López Navarro, Juan Manuel and Ruiz González, Mariano and Arcas Castro, Guillermo de and Dormido-Canto, Sebastián and Vega Sánchez, Jesús and Murari, Andrea and Ramirez Pérez, Jesús Manuel
Simulator of the JET real-time disruption predictor.
In: "7 th Workshop on Fusion Data Processing Validation and Analysis, March 27, 2012", 26/03/2012 - 28/03/2012, Frascati, Roma, Italia. pp. 1-20.
A disruption predictor based on support vector machines
(SVM) has been developed to be used in JET. The training
process uses thousands of discharges and, therefore, high performance computing has been necessary to obtain the models. To this respect, several models have been generated with data from different JET campaigns. In addition, various kernels (mainly linear and RBF) and parameters have been tested. The main objective of this work has been the
implementation of the predictor model under real-time constraints. A “C-code” software application has been developed to simulate the real-time behavior of the predictor. The application reads the signals from the JET
database and simulates the real-time data processing, in particular, the specific data hold method to be developed when reading data from the JET ATM real time network. The
simulator is fully configurable by means of text files to select models, signal thresholds, sampling rates, etc. Results with data between campaigns C23and C28 will be shown.