One of the major problems in present tokamaks is the presence of disruptions. If disruptions are not mitigated, they can produce serious damage to th e device. Therefore, disruption predictors are needed in order to apply the mitigation techniques in time. In this paper, the real - time implementation in JET of a new type of disruption predictor is presented. The new predictor, Single signal Predictor bas ed on Anomaly Detection (SPAD), does not require past discharges for training purposes. The implementation is based on the Multi - threaded Application Real - Time executor (MARTe) framework. Analysis over all JET's ITER - like Wall campaigns (C28 - C34) show that SPAD was able to predict 83.57% of the disruptions with enough time to apply mitigation techniques. The average anticipation time was 389 ms. In this paper the real - time implementation will be discussed, as well as the optimizations developed to make the algorithm suitable for real - time processing. Performance results and possible improvements will also be analyzed.
Real-time implementation in JET of the SPAD disruption predictor using MARTe
Murari A;
2016
Abstract
One of the major problems in present tokamaks is the presence of disruptions. If disruptions are not mitigated, they can produce serious damage to th e device. Therefore, disruption predictors are needed in order to apply the mitigation techniques in time. In this paper, the real - time implementation in JET of a new type of disruption predictor is presented. The new predictor, Single signal Predictor bas ed on Anomaly Detection (SPAD), does not require past discharges for training purposes. The implementation is based on the Multi - threaded Application Real - Time executor (MARTe) framework. Analysis over all JET's ITER - like Wall campaigns (C28 - C34) show that SPAD was able to predict 83.57% of the disruptions with enough time to apply mitigation techniques. The average anticipation time was 389 ms. In this paper the real - time implementation will be discussed, as well as the optimizations developed to make the algorithm suitable for real - time processing. Performance results and possible improvements will also be analyzed.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.