Time-domain global similarity method for automatic data cleaning for multi-channel measurement systems in magnetic confinement fusion devices

摘要

To guarantee the availability and reliability of data source in MagneticConfinement Fusion (MCF) devices, incorrect diagnostic data, which cannotreflect real physical properties of measured objects, should be sorted outbefore further analysis and study. Traditional data sorting cannot meet thegrowing demand of MCF research because of the low-efficiency, time-delay, andlack of objective criteria. In this paper, a Time-Domain Global Similarity(TDGS) method based on machine learning technologies is proposed for theautomatic data cleaning of MCF devices. Traditional data sorting aims to theclassification of original diagnostic data sequences, which are different inboth length and evolution properties under various discharge parameters. Hencethe classification criteria are affected by many discharge parameters and varyshot by shot. The focus of TDGS method is turned to the physical similaritybetween data sequences from different channels, which are more essential andindependent of discharge parameters. The complexity arisen from real dischargeparameters during data cleaning is avoided in the TDGS method by transformingthe general data sorting problem into a binary classification problem about thephysical similarity between data sequences. As a demonstration of itsapplication to multi-channel measurement systems, the TDGS method is applied tothe EAST POlarimeter-INterferomeTer (POINT) system. The optimized performanceof the method has reached 0.9871.