Combined retrieval of multiple discharge signal waveforms based on distributed architecture

Processing time-series data, such as discharge signal waveforms, is essential for conducting data analysis in tokamak discharge experiments. Typically, researchers manually identify and filter shots by observing waveforms, which can be a time-consuming task involving data preprocessing and classification. In addition, there is often no intuitive interface for complex retrieval of distinct characteristic waveforms. This study proposes a novel method for retrieving multiple discharge signal waveforms in tokamak discharge experimental data. This method leverages several waveform features, including the slope of waveform segments, drop distance of segment head (DDSH), and mean absolute error of segments (MAES). It also offers a strategy for optimizing waveform retrieval based on the waveform length and average DDSH value. The experimental results demonstrate an average reduction in waveform retrieval time of 39.9 %. Building upon this method, the study presents the design and implementation of a distributed discharge signal waveform retrieval platform within a visual environment. This platform has undergone preliminary deployment in the EAST tokamak for tasks such as data screening and abnormal waveform detection, highlighting its significant contribution to simplifying data analysis processes.