The electron cyclotron emission(ECE)diagnostic system has been developed on the ENN spherical torus(EXL-50).The ECE system is designed to detect radiation emitted by energetic electrons,rather than conventional 1D ele...The electron cyclotron emission(ECE)diagnostic system has been developed on the ENN spherical torus(EXL-50).The ECE system is designed to detect radiation emitted by energetic electrons,rather than conventional 1D electron temperature profile measurement,in the frequency range of 4-40 GHz.The system is composed of five subsystems,each covering a different frequency band,including the C-band(4-8 GHz),X-band(8-12 GHz),Ku-band(12-18 GHz),K-band(18-26.5 GHz)and Kα-band(26.4-40 GHz).The system uses heterodyne detection to analyze the received signals.The K-band and Kα-band subsystems are located horizontally in the equatorial plane of the EXL-50,while the C-band,X-band and Ku-band subsystems are located under the vacuum vessel of the EXL-50.To direct the microwaves from the plasma to the antennas for the horizontal detection subsystems,a quasi-optical system has been developed.For the vertical detection subsystems,the antennas are directly attached to the port located beneath the torus at R=700 mm,which is also the magnetic axis of the torus.The system integration,bench testing and initial experimental results will be thoroughly discussed,providing a comprehensive understanding of the ECE system s performance and capabilities.展开更多
A neural network model with a classical annotation method has been used on the EXL-50tokamak to predict impending disruption.However,the results revealed issues of overfitting and overconfidence in predictions caused ...A neural network model with a classical annotation method has been used on the EXL-50tokamak to predict impending disruption.However,the results revealed issues of overfitting and overconfidence in predictions caused by inaccurate labeling.To mitigate these issues,an improved training framework has been proposed.In this approach,soft labels from previous training serve as teachers to supervise the further learning process;this has lead to a significant improvement in predictive model performance.Notably,this enhancement is primarily attributed to the coupling effect of the soft labels and correction mechanism.This improved training framework introduces an instance-specific label smoothing method,which reflects a more nuanced model assessment on the likelihood of a disruption.It presents a possible solution to effectively address the challenges associated with accurate labeling across different machines.展开更多
基金performed under the auspices of National Natural Science Foundation of China(No.11605244)supported by the High-End Talents Program of Hebei Province,Innovative Approaches towards Development of CarbonFree Clean Fusion Energy(No.2021HBQZYCSB006)。
文摘The electron cyclotron emission(ECE)diagnostic system has been developed on the ENN spherical torus(EXL-50).The ECE system is designed to detect radiation emitted by energetic electrons,rather than conventional 1D electron temperature profile measurement,in the frequency range of 4-40 GHz.The system is composed of five subsystems,each covering a different frequency band,including the C-band(4-8 GHz),X-band(8-12 GHz),Ku-band(12-18 GHz),K-band(18-26.5 GHz)and Kα-band(26.4-40 GHz).The system uses heterodyne detection to analyze the received signals.The K-band and Kα-band subsystems are located horizontally in the equatorial plane of the EXL-50,while the C-band,X-band and Ku-band subsystems are located under the vacuum vessel of the EXL-50.To direct the microwaves from the plasma to the antennas for the horizontal detection subsystems,a quasi-optical system has been developed.For the vertical detection subsystems,the antennas are directly attached to the port located beneath the torus at R=700 mm,which is also the magnetic axis of the torus.The system integration,bench testing and initial experimental results will be thoroughly discussed,providing a comprehensive understanding of the ECE system s performance and capabilities.
基金supported by National Natural Science Foundation of China(Nos.12175277 and 11975271)the National Key R&D Program of China(No.2022YFE 03050003)。
文摘A neural network model with a classical annotation method has been used on the EXL-50tokamak to predict impending disruption.However,the results revealed issues of overfitting and overconfidence in predictions caused by inaccurate labeling.To mitigate these issues,an improved training framework has been proposed.In this approach,soft labels from previous training serve as teachers to supervise the further learning process;this has lead to a significant improvement in predictive model performance.Notably,this enhancement is primarily attributed to the coupling effect of the soft labels and correction mechanism.This improved training framework introduces an instance-specific label smoothing method,which reflects a more nuanced model assessment on the likelihood of a disruption.It presents a possible solution to effectively address the challenges associated with accurate labeling across different machines.
