Advanced signal processing methods and learning systems applied to monitor the NPP reactor core condition were the theme of an open workshop organized in cooperation with the European Horizon 2020 CORTEX project in UJV Rez, Czech Republic.
The Workshop, dedicated to nuclear power station research staff, research organizations and academia, was attended by over four dozen experts from 10 European countries, including experts from NPPs, universities, research organizations and the Nuclear Research Institute.
The full-day workshop agenda was divided into 5 sessions, including 14 lectures presented by 10 lecturers.
DOWNLOAD PRESENTATIONS
- Welcome, UJV Introduction (V. Fiser – UJV Rez)
- Basic EU Cortex information (C. Demazière – Chalmers University of Technology)
- UJV activity in the field of fuel cycles and noise plant diagnostics (F. Havluj – UJV Rez)
- UJV activity in the field of fuel cycles and noise plant diagnostics (P. Stulik – UJV Rez)
- CIIRC introduction (L. Lhotska – Czech Institute of Informatics, Robotics and Cybernetics, Czech Technical University in Prague)
- Neutron noise simulation for the development and testing (C. Demazière – Chalmers University of Technology)
- Challenges in signal processing in complex applications (L. Lhotska, V. Gerla – CIIRC CVUT)
- Noise analysis in plant data processing (P. Stulik – UJV Rez)
- Reconstruction of ex- and in-core neutron signals (J. Machek, UJV Rez)
- Signal processing methodologies for nuclear reactor data (G. Alexandridis – National Technical University of Athens)
- The computational intelligence paradigm for the efficient, accurate and robust processing of nuclear reactor data (T. Tambouratzis – University of Piraeus)
- Learning methodologies: issues of data, features, and long-term trends (M. Macas – CIIRC CVUT)
- State-of-the-art machine learning methodologies (G. Leontidis – University of Lincoln)
- Deep learning: 3D convolutional and recurrent neural networks in reactor perturbations unfolding and anomaly detection (G. Leontidis – University of Lincoln)
The alive and vital discussions confirmed that the new advanced processing and learning methodologies are the crucial aspect of the effective and reliable core diagnostics based on the knowledge of reactor inversion function.
The one day workshop was followed by a working meeting of WP3, which was attended by 10 specialists, who actively participate in the project. The meeting was concentrated on further progress in WP3 and WP4, discussions about the new simulation data sets and on real NPP data processing challenges.