我国目前高层建筑大量涌现，合理评估高层建筑在遭受地震后的性能具有重要意义。本项目基于新的全概率PBEE方法对钢筋混凝土高层结构进行性能评估，重点解决其中的关键问题，包括地震动谱形特性对高层建筑结构性能的影响、弯矩与剪力耦合作用对剪力墙性能的影响、开洞后由于连梁作用引起的联肢剪力墙复杂非线性问题。针对以上问题分别提出：采用考虑地震动频谱成分和峰值加速度的双参数地震动指标；考虑高阶振型影响，按振型质量参与系数加权平均选择条件均值谱中的条件周期T*和地震动谱型参数ε(T*)，定量给出地震动谱形特性对高层建筑结构易损性曲面和结构倒塌率的影响；将弯矩等效成力偶的形式加载的剪力墙弯剪耦合低周往复试验；分别开发出普通配筋和对角斜筋连梁的非线性剪切滞回模型；采用机器学习方法筛选合理的剪力墙的损伤指标。最后，量化各种不确定性参数(地震动输入、结构、损伤和损失不确定性)，计算相关概率和累计损失。

With the rapid development of economy in China, tall buildings have been widely constructed and distributed in the past decades. As earthquake frequently occurs in China, it is necessary and has significantly social meaning to reasonably evaluate the seismic performances of these tall buildings subjected to earthquake. This project focuses on seeking the solutions to key techniques in the seismic performance evaluation of reinforced concrete (RC) tall buildings based on a novel law of total probability used in performance-based earthquake engineering (PBEE), including the effect of ground motion (GM) frequency spectrums on the seismic performances, the effect of interaction between moment and shear force on the RC shear walls and the effect of coupling beams on coupled wall systems. To solve these key issues, the double-parameters GM indicator considering frequency spectrums and peak ground acceleration (PGA) is adopted to quantitatively present the effect of GM frequency spectrums on the fragility curve surface and collapse probability of RC tall building structures; in order to consider the impaction of the high mode, modal mass participating in the coefficient weighted to uniformly is used to select conditional period T* and GM spectrum shape parameter ε(T*) of conditional mean spectrums; the moment is equivalent to a force pair loading on the RC shear walls quasi-static cyclic test to study the effect of interaction between moment and shear force; developing nonlinearly analytical model of conventionally and diagonally reinforced coupling beams to couple the effect of coupling beams on coupled wall systems; meanwhile, machine learning algorithm is utilized to rationally choose the damage indicators of RC shear walls. Finally, quantify various uncertain parameters (e.g., uncertainties in GM inputs, structures, damages, and loss) and calculate the relevant probability and cumulative loss to assess the seismic performance of reinforced concrete tall buildings.