Single-sensor AI maps quake risks across 139 points inside a nuclear power plant in real time
Civil EngineeringMaterials ScienceOperations
THE AI ANGLE
Inferring real-time structural vibrations and quantifying inspection risks from a single seismometerResearchers from KRISS and UNIST developed a deep learning virtual sensing technology that uses data from a single seismometer to infer real-time vibrations across 139 unsensored points inside a nuclear power plant. By quantifying uncertainty and the probability of exceeding risk thresholds, the model helps operators prioritize post-earthquake inspections and significantly mitigate costly operational downtime. This approach addresses practical constraints where dense sensor installation is prevented by radiation hazards and licensing requirements, and computational analyses take days to complete.
THE TEACHING ANGLE
Students can examine the trade-offs between physical sensing limitations in high-hazard environments and relying on AI-driven virtual sensing with quantified uncertainty to direct critical operational inspection workflows.Read the original at techxplore.com Generate teaching or study materials
More in Civil Engineering
- The ERP reckoning: Decades of customization could block AI valueCIO.com · September 15, 2026
- HPE combats memory constraints with supplier help, better forecastingSupply Chain Dive · September 15, 2026
- The biggest issues with delivery robots are exactly what you'd thinkEngadget · September 15, 2026
- AI-powered inspection system gives 3D printers 'a brain behind the eyes'Phys.org — Technology · September 15, 2026
- Robots and AI uncover unexpected pathway in classic 135-year-old chemical reactionPhys.org — Chemistry · September 15, 2026