Cyber-physical model of civil infrastructure system subjected to extreme loads

现代社会的公共安全和经济的稳定发展在很大程度上依赖于可靠的民用基础设施，如桥梁、隧道、核电站和高层建筑。虽然这些基础设施的设计寿命为几十年，但在遭受地震、火灾等极端事件以及频繁且日益不可预测的天气事件时，采用过时设计规范建造的基础设施是脆弱的。准确预测基础设施在极端负荷下的性能对于优化设计新基础设施以及制定现有基础设施的维护和改造策略至关重要。然而，预测民用基础设施在极端载荷下的性能一直是一个挑战，特别是在预测基础设施的故障时。挑战来自各种来源，例如缺乏：(1)结构部件的精确模型，可以复制部件在极端载荷下的失效过程；(2)民用基础设施竣工情况资料；（三）能对多轴边界条件下的全尺寸结构构件进行试验的试验设施；(4)外荷载的固有随机性和预测模型的不确定性。该提案的主要目标是将PI研究小组开发的UT-SIM框架扩展为民用基础设施的网络物理（CP）模型框架，以应对上述挑战。为实现这一目标，未来五年将开展四个项目：(1)超大规模多轴试验设备的子结构混合模拟控制方案；(2)大风或火灾作用下结构的鲁棒混合模拟方法；(3)基于机器学习算法和物联网传感器数据的网络模型的模型更新；(4)与物理模型相互作用的大型城市街区尺度网络模型。CP模型是混合（数值-实验）模型在概念上的进步。混合模型的中心前提是数字表示的组件足够精确，但通常情况并非如此。此外，物理表示元素的数量受到实验设备能力的限制。在CP模型中，民用基础设施的数值模型是通过使用各种支持物联网的传感器对真实结构或物理样本进行测量来更新的，而预计在非线性范围内运行且难以进行数值建模的主要部件则用物理样本来表示。拟议的研究计划和拟议的项目将形成一个模拟框架，用于准确预测民用基础设施在各种自然灾害中的性能。通过这个研究项目，不同的hqp将被训练成一套独特的技能，为行业做出贡献。

The safety of the public and stable economic development of a modern society depend highly on reliable civil infrastructure such as bridges, tunnels, nuclear power plants, and highrise buildings. While these infrastructures are designed to be operational for several decades, infrastructures constructed with outdated design codes are vulnerable when subjected to extreme events such as earthquakes, fires, as well as frequent and increasingly unpredictable weather events. Accurate prediction of infrastructures' performance when subjected to extreme loads is of the utmost importance to optimally design new infrastructures and to develop maintenance and retrofit strategies for existing ones. However, the prediction of civil infrastructures' performance subjected to extreme loads has been challenging, especially when predicting infrastructures' failure. The challenges arise from various sources such as a lack of: (1) accurate models of structural components that can replicate the failure process of the components subjected to extreme loads; (2) data on the as-built condition of civil infrastructures; (3) testing facilities that can test full scale structural elements subjected to multi-axial boundary conditions; and, (4) inherent randomness in the external loads and uncertainties in the prediction models. This proposal's main objective is to expand the UT-SIM framework developed in the PI's research group toward a framework for the cyber-physical (CP) model of civil infrastructure to address aforementioned challenges. To achieve the objective, four projects will be carried out in the next five years: (1) a control scheme for substructure hybrid simulation with mega-scale multi-axial testing equipment; (2) a robust hybrid simulation methods for structures subjected to strong wind or fire; (3) a model updating of a cyber-model based on a machine-learning algorithm and IoT sensor data; and (4) a large-scale city block scale cyber-model that interacts with a physical model. The CP model is a conceptual advancement of a hybrid (numerical-experimental) model. The central premise in the hybrid model is that the numerically represented components are sufficiently accurate, which is often not the case. Also, the number of physically represented elements is limited by the capacities of experimental facilities. In the CP model, the numerical model of civil infrastructure is updated through measurements from real structures or physical specimens instrumented with various IoT-enabled sensors, while the main components that are expected to behave in the nonlinear range and are difficult to be modelled numerically are represented with physical specimens. The proposed research program and the proposed projects will lead to a simulation framework for the accurate prediction of civil infrastructure's performance in various natural hazards. Through the research program, diverse HQPs will be trained with a unique set of skills to contribute to the industry.