How to Extract Information and Knowledge from Fusion Massive Databases
- Consorzio RFX-Associazione EURATOM ENEA per la Fusione, 1-35127 padova (Italy)
- Asociacion EURATOM-CIEMAT para Fusion, CIEMAT, Madrid (Spain)
- EURATOM/UKAEA Fusion Association, Culham Science Centre, Abingdon (United Kingdom)
- Max-Planck-Institut fuer Plasmaphysik, Teilinstitut Greifswald, EURATOM Association, Wendelsteinstr. 1, 17491 Greifswald (Germany)
The need to understand and control the dynamics of reactor grade fusion plasmas requires the analysis of increasing amounts of data, which at JET can reach easily the level of several GBytes per shot. Therefore a series of new approaches are being pursued to store the data and to retrieve the required information. They range from loss less data compression techniques, to wavelets and Structural Pattern Recognition methods. Since the information available is very often affected by high level of uncertainties and the phenomena to be studied are complex and nonlinear, the inference problems in this field of plasma physics are particularly delicate. Even in this perspective innovative solutions are under development. In particular a range of Soft Computing approaches have already been implemented at JET. The most successful are Bayesian statistics for the integration of diagnostic measurements, Data Mining techniques to study the nonlinear correlation of various variables and Fuzzy Logic to include the knowledge of the experts even if formulated in linguistics terms. Specific methodologies are being investigated for real time control, since it poses some specific issues. In this field, a combination of hardware and software solutions, like Cellular Nonlinear Networks, are often necessary to satisfy the needs of both speed and reliability.
- OSTI ID:
- 21136817
- Journal Information:
- AIP Conference Proceedings, Vol. 988, Issue 1; Conference: International conference on burning plasma diagnostics, Varenna (Italy), 24-28 Sep 2007; Other Information: DOI: 10.1063/1.2905116; (c) 2008 American Institute of Physics; Country of input: International Atomic Energy Agency (IAEA). JET EFDA Contributors; ISSN 0094-243X
- Country of Publication:
- United States
- Language:
- English
Similar Records
PipeSight: A High-Performance Computing Platform for Pipeline Integrity Management
Qualitative information modeling: The role of fuzzy logic in project economic evaluations