I-Corps: AI-Based Decision Support for Management of Bridge Networks

这个I-Corps项目更广泛的影响/商业潜力是开发决策支持软件系统，以优化运输网络的管理。这些系统解决了与网络资产相关的风险，以及可能导致自然灾害的紧急交通动态。包括国家机构和市政当局在内的各种实体的采购商可以根据物理基础设施的当前条件和预测的高优先级区域，使用该系统对交通基础设施进行最佳资产管理。该决策支持系统还可以增强疏散规划和灾后网络管理程序，因为它可以提供移动地图，在灾难期间和之后更新传入的数据。这个I-Corps项目是基于图形神经网络（GNN）模型的开发，该模型可用于估计概率自然灾害下的网络响应，计算时间最短。这些神经网络模型旨在计算各种响应措施，例如连接性，最短距离和旅行时间。GNN的模块化特征允许使用来自一个城市或地区的网络的数据开发的模型在另一个地区使用。这些模型可用于极端事件，如地震、洪水、飓风和龙卷风。所提出的技术是建立在成功的数字实验上的交通网络在加州，纽约，和佛罗里达的地震和洪水灾害。这些GNN模型的准确性，计算效率和鲁棒性的文件。基于这些GNN模型的决策支持将采用智能工具，评估物理基础设施的当前状况，对与网络规划相关的投资决策进行排名，以关注高优先级问题。该奖项反映了NSF的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

The broader impact/commercial potential of this I-Corps project is the development of decision support software systems for optimal management of transportation networks. These systems address risks associated with network assets as well as the emergent traffic dynamics that would likely ensue in natural hazards. Planners from various entities, including state agencies and municipalities may use this system for optimal asset management for transportation infrastructure, based on the current conditions of the physical infrastructure and the predicted high priority areas. This decision support system also may enhance evacuation planning and post-disaster network management procedures, as it can provide mobility maps that are updated with incoming data during and after a disaster.This I-Corps project is based on the development of graph neural network (GNN) models that may be used to estimate the network response under probabilistic natural hazards with minimal computational time. These neural network models are designed to calculate various response measures, such as connectivity, shortest distance, and travel times. The modular feature of GNNs allows for models that are developed using data from networks in one city or region to be usable in another region. These models may be used for extreme events such as earthquakes, floods, hurricanes, and tornadoes. The proposed technology is building upon successful numerical experiments on transportation networks in California, New York, and Florida for seismic and flood hazards. The accuracy, computational efficiency, and robustness of these GNN models are documented. The decision support that will be built on these GNN models will feature intelligent tools that evaluate the current conditions of the physical infrastructure, rank investment decisions related to network planning to focus on high priority concerns. It also may facilitate effective emergency response by combining the same predictive capabilities with real-time network data to effectively update response plans.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.