http://www.cnr.it/ontology/cnr/individuo/prodotto/ID196217

Improved feature selection based on genetic algorithms for real time disruption prediction on JET (Articolo in rivista)

Type

Label

Improved feature selection based on genetic algorithms for real time disruption prediction on JET (Articolo in rivista) (literal)

Anno

Http://www.cnr.it/ontology/cnr/pubblicazioni.owl#doi

Alternative label

G.A. Rattá; J. Vega; A. Murari; JET-EFDA Contributors (2012)
Improved feature selection based on genetic algorithms for real time disruption prediction on JET
in Fusion engineering and design; ELSEVIER SCIENCE SA, PO BOX 564, 1001 LAUSANNE (Svizzera)
(literal)

Http://www.cnr.it/ontology/cnr/pubblicazioni.owl#autori

Pagina inizio

Pagina fine

Http://www.cnr.it/ontology/cnr/pubblicazioni.owl#altreInformazioni

http://biblioproxy.cnr.it:2127/S0920379612003201/1-s2.0-S0920379612003201-main.pdf?_tid=9c310ad4-cf7c-11e2-ab5d-00000aacb362&acdnat=1370614668_4c6adc704977f832d18863f28c508513 La rivista è pubblicata anche online con ISSN 1873-7196 (literal)

Http://www.cnr.it/ontology/cnr/pubblicazioni.owl#url

Http://www.cnr.it/ontology/cnr/pubblicazioni.owl#numeroVolume

Rivista

Http://www.cnr.it/ontology/cnr/pubblicazioni.owl#pagineTotali

Http://www.cnr.it/ontology/cnr/pubblicazioni.owl#numeroFascicolo

Note

Http://www.cnr.it/ontology/cnr/pubblicazioni.owl#affiliazioni

a GATEME, Facultad de Ingeniería, Universidad Nacional de San Juan, Avda. San Martín 1109 (O), 5400 San Juan, Argentina; b Asociación EURATOM/CIEMAT Para Fusión, Avda. Complutense 40, 28040 Madrid, Spain; c Associazione EURATOM-ENEA per la Fusione, Consorzio RFX, 4-35127 Padova, Italy; d JET-EFDA, Culham Science Centre, OX14 3DB Abingdon, United Kingdom. (Rattá, G.A.ad , Vega, J.bd, Murari, A.cd) (literal)

Titolo

Improved feature selection based on genetic algorithms for real time disruption prediction on JET (literal)

Abstract

The early prediction of disruptions is an important aspect of the research in the field of Tokamak control. A very recent predictor, called \"Advanced Predictor Of Disruptions\" (APODIS), developed for the \"Joint European Torus\" (JET), implements the real time recognition of incoming disruptions with the best success rate achieved ever and an outstanding stability for long periods following training. In this article, a new methodology to select the set of the signals' parameters in order to maximize the performance of the predictor is reported. The approach is based on \"Genetic Algorithms\" (GAs). With the feature selection derived from GAs, a new version of APODIS has been developed. The results are significantly better than the previous version not only in terms of success rates but also in extending the interval before the disruption in which reliable predictions are achieved. Correct disruption predictions with a success rate in excess of 90% have been achieved 200 ms before the time of the disruption. The predictor response is compared with that of JET's Protection System (JPS) and the ADODIS predictor is shown to be far superior. Both systems have been carefully tested with a wide number of discharges to understand their relative merits and the most profitable directions of further improvements. (literal)

Editore