Real-Time AI Decision Support for Emergency ResponseThe NeedDuring rapidly evolving industrial emergencies, operators must interpret complex, time-dependent sensor data and determine which accident pathways could produce severe consequences. Conventional event-tree and fault-tree methods can struggle to represent interactions among equipment, processes, software, and human actions in real time. This can delay identification of high-consequence scenarios and limit the technical guidance available for emergency classification, mitigation, and protective-action decisions. The TechnologyOSU engineers have developed a real-time decision-support platform that filters monitored and simulated scenario data, groups scenarios with similar behavior, and applies trained neural networks to identify pathways most likely to produce undesirable outcomes. The platform combines scenario screening with conventional or modular neural-network architectures and can present results for human review. Demonstrations described in the documents use simulated nuclear-plant accident scenarios and radiological-consequence classifications. Commercial Applications
Benefits/Advantages
Patents
|
Tech IDT2018-197 CollegeLicensing ManagerZinn, Ryan InventorsCategoriesExternal Links |