Automatic recognition of plasma relevant events: Implications for ITER

Abstract

This work makes a proposal about the use of big data techniques for the automatic recognition and classification

of plasma relevant events in huge databases of nuclear fusion devices. A relevant event can be any kind of

anomaly (or perturbation) in the plasma evolution. This is revealed in the temporal evolution signals as (typically) abrupt variations (for instance in amplitude, noise, or sudden presence/suppression of patterns with

periodical structure). A general algorithm based on five steps is presented here for the automatic location and

unsupervised classification of plasma events: dataset selection, location of anomalies in individual signals, definition of multi-signal patterns, unsupervised clustering of multi-signal patterns and creation of supervised

classifiers. It is important to note that the algorithm implementation is for off-line analysis but supervised

classifiers could be implemented under real-time conditions.