English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Tools annual

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

In this paper, we present a new deep learning disruption prediction algorithmbased on important findings from explorative data analysis which effectivelyallows knowledge transfer from existing devices to new ones, thereby predictingdisruptions using very limited disruptive data from the new devices. Theexplorative data analysis conducted via unsupervised clustering techniquesconfirms that time-sequence data are much better separators of disruptive andnon-disruptive behavior than the instantaneous plasma state data with furtheradvantageous implications for a sequence-based predictor. Based on suchimportant findings, we have designed a new algorithm for multi-machinedisruption prediction that achieves high predictive accuracy on the C-Mod(AUC=0.801), DIII-D (AUC=0.947) and EAST (AUC=0.973) tokamaks with limitedhyperparameter tuning. Through numerical experiments, we show that boostedaccuracy (AUC=0.959) is achieved on EAST predictions by including in thetraining only 20 disruptive discharges, thousands of non-disruptive dischargesfrom EAST, and combining this with more than a thousand discharges from DIII-Dand C-Mod. The improvement of predictive ability obtained by combiningdisruptive data from other devices is found to be true for all permutations ofthe three devices. Furthermore, by comparing the predictive performance of eachindividual numerical experiment, we find that non-disruptive data aremachine-specific while disruptive data from multiple devices containdevice-independent knowledge that can be used to inform predictions fordisruptions occurring on a new device.

Hybrid deep learning architecture for general disruption prediction across tokamaks