Overview of machine learning applications in fusion plasma experiments on J-TEXT tokamak

Machine learning research and applications in fusion plasma experiments are one of the main subjects on J-TEXT. Since 2013, various kinds of traditional machine learning, as well as deep learning methods have been applied to fusion plasma experiments. Further applications in the real-time experimental environment have proved the feasibility and effectiveness of the methods. For disruption prediction, we started by predicting disruptions of limited classes with a short warning time that could not meet the requirements of the mitigation system. After years of study, nowadays disruption prediction methods on J-TEXT are able to predict all kinds of disruptions with a high success rate and long enough warning time. Furthermore, cross-device disruption prediction methods have obtained promising results. Interpretable analysis of the models are studied. For diagnostics data processing, efforts have been made to reduce manual work in processing and to increase the robustness of the diagnostic system. Models based on both traditional machine learning and deep learning have been applied to real-time experimental environments. The models have been cooperating with the plasma control system and other systems, to make joint decisions to further support the experiments.     

Implementation of Automatic Process of Edge Rotation Diagnostic System on J-TEXT Tokamak

A spectral diagnostic control system （SDCS） is developed to implement automatic process of the edge rotation diagnostic system on the J-TEXT tokamak. The SDCS contains a control module, data operation module, data analysis module, and data upload module. The core of this system is a newly developed software ＂Spectra Assist＂, which completes the whole process by coupling all related subroutines and servers. The results of data correction and calculated rotation are presented. In the daily discharge of J-TEXT, SDCS is proved to have a stable performance and high efficiency in completing the process of data acquisition, operation and results output.     

A Visible Light Imaging System for the Estimation of Plasma Vertical Displacement in J-TEXT

A wide-viewing-angle visible light imaging system (VLIS) was mounted on the Joint Texas Experimental Tokamak (J-TEXT) to monitor the discharge process. It is proposed that by using the film data recorded the plasma vertical displacement can be estimated. In this paper installation and operation of the VLIS are presented in detailed. The estimated result is further compared with that measured by using an array of magnetic pickup coils. Their consistency verifies that the estimation of the plasma vertical displacement in J-TEXT by using the imaging data is promising.     

Design of the Control System of Toroidal Field （TF） Power Supply for J-TEXT Tokamak

J-TEXT, formerly TEXT-U at the University of Texas at Austin in USA, is a medium-sized tokamak at the Huazhong University of Science and Technology. The toroidal field （TF） power supply of this tokamak should provide a current of 160 kA and a flat duration of 500 ms for the toroidal field coils to generate a maximum toroidal field of 3 T at the geometric center of the vaccum vessel. This paper presents a design of a new control system which takes the real-time feedback control option for the TF power supply operation. The system was tested successfully during the commissioning. In the first experimental campaign of J-TEXT, the system effectively controlled the power supply to provide a fiat current up to 92.5 kA and therein the TF produced reached 1.74 T, which enabled the machine to generate the first plasma successfully.     

Development of the pellet injection system on the J-TEXT tokamak

Pellet injection is an attractive technology for core-fueling and magnetohydrodynamic study in magnetic-confinement fusion devices like tokamaks and stellarators. It can inject solid hydrogen/deuterium pellets into the plasma with deeper density deposition compared with other fueling methods, such as gas puffing. A three-barrel H₂ pellet injection system was installed on the J-TEXT tokamak and experiments were carried out. The pellets are formed in three barrels cooled by a cryocooler and compressor system at around 9 K, and are 0.8 mm/1 mm diameter and 0.8 mm length. The pellet is launched by helium propellant gas and injected from the low-field side of the plasma. The normal range of pellet speed is 210–310 m s⁻¹ for different propellant gas pressures. Due to the three-barrel structure, the number of injected pellets can be adjusted between one and three. Pellets can be launched sequentially with arbitrary time intervals, which enables flexible applications. The results of the experiments show that pellet fueling efficiency can reach 50%. The energy confinement time increased by about 7.5‒10 ms after pellet injection.     

The anode power supply for the ECRH system on the J-TEXT tokamak

The electron cyclotron resonance heating(ECRH) system with a 60 GHz/200 k W/0.5 s gyrotron donated by the Culham Science Center is being developed on the J-TEXT tokamak for plasma heating, current drive and MHD studies. Simultaneously, an anode power supply(APS) has been rebuilt and tested for the output power control of the gyrotron, of which the input voltage is derived from an 80 k V negative cathode power supply. The control strategy by controlling the grid voltage of the tetrode TH5186 is applied to obtain an accurate anode climbing voltage, of which the output voltage can be obtained from 0-30 k V with respect to the cathode power supply. The characteristics of the APS, including control, protection, modulation, and output waveform, were tested with a100 k V/60 A negative cathode power supply, a dummy load and the ECRH control system. The results indicate that the APS can meet the requirements of the ECRH system on J-TEXT.     

The timing system on the J-TEXT tokamak

This paper describes the timing system designed to control the operation time-sequence and to generate clocks for various sub-systems on J-TEXT tokamak. The J-TEXT timing system is organized as a distributed system which is connected by a tree-structured optical fiber network. It can generate delayed triggers and gate signals (0 μs–4000 s), while providing reference clocks for other sub-systems. Besides, it provides event handling and timestamping functions. It is integrated into the J-TEXT Control, Data Access and Communication (J-TEXT CODAC) system, and it can be monitored and configured by Experimental Physics and Industrial Control System (EPICS). The configuration of this system including tree-structured network is managed in XML files by dedicated management software. This system has already been deployed on J-TEXT tokamak and it is serving J-TEXT in daily experiments.     

Design and realization of the J-TEXT tokamak central control system

The Joint Texas Experimental Tokamak (J-TEXT), a medium-sized conventional tokamak, serves as a user experimental facility in the China-USA fusion research community. Development of a flexible and easy-to-use J-TEXT central control system (CCS) is of supreme importance for users to coordinate the experimental scenarios with full integration into the discharge operation. This paper describes in detail the structure and functions of the J-TEXT CCS system as well as the performance in practical implementation. Results obtained from both commissioning and routine operations show that the J-TEXT CCS system can offer a satisfactory and effective control that is reliable and stable. The J-TEXT tokamak achieved high-quality performance in its first-ever experimental campaign with this CCS system.     

Quasi-3D electron cyclotron emission imaging on J-TEXT

Electron cyclotron emission imaging(ECEI) can provide measurements of 2D electron temperature fluctuation with high temporal and spatial resolution in magnetic fusion plasma devices. Two ECEI systems located in different toroidal ports with 67.5 degree separation have been implemented on J-TEXT to study the 3D structure of magnetohydrodynamic(MHD) instabilities. Each system consists of 12(vertical) × 16(horizontal) = 192 channels and the image of the 2nd harmonic X-mode electron cyclotron emission can be captured continuously in the core plasma region. The field curvature adjustment lens concept is developed to control the imaging plane for receiving optics of the ECEI systems. Field curvature of the image can be controlled to match the emission layer. Consequently, a quasi-3D image of the MHD instability in the core of the plasma has been achieved.     

Introduction to resonant magnetic perturbation coils of the J-TEXT Tokamak

To investigate the interactions between both the static and rotating resonant magnetic perturbations (RMP) and the tokamak plasma, two sets of coils, namely static RMP (SRMP) and dynamic RMP (DRMP), are constructed on the J-TEXT tokamak. SRMP is reconstructed from TEXT-U and mainly produces static m/n = 1/1, 2/1 and 3/1 resonant perturbation field, where m and n are the poloidal and toroidal mode numbers, respectively. DRMP, newly designed and installed inside the vacuum vessel, can generate pure 2/1 RMP. DRMP is also designed to operate in the AC mode and can produce rotating 2/1 RMP which will be used to study the tearing mode control. Due to the effect of the eddy current in the vacuum vessel wall, the amplitudes of the 2/1 component will be attenuated to about 1/3.6 of the DC value when the operation frequency is larger than 500 Hz. However, DRMP can still provide sufficient large rotating 2/1 perturbation for tearing mode related studies.     

强子探测器快速读出系统的数据压缩和并行处理

在强子探测器快速读出系统中,由于采用了一级预相符触发的方法,达到了对数据流消零(Zero Suppression)的目的,从而大大降低了数据流量;由于采用了流水线结构(PipelineArchitecture)和高速查询表(Look-up Table)相结合的并行处理方法,二级触发器可处理数据率高达16MHz以上的数据。     

Design and first result of combined Langmuir-magnetic probe on J-TEXT tokamak

In order to measure boundary electrostatic and magnetic fluctuations simultaneously, a combined Langmuir-magnetic probe (CLMP) has been designed and built on joint-Texas experimental tokamak. The probe consists of 8 graphite probe pins and a 3D magnetic probe, driven by a mechanical pneumatic device. By means of simulation, the shielding effect of the graphite sleeve on the magnetic fluctuation signal is explored, and the influence of the eddy current was reduced by cutting the graphite sleeve. In the experiment, it has been verified that the mutual inductance of electromagnetic signals can be ignored, and a 70–90 kHz electromagnetic mode is observed around the last closed magnetic surface. The establishment of CLMP provides data for the exploration of the coupling of electrostatic and magnetic fluctuations.     

反应堆中子活化分析数据自动处理系统

介绍了一个在微机上建立的反应堆中子活化分析数据自动处理系统。它具有放射性核素的自动识别、反应堆参数ｆ和α值的计算、实验测定Ｋ０值、元素含量的定量计算、核素数据管理、参数选取建议等功能。元素分析方法包括绝对法、相对法和Ｋ０值法。利用该系统进行了金、锆和岩石标准物质（ＧＢＷ０７１０７）的活化分析,计算了中国原子能科学研究院微型反应堆的ｆ和α值参数,分析了岩石样品中的元素含量,取得了较满意的结果。     

Machine learning application to predict the electron temperature on the J-TEXT tokamak

The reliability of diagnostic systems in tokamak plasma is of great significance for physics researches or fusion reactor. When some diagnostics fail to detect information about the plasma status, such as electron temperature, they can also be obtained by another method: fitted by other diagnostic signals through machine learning. The paper herein is based on a machine learning method to predict electron temperature, in case the diagnostic systems fail to detect plasma temperature. The fully-connected neural network, utilizing back propagation with two hidden layers, is utilized to estimate plasma electron temperature approximately on the J-TEXT. The input parameters consist of soft x-ray emission intensity, electron density, plasma current, loop voltage, and toroidal magnetic field, while the targets are signals of electron temperature from electron cyclotron emission and x-ray imaging crystal spectrometer. Therefore, the temperature profile is reconstructed by other diagnostic signals, and the average errors are within 5%. In addition, generalized regression neural network can also achieve this function to estimate the temperature profile with similar accuracy. Predicting electron temperature by neural network reveals that machine learning can be used as backup means for plasma information so as to enhance the reliability of diagnostics.     

New Current Control Method of DC Power Supply for Magnetic Perturbation Coils on J-TEXT

In order to advance the research on suppressing tearing modes and driving plasma rotation, a DC power supply （PS） system has been developed for dynamic resonant magnetic perturbation （DRMP） coils and applied in the J-TEXT experiment. To enrich experimental phe- nomena in the J-TEXT tokamak, applying the circulating current four-quadrant operation mode in the DRMP DC PS system is proposed. By using the circulating current four-quadrant oper- ation, DRMP coils can be smoothly controlled without the dead-time when the current polarity reverses. Essential circuit analysis, control optimization and simulation of desired scenarios have been performed for normal current. Relevant simulation and test results are also presented.     

Design of the high voltage isolation transmission module with low delay for ECRH system on J-TEXT

As a flexible auxiliary heating method,the electron cyclotron resonance heating(ECRH) has been widely used in many tokamaks and also will be applied for the J-TEXT tokamak.To meet requirements of protection and fault analysis for the ECRH system on J-TEXT,signals of gyrotrons such as the cathode voltage and current,the anode voltage and current,etc should be transmitted to the control and data acquisition system.Considering the high voltage environment of gyrotrons,isolation transmission module based on FPGA and optical fiber communication has been designed and tested.The test results indicate that the designed module has strong anti-noise ability,low error rate and high transmission speed.The delay of the module is no more than 5 μs which can fulfill the requirements.     

Impact of Nuclear Data Uncertainties on Innovative Fast Reactors and Required Target Accuracies

It is widely accepted that the current status of neutronics calculations for fast reactor design is such that the present uncertainties on nuclear data should still be significantly reduced, in order to get the full benefit from advances in modeling and simulation. Only a parallel effort in advanced simulation, high-accuracy validation experiments, and nuclear data improvement will provide designers with more general and wellvalidated calculation tools to meet tight design target accuracies to further improve safety and economics. The present paper presents very recent results related to nuclear data uncertainty impact assessment and target accuracy requirements for advanced reactor systems.     

Overview of the J-TEXT progress on RMP and disruption physics

The J-TEXT tokamak has been operated for ten years since its first plasma obtained at the end of 2007. The diagnostics development and main modulation systems, i.e. resonant magnetic perturbation (RMP) systems and massive gas injection (MGI) systems, will be introduced in this paper. Supported by these efforts, J-TEXT has contributed to research on several topics, especially on RMP physics and disruption mitigation. Both experimental and theoretical research show that RMP could lock, suppress or excite the tearing modes, depending on the RMP amplitude, frequency difference between RMP and rational surface rotation, and initial stabilities. The plasma rotation, particle transport and operation region are influenced by the RMP. Utilizing the MGI valves, disruptions have been mitigated with pure He, pure Ne, and a mixture of He and Ar (9:1). A significant runaway current plateau could be generated with moderate amounts of Ar injection. The RMP has been shown to suppress the generation of runaway current during disruptions.     

Effect of edge magnetic island on carbon screening in the J-TEXT tokamak

The effect of externally applied resonant magnetic perturbation(RMP)on carbon impurity behavior is investigated in the J-TEXT tokamak.It is found that the m/n=3/1 islands have an impurity screening effect,which becomes obvious while the edge magnetic island is generated via RMP field penetration.The impurity screening effect shows a dependence on the RMP phase with the field penetration,which is strongest if the O point of the magnetic island is near the low-field-side(LFS)limiter plate.By combining a methane injection experimental study and STRAHL impurity transport analysis,we found that the variation of the impurity transport dominates the impurity screening effect.The impurity diffusion at the inner plasma region(r/a＜0.8)is enhanced with a significant increase in outward convection velocity at the edge region in the case of the magnetic island's O point near the LFS limiter plate.The impurity transport coefficient varies by a much lower level for the case with the magnetic island's X point near the LFS limiter plate.The interaction of the magnetic island and the LFS limiter plate is thought to contribute to the impurity transport variation with the dependence on the RMP phase.A possible reason is the interaction between the magnetic island and the LFS limiter.