【作者】 尤杰；

【导师】 车轶；

【作者基本信息】 大连理工大学 ， 结构工程， 2010， 硕士

【摘要】 随着服役时间的推移,城市既有钢筋混凝土建筑的性能不断退化,影响结构的安全使用,衰减结构的预定功能。研究既有钢筋混凝土结构的性能,不仅可以揭示潜在危险,而且研究成果还可以直接用于指导结构设计。本文以我国存在大量城市既有建筑为背景,利用人工神经网络模型和非线性有限元分析软件OpenSEES为工具,从材料、构件和整体结构三个层次研究了城市既有混凝土结构的性能。首先,总结分析国内外有关混凝土强度预测研究成果的基础上,运用MATLAB神经网络工具箱,建立了既有建筑混凝土强度退化的计算模型,研究了既有建筑材料性能退化的情况。本文所建立的模型主要考虑了混凝土强度随时间的退化规律,其成果可用于大面积普查既有建筑的抗震性能。其次,在收集国外大量抗剪试验数据的基础上,用神经网络模型来研究钢筋混凝土梁的抗剪性能。通过对各国规范的抗剪设计方法和神经网络模型预测方法的对比分析,总结得出各国规范的适用性,而且研究表明,本文的神经网络模型具有很好的计算精度,计算结果明显好于各国的经验公式和理论计算方法。最后,本文在地震工程模拟平台OpenSEES上建立了—榀服役期为10年的十层三跨钢筋混凝土框架纤维模型,对其进行Pushover分析,研究了现浇楼板对结构破坏模式的影响。根据是否考虑现浇楼板效应,分为两种工况,分别进行静力弹塑性分析。分析结果表明：现浇楼板的存在增大柱的转动、减小梁的转动,削弱了设计时所期望的“强柱弱梁”效果,改变了梁柱达到屈服状态与极限状态的先后顺序。更多还原

【Abstract】 As the service time, the performance of city’s existing reinforced concrete buildings continue to deteriorate, affect the safe use of the structure, attenuate its intended function. Study the performance of existing reinforced concrete structures can not only reveal the potential risk, but research also can be directly used to guide the structural design.In this paper, against the background of having a lot of cities’existing buildings in China, using artificial neural network model and finite element analysis software OpenSEES as a tool, from the materials, components and overall structure of the three levels to study existing concrete structure of the performance of the city.First, on the basis of summarize and analyze the research results of prediction of concrete strength results concrete strength, using MATLAB neural network toolbox, established the degradation model of concrete strength of existing buildings to study the performance degradation of the existing building materials. The model in this paper mainly considers the degradation of concrete strength with time rules, the results can be used to survey a large area of the seismic performance of existing buildings.Second, based on a large number of shear test data abroad, using the neural network model to study the components of reinforced concrete beam shear properties. Shear design of the national standard method and neural network prediction method were compared, the results show that this neural network model has good accuracy, its results significantly better than the experience and theoretical calculation formula.Finally, a three-span and 10 layers reinforced concrete frame was built in OpenSEES with fiber model, by pushover analysis to study the mode of casting slab on the impact of structural damage. Whether to consider casting slab or not, two conditions is divided, and respectively carry on Static Analysis. The presence of cast slab weakened design expected "strong column and weak beam" effect.Through study of this article, we can effectively reveal degradation laws of existing buildings’material and component, and the affect of cast slab on the seismic capacity of the whole structure.更多还原