识别、可靠度评估具有重要意义。

Most of Tibetan-style ancient structures are age-old, and there are no design drawings in the construction process, so a lot of uncertain factors may appear in the structural analysis. The results obtained by finite element analysis usually can not be in good agreement with experimental results. When the error between them is larger, the dynamic characteristics of the structural model acquired by the finite element method and the actual measurement may have greater discrepancy, even beyond the engineering precision required in practice. The finite element model of the measured structure was modified to make it closer to the true state by using the artificial neural network (ANN) method. In the method, the measured modal parameters were adopted as the target to update some uncertain factors of the finite element model of a typical Tibetan structure, such as the height of equivalent non-uniform beam, the material density and the elastic modulus. The finite element model after updating is significant for damage identification and reliability assessment of typical Tibetan structures.