Manifold Learning to Interpret JET High-Dimensional Operational Space

In this paper, the problem of visualization and exploration of JET high-dimensional operational space is considered. The data comes from plasma discharges selected from JET campaigns from C15 (year 2005) and up to C27 (year 2009). The aim is to learn the possible manifold structure embedded in the data, to create some representations of the plasma parameters on low-dimensional maps, which are understandable and which preserve the essential properties owned by the original data. A crucial issue for the design of such mappings is the quality of the data set. This paper reports the details of the criteria used to properly select the more suitable signals downloaded from JET data bases, and the algorithms to pre-process the diagnostic signals in order to obtain a data set of real­time reliable observations. Moreover, a statistical analysis is performed in order to recognize the presence of outliers. Finally, data reduction, based on clustering methods, is performed to select a limited and representative number of samples for the operational space mapping. Among the large number of manifold learning methods, in this paper three have been investigated: two linear algorithms (Principal Component Analysis and Grand Tour) and one non linear (Self Organizing Maps). The obtained maps can be used to identify characteristic regions of the plasma scenario, allowing to discriminate between the regions with high risk of disruption and those with low risk of disruption.
Name Size  
EFDP12011 1.12 Mb