学术报告第152期—熊寿齐 张智超 论文投稿报告

时间：2021年11月17日 周三14:30

地点：326会议室

论文投稿报告

报告人：熊寿齐 研究生

题目：Study of a 105GHz short-pulse dummy load for the electron cyclotron resonance heating system with the quasi-optical method

摘要：Power measurement is necessary for an electron cyclotron resonance heating (ECRH) system. The dummy load is one of the significant equipment for the millimeter wave power measurement. In this paper, the dummy load is analyzed based on the quasi-optical method and the ray tracing method. The reflectivity and thermal deposition of the dummy load has been considered to ensure the safety of the whole system. High-power tests have been carried out at a 105GHz/500kW ECRH system. The results of the tests indicate that the dummy load is stable and valid.

报告人：张智超 研究生

题目：Realization of automatic data cleaning and feedback conditioning for J-TEXT ECEI signals based on machine learning

摘要：In recent years, Electron Cyclotron Emission Imaging (ECEI) diagnostic and many other imaging diagnostics become more and more important in magnetic confinement fusion research. The image’s quality gets worse by the bad channels, it is a nonignord issue for imaging diagnostics. To automatically identify and classify the ECEI abnormal signals, the classifier of ECEI signals based on machine learning tools has been developed for the J-TEXT ECEI diagnostic. Based on the digital control function of J-TEXT ECEI, it can also correct the channels of supersaturated signals and weak signals by adjusting the attenuation levels. At present, the system has been set up, the overall accuracy rate of the classification algorithm on external test set can reach 93.8%, which can meet the requirement of signals identification. Feedback conditioning can be completed between two discharge shots. Such automatic data cleaning and feedback conditioning unit has been applied on the J-TEXT ECEI. It can preprocess the diagnostic data for physical analysis and improve the quality of ECEI signals.