融和核能与人工智能，助力能源技术发展

Wei HUANGDirector, Division of Planning Information and Knowledge ManagementDepartment of Nuclear Energy, IAEA 29-04-2025Fusion of Neutrons and Neural Nets: Energizing energy frontiers.Global AI landscape Stakeholder EngagementOrganizational and Safety CultureChange and Knowledge ManagementDiversity and Inclusion§ Rapid expansion in all sectors driven by advancements in machine learning, big data, and computing power. § Growing investment from government and tech giants to race AI harness.§ Regulatory efforts to ensure ethical and safe AI use.§ Enormous potential to accelerate nuclear technological development, from science to energy to medicine and others§ Nuclear Power Opportunities from AI expansion. . Harness AI for nuclear to address pressing global challenges Stakeholder EngagementOrganizational and Safety CultureChange and Knowledge ManagementDiversity and InclusionAI in Nuclear PowerAI in Food and AgricultureAI in Water and the environmentAI in Nuclear SecurityAI in Nuclear FusionAI in SafeguardsAI in Human healthNuclear Power empowered by AIIAEAPREVENTIONDETECTIONRESPONSE• Design: supporting advanced nuclear reactor design and analysis with AI and ML, e.g. more accurate models in fewer changes, improved characterization of safety margin and efficient and faster analysis.• Construction: Identifying potential risks (schedule delays, cost overruns and supply chain disruptions) for enhanced real-time decision-making• Operations: Detecting anomalies, forecasting potential equipment failures, reducing human error and improving reliability of operation.• Maintenance: Enabling predictive maintenance and reducing unplanned downtime with AI powered prediction and prognostics. • Hazardous environment: Reducing human exposure to hazardous with drones, robots, and AI-driven machines and automation.• Knowledge and competency: preserving critical knowledge for sustainable operation and maintaining HR competency.Specific applications in NPP operationStakeholder EngagementOrganizational and Safety CultureChange and Knowledge ManagementDiversity and Inclusion§ In-Service Inspections: AI automates analysis of non-destructive testing data (e.g., ultrasonic, radiographic inspections).§ Control Rod Drive Mechanism (CRDM) Monitoring: ML detects anomalies in coil currents (~96% accuracy), preventing rod drops.§ Anomaly Detection: AI identifies early warning signs of equipment failures before human operators notice them.§ Work Process Automation: Drones with computer vision reduce the need for manual plant inspections.§ Human-Computer Interaction: Speech/gesture recognition in control rooms improves operator efficiency.§ Natural Language Processing (NLP): Automates report analysis and decision-making from event logs.§ Fuel Management & Core Design: AI optimizes fuel loading patterns and reactor control strategies.AI concerns and uncertaintiesIAEA§ Regulatory & Safety Concerns: Lack of standards for AI in safety-critical nuclear applications.§ Explainabilit