POWRE: A Bayesian Approach in Developing Models for the Spatial Variability of Seismic Ground Motions

9870509 Zerva A POWRE奖将支持PI在加州理工学院进行的研究，为期一年，她将在那里与J.L.贝克教授合作。这项联合研究将把PI在确定地震地面运动空间变异性原因方面的专业知识与Beck博士在统计系统识别方面的专业知识结合起来。地面运动的空间变化可能会影响桥梁和管道等生命线的响应，而不是均匀运动所能承受的荷载。现有的模型一般是基于对单一地点和单一事件记录的数据的回归分析。因此，它们不能轻易地应用于不同的地点和事件，这限制了它们的用途。为了克服这一限制，本研究将把物理参数纳入空间变异性模型，必要时使用系统识别来估计这些参数。该方法将系统辨识方案嵌入到一个统计框架中，该框架考虑了模型参数和模型本身预测中的不确定性。将使用贝叶斯方案，允许使用来自不同地点和事件的额外信息来更新预测概率，直到模型稳定或收敛。该项目将通过拓宽她的研究视野来促进她的职业生涯，特别是在统计系统识别、基于可靠性的优化设计和稳健结构控制领域。她希望将这些知识应用于未来与东海岸老旧破旧基础设施的安全和可靠性相关的研究中，她的机构就位于东海岸。***

9870509 Zerva A POWRE award will support for one year the conduct of research by PI at California Institute of Technology, where she will collaborate with Prof. J.L. Beck. The joint study will combine the PI's expertise in identifying the causes of spatial variability in seismic ground motions with Dr. Beck's expertise in statistical system identification. The spatial variation of ground motions may affect the response of lifelines, such as bridges and pipelines, beyond the loads expected from uniform motions. Existing models are based, generally, on regression analyses of data recorded at a single site and for a single event. Thus, they cannot readily be applied to different sites and events, which limits their usefulness. In order to overcome this limitation the present study will incorporate physical parameters in the spatial variability model, parameters which are estimated, where necessary, using system identification. The approach will embed the system identification scheme in a statistical frame work, that accounts for uncertainties in the model parameters and in the predictions from the model itself. A Bayesian scheme will be used, allowing the predictive probabilities to be updated with additional information from various sites and events, until the models stabilize or converge. The project will enhance the PI's career by broadening her research horizons, specifically in the areas of statistical system identification, reliability-based optimal design and robust structural control. She expects to apply this knowledge in future research related to the safety and reliability of the older and decaying infrastructure of the East Coast, where her home institution is located. ***