Real-Time AI Decision Support for Emergency Response

The Need

During 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 Technology

OSU 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

  • Nuclear power plant accident assessment and emergency-response support
  • Severe-accident management and operator decision-support systems
  • Radiological consequence classification and off-site response planning
  • Potential adaptation to other sensor-intensive infrastructure, process-safety, or anomaly-detection environments

Benefits/Advantages

  • Supports real-time prioritization
  • Handles time-dependent complexity
  • Reduces the scenario search space
  • Adaptable architecture

Patents

Patent # Title Country
11156995 SYSTEMS AND METHODS FOR SUPPORTING REAL-TIME EMERGENCY PLANNING United States of America
11720091 SYSTEMS AND METHODS FOR SUPPORTING REAL-TIME EMERGENCY PLANNING United States of America

Loading icon