The Integrated Knowledge Engine, a tool for rapid diversion analysis of complex data in an operating facility

The domestic (NRC) requirements and international (IAEA) goals for spent fuel reprocessing plants require unprecedented levels of measurement precision and process monitoring for existing high throughput facilities, including Japan’s Rokkasho Reprocessing Plant (RRP). To address these requirements, the IAEA employs additional monitoring techniques for enhancement of safeguards and determination of correlations between systems for diversion analysis. What is actually needed is the ability to credibly and rapidly draw definitive safeguards conclusions from all monitoring systems. This effort requires massive data collection, integration, and timely assessment from many sensors. Sensors include cameras, radiation sensors, sensors for density, pressure, weight as well as images. This paper describes a methodology by which we can begin to achieve the above. Specifically, we look at extension of a system called IKE, the Integrated Knowledge Engine. IKE is a model based data fusion system that integrates multiple diverse information sources and sensors and could thus provide the following: 1) Fusion of sensor data to assess the probability of specified diversion scenarios. 2) Rapid analysis of potential diversion scenarios to rule out false alarms (i.e. atypical process operations). 3) Real time incorporation of IKE’s unique capability to request additional data for optimized determination of the probability of diversion. 4) Support for development of a combined safeguards figure of merit that can integrate qualitative techniques and quantitative measures for all sensor platforms (e.g. cameras and non-destructive assay) and quantify their relative worth.