Up to now, the occurrence of disruptions has proven to be an unavoidable aspect of Tokamak operation, particularly in high performance configurations, and therefore being able to predict them and undertake mitigation actions is a top priority on the route to the next step devices like ITER. In particular, it is now considered essential not only to predict their occurrence but also to be able to determine, with high probability, the type of disruption about to occur. This paper reports the results obtained using Generative Topographic Maps (GTMs) [1] to automatically classify several disruption classes manually classified by the JET physicists [2]. Using GTMs high dimensional data reduces to a small enough number of dimensions and can be visualized in the low dimensional (usually 2-D) embedding space. The 2-D GTM of the 10-D operational space of JET has been obtained using an extensive database of discharges selected from JET campaigns from C15 up to C27 (year 2005-2009) performed with the carbon wall. The obtained map shows not only a natural discrimination between safe and disruptive regions, but also that the different classes of disruptions tend to occupy different regions in the map. Monitoring the evolution of each disruptive discharge on the GTM it has been found that many of them evolves within the same region with an associated class of disruption. The success rate of the classification is extremely high, sometimes reaching 100%. Using such GTM with discharges selected from more recent campaigns (C28-C30), when JET changes the carbon wall with the new ITER-like wall, significantly deteriorates the classification performance, particularly on certain disruption classes. Hence, a new GTM has been built to represent the new operational space of JET. Using this map with all the disruptions occurred in these last campaigns the very high classification performance is confirmed and therefore the prospects for the deployment of this tool in real time are very promising

Advances in disruption classification at JET

PAU, ALESSANDRO;CANNAS, BARBARA;FANNI, ALESSANDRA;SIAS, GIULIANA
2013

Abstract

Up to now, the occurrence of disruptions has proven to be an unavoidable aspect of Tokamak operation, particularly in high performance configurations, and therefore being able to predict them and undertake mitigation actions is a top priority on the route to the next step devices like ITER. In particular, it is now considered essential not only to predict their occurrence but also to be able to determine, with high probability, the type of disruption about to occur. This paper reports the results obtained using Generative Topographic Maps (GTMs) [1] to automatically classify several disruption classes manually classified by the JET physicists [2]. Using GTMs high dimensional data reduces to a small enough number of dimensions and can be visualized in the low dimensional (usually 2-D) embedding space. The 2-D GTM of the 10-D operational space of JET has been obtained using an extensive database of discharges selected from JET campaigns from C15 up to C27 (year 2005-2009) performed with the carbon wall. The obtained map shows not only a natural discrimination between safe and disruptive regions, but also that the different classes of disruptions tend to occupy different regions in the map. Monitoring the evolution of each disruptive discharge on the GTM it has been found that many of them evolves within the same region with an associated class of disruption. The success rate of the classification is extremely high, sometimes reaching 100%. Using such GTM with discharges selected from more recent campaigns (C28-C30), when JET changes the carbon wall with the new ITER-like wall, significantly deteriorates the classification performance, particularly on certain disruption classes. Hence, a new GTM has been built to represent the new operational space of JET. Using this map with all the disruptions occurred in these last campaigns the very high classification performance is confirmed and therefore the prospects for the deployment of this tool in real time are very promising
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: http://hdl.handle.net/11584/108381
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo

Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
social impact