Current tokamaks make routinely use of few individual diagnostic signals, such as locked mode and vertical stabilization measurements, for triggering disruption mitigation systems. In ITER it will likely be necessary to employ a wider set of signals for disruption prediction. This is because presumably individual diagnostic signals will not be capable of detecting all disruptions type with enough warning time to allow effective mitigation actions. Nowadays, the interpretation of the physical mechanisms leading to disruptions in the existing experiments is the only viable way for extrapolating results to ITER.To this end,a comparison among similar disruption types and their corresponding precursors at JET and AUG would be valuable.Recently,criteria for manual disruption classification have been proposed both for JET [1]and AUG [2]. The analysis of disruption causes performed in [2] highlights that several physics instabilities in AUG are also typical disruption precursors in JET.In this paper,121 disruptive discharges occurred in AUG during the 2013-2014 experimental campaigns have been clustered.In particular,the chain of events characterizing an upcoming disruption has been identified and disruptions which follow similar path shave been categorized following the classification adopted for JET. A comparative multivariate statistical analysis of similar disruption types on JET and AUG has been performed in order to provide information on how the mechanisms leading to a disruption type can be generalized. Then, for each disruption class,the most significant precursors have been identified on both machines. Moreover, multisensor data fusion strategies will be applied with the aim of developing a real time detection system based on the complementary information of several diagnostic signals.
A multivariate analysis of disruption precursors on JET and AUG
Murari A;
2015
Abstract
Current tokamaks make routinely use of few individual diagnostic signals, such as locked mode and vertical stabilization measurements, for triggering disruption mitigation systems. In ITER it will likely be necessary to employ a wider set of signals for disruption prediction. This is because presumably individual diagnostic signals will not be capable of detecting all disruptions type with enough warning time to allow effective mitigation actions. Nowadays, the interpretation of the physical mechanisms leading to disruptions in the existing experiments is the only viable way for extrapolating results to ITER.To this end,a comparison among similar disruption types and their corresponding precursors at JET and AUG would be valuable.Recently,criteria for manual disruption classification have been proposed both for JET [1]and AUG [2]. The analysis of disruption causes performed in [2] highlights that several physics instabilities in AUG are also typical disruption precursors in JET.In this paper,121 disruptive discharges occurred in AUG during the 2013-2014 experimental campaigns have been clustered.In particular,the chain of events characterizing an upcoming disruption has been identified and disruptions which follow similar path shave been categorized following the classification adopted for JET. A comparative multivariate statistical analysis of similar disruption types on JET and AUG has been performed in order to provide information on how the mechanisms leading to a disruption type can be generalized. Then, for each disruption class,the most significant precursors have been identified on both machines. Moreover, multisensor data fusion strategies will be applied with the aim of developing a real time detection system based on the complementary information of several diagnostic signals.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
