本项目关注既有砌体构件性能的量化分析，该问题涉及砌体跨尺度的力学行为和不确定性跨尺度传播的机制。首先，引入近场动力学方法模拟砌体微观损伤累积、细观裂缝自发产生和发展和宏观受力变形过程，并用于确定构件性能指标。其次，采用贝叶斯方法结合先验判断和实测数据，建立和修正砌体力学参数的联合概率模型，量化参数的不确定性。再次，通过合理方法从联合概率模型中抽取力学参数样本、并代入近场动力学模型，采用高性能计算获取构件性能指标的充足样本。在此基础上，建立高斯过程模型描述力学参数总体和构件性能指标总体间的关系，量化参数不确定性传播的机制。然后，按高斯过程模型的不确定性程度进行加权组合、基于贝叶斯方法建立和修正渐进全概率模型，量化模型的不确定性。最后，基于渐进全概率模型重构性能指标向量的联合概率密度函数。

The proposal focuses on the quantitative performance analysis of existing masonry members. The focus relates to the trans-scale mechanical behavior of masonry and the trans-scale propagation of uncertainty. First, we introduce the peridynamic method to simulate the cumulation of micro damages, the spontaneous formation and development of cracks in masonry, and the deformation of masonry members under loading. The simulation helps to determine the performance index of masonry members. Second, we use the Bayesian method to combine the prior judgement and measuring data, and then establish and update the joint probability density function (PDF) for the mechanical parameters of masonry. We quantify the uncertainty in this way. Third, we sample from the joint PDF of the mechanical parameters, substitute the samples into the peridynamic model, and then generate enough results for the performance index by high-performance computation. Based on enough samples, the Gaussian process (GP) models are established to represent the relationship between the mechanical parameters and performance index. We quantify the propagation of uncertainty in this way. As the next step, we combine the GP models according to their extent of uncertainty, and then establish and update the asymptotic total probability model following the Bayesian method. Finally, we determine the joint PDF of the vector of performance index using the asymptotic total probability model.