Disruption prediction has made rapid progress in recent years, especially in machine learning (ML)-based methods. If a disruption prediction model can be interpreted, it can tell why certain samples are classified as disruption precursors. This allows us to tell the types of incoming disruption for disruption avoidance and gives us insight into the mechanism of disruption. This paper presents a disruption predictor called Interpretable Disruption Predictor based on Physics-Guided Feature Extraction (IDP-PGFE) and its results on J-TEXT experiment data. The prediction performance of IDP-PGFE with physics-guided features is effectively improved (TPR = 97.27%, FPR = 5.45%, AUC = 0.98) compared to the models with raw signal input. The validity of the interpretation results is ensured by the high performance of the model. The interpretability study using an attribution technique provides an understanding of J-TEXT disruption and conforms to our prior comprehension of disruption. Furthermore, IDP-PGFE gives a possible means of inferring the underlying cause of the disruption and how interventions affect the disruption process in J-TEXT. The interpretation results and the experimental phenomenon have a high degree of conformity. The interpretation results also give a possible experimental analysis direction that the RMPs delay the density limit disruption by affecting both the MHD instabilities and the radiation profile. PGFE could also reduce the data requirement of IDP-PGFE to 10% of the training data required to train a model on raw signals. This made it possible to be transferred to the next-generation tokamaks, which cannot provide large amounts of data. Therefore, IDP-PGFE is an effective approach to exploring disruption mechanisms and transferring disruption prediction models to future tokamaks.

Tokamaks are the most promising way for nuclear fusion reactors. Disruption in tokamaks is a violent event that terminates a confined plasma and causes unacceptable damage to the device. Machine learning models have been widely used to predict incoming disruptions. However, future reactors, with much higher stored energy, cannot provide enough unmitigated disruption data at high performance to train the predictor before damaging themselves. Here we apply a deep parameter-based transfer learning method in disruption prediction. We train a model on the J-TEXT tokamak and transfer it, with only 20 discharges, to EAST, which has a large difference in size, operation regime, and configuration with respect to J-TEXT. Results demonstrate that the transfer learning method reaches a similar performance to the model trained directly with EAST using about 1900 discharge. Our results suggest that the proposed method can tackle the challenge in predicting disruptions for future tokamaks like ITER with knowledge learned from existing tokamaks.

China has been suffering from water shortage for a long time. Weather modification and rainfall enhancement via cloud seeding has been proved to be effective to alleviate the problem. Current cloud seeding methods mostly rely on solid carbon dioxide and chemicals such as silver iodide and hygroscopic salts, which may have negative impacts on the environment and are expensive to operate. Lab experiments have proved the efficiency of ion-based cloud seeding compared with traditional methods. Moreover, it is also more environmentally friendly and more economical to operate at a large scale. Thus, it is necessary to carry out a field experiment to further investigate the characteristics and feasibility of the method. This paper provides the design and implementation of the ion-based cloud seeding and rain enhancement trial currently running in Northwest China. It introduces the basic principle of the trial and the devices developed for it, as well as the installation of the bases and the evaluation method design for the trial.

Machine learning research and applications in fusion plasma experiments is 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 with predicting disruptions of limited classes with a short warning time that was not able to meet the requirements of the mitigation system. With years of study, nowadays disruption prediction methods on J-TEXT are able to predict all kinds of disruptions on J-TEXT with a high success rate and long enough warning time. Furthermore, cross-device disruption prediction methods and interpretable models are studied and have obtained promising results. For diagnostics data processing, efforts have been made to reduce manual work in processing and to increase the robustness of the diagnostic system. Both traditional machine learning and deep learning-based models have been applied to the real-time experimental environment and have been cooperating with plasma control and other systems to make joint decisions to further support the experiments.

Background Acute lung injury (ALI) is a severe inflammatory disease, underscoring the urgent need for novel treatments. Nauclea officinalis Pierre ex Pitard (Danmu in Chinese, DM) is effective in treating inflammatory respiratory diseases. However, there is still no evidence of its protective effect against ALI. Methods Metabolomics was applied to identify the potential biomarkers and pathways in ALI treated with DM. Further, network pharmacology was introduced to predict the key targets of DM against ALI. Then, the potential pathways and key targets were further verified by immunohistochemistry and western blot assays. Results DM significantly improved lung histopathological characteristics and inflammatory response in LPS-induced ALI. Metabolomics analysis showed that 16 and 19 differential metabolites were identified in plasma and lung tissue, respectively, and most of these metabolites tended to recover after DM treatment. Network pharmacology analysis revealed that the PI3K/Akt pathway may be the main signaling pathway of DM against ALI. The integrated analysis of metabolomics and network pharmacology identified 10 key genes. These genes are closely related to inflammatory response and cell apoptosis of lipopolysaccharide (LPS)-induced ALI in mice. Furthermore, immunohistochemistry and western blot verified that DM could regulate inflammatory response and cell apoptosis by affecting the PI3K/Akt pathway, and expression changes in Bax and Bcl-2 were also triggered. Conclusion This study first integrated metabolomics, network pharmacology and biological verification to investigate the potential mechanism of DM in treating ALI, which is related to the regulation of inflammatory response and cell apoptosis. And the integrated analysis can provide new strategies and ideas for the study of traditional Chinese medicines in the treatment of ALI. Graphical Abstract