基于径向基神经网络的系杆拱桥索力确定与模型修正

发布时间：2018-05-16 10:00

本文选题：有限元模型修正 + 径向基神经网络　；参考：《合肥工业大学》2015年硕士论文

【摘要】：跟随着有限元技术的深入发展,有限元模型修正技术在过去的20多年也得到了广泛研究及应用。目前流行的有限元模型参数型修正法通过对模型参数的选取与修正,使模型结构响应与实际结构响应趋于一致,从而建立能够反映以及预测实际结构行为的有限元模型。在土木工程领域,由于针对结构建立有限元模型时基于一定的假设,以及实际结构在建设和运营中受外界荷载及环境的影响,有限元模型与实际结构存在误差,该模型通常不能够反映结构的实际工作状态。因此,为了确保模型计算结果能够表达实际结构状况,应对其进行有限元模型修正。近年来,大跨桥梁结构的健康监测与安全状态评估成为了国内外的热点研究领域,而有限元模型修正技术能够为此提供一个可靠且实效的模型来反演桥梁结构的工作及健康状况甚至识别结构损伤部位与程度等等。因此,有限元修正技术具有重要的工程研究意义。全文研究内容及安排如下：(1)回顾有限元模型修正技术发展历程,归纳总结有限元模型修正方法,分析各种方法优缺点。(2)利用径向基函数神经网络(radial basis function, RBF)进行有限元模型修正,并结合有限元模型修正问题的特点,对网络输入输出参数进行分析,选择构建RBF网络结构参数,确定径向基神经网络修正有限元模型以及确定施工索力的技术路线。(3)通过假设既有状态,利用RBF神经网络对一预应力混凝土简支小箱梁有限元模型进行修正,从而验证该方法的有效性。(4)利用RBF神经网络确定系杆拱桥吊杆施工索力张拉,完成有限元模型修正扩展应用。(5)根据图纸采用梁格法建立梁拱组合体系桥梁,首先利用主成分分析法对网络训练样本进行预处理网络输出输入样本对,然后根据实测桥梁结构响应数据,利用训练检验后的RBF神经网络进行有限元模型修正。修正后有限元模型计算值与实测值之间误差较原有设计模型有很大程度改进。
[Abstract]:With the further development of finite element technology, finite element model modification technology has been widely studied and applied in the past 20 years. By selecting and modifying the parameters of finite element model, the structural response of the model is consistent with the response of the actual structure, and the finite element model which can reflect and predict the behavior of the actual structure is established. In the field of civil engineering, because the finite element model is based on certain assumptions, and the actual structure is affected by the external load and environment in the construction and operation, there are errors between the finite element model and the actual structure. The model does not usually reflect the actual working state of the structure. Therefore, in order to ensure that the model results can express the actual structure, it is necessary to modify the finite element model. In recent years, health monitoring and safety state assessment of long-span bridge structures have become a hot research field at home and abroad. The finite element model modification technique can provide a reliable and effective model for retrieving the work and health of bridge structure and even identifying the damage location and degree of the structure and so on. Therefore, the finite element correction technique has important engineering significance. The contents and arrangements of this paper are as follows: (1) reviewing the development of finite element model modification technology, summarizing the finite element model modification methods, analyzing the advantages and disadvantages of each method. 2) using radial basis function neural network radial basis function, RBF) to modify the finite element model. Combined with the characteristics of the finite element model correction problem, the input and output parameters of the network are analyzed, and the parameters of the RBF network structure are selected. Determining the modified finite element model of radial basis function neural network and determining the technical route of cable force in construction. The finite element model of a prestressed concrete simply supported small box girder is modified by RBF neural network. Therefore, the validity of the method is verified. The RBF neural network is used to determine the cable tension in the construction of tied arch bridge, and the finite element model modification and extension application. 5) according to the drawing, the beam lattice method is used to establish the beam arch composite system bridge. Firstly, the network training samples are preprocessed by principal component analysis (PCA), and then the finite element model is modified by using the RBF neural network after training and checking according to the measured bridge structure response data. The error between the calculated value and the measured value of the modified finite element model is much improved than that of the original design model.
【学位授予单位】：合肥工业大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：U441;U448.225

【相似文献】