A Novel Approach for Plasma Current Quench Studies Via Synthetic Data Generation

Abstract

Electromagnetic forces, thermal, and radiation loads experienced by the in-vessel components/vacuum vessel at the time of the tokamak plasma current quench (CQ) greatly affect the health of the overall plasma device. Thus the mitigation is of paramount importance, which requires a proper identification of the disruption precursors. Using new machine learning (ML) and artificial intelligence (AI) approaches, it is possible to identify disruption precursors, however, such approaches require training the ML/AI models. This training of models requires a huge amount of experimental data, which sometimes may not be sufficient due to variation in the current quench profile across different tokamak. This necessitates the need for accurate synthetic current quench data generation for different types of the CQ profiles observed experimentally. We propose a novel approach for synthetic CQ data generation, wherein the set of experimental current quench data is clustered into groups with similar patterns and each cluster, rather than the entire set, is employed as a base for generating more realistic and accurate synthetic data consisting of all important current quench shape aspects. The synthetically generated data is evaluated and categorized as disruption, soft landing, or step landing with multiple slopes. The disruption data set is further analyzed qualitatively for the CQ nature - exponential, linear, or Gaussian. The quantification of the minimum number of experimental data sets and it's time resolution for generating meaningful synthetic data are also explored. The typical current quench parameters namely average CQ rate <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$(QR_{90-10})$</tex> , instantaneous CQ <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$((dI_{p}/dt)$</tex> , CQ time ( <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\tau$</tex> ) have been estimated for the synthetic data, and subsequent comparison with the experimental parameter across different tokamak CQ has been found to be in a fair agreement.