Clean Energy Conversion Research Section
概要
In Heliotron-J, a large amount of time series data has been obtained by the progress of simultaneous multi-point plasma measurements for many years. It is necessary to coarse-grain and visualize this large amount of data to extract deep correlation and hidden causarity. In order to coarse-grain the data under the assumption that "physics is simple", it is necessary to define some kind of simplified model and to quantify and minimize the statistical distance between the observation and the model [1]. Following this approach, we focus on time series data in particular, and extract some temporal patterns inherent in the data. This method can represent the non-stationarity, intermittence, and non-linearity of the plasma with fewer degrees of freedom than the Fourier analysis, which has been the main tool for time series data analysis so far. In this study, we develop such a new analysis tool using Heliotron-J time series data, and discuss the non-stationarity, suddenness, and non- linearity of Heliotron-J plasma. Heliotron-J has accumulated a large amount of time series data as described above, and in addition, it has large-scale data such as digital ECE using ultrafast oscilloscopes, which has been conducted in previous collaborative research. There are few projects in Japan where such data sets are available. Therefore, it is necessary to conduct this project as a collaborative research at Heliotron J.