真空抽气系统是托卡马克装置中不可缺少的核心子系统之一，真空抽气系统直接影响到聚变装置的运行安全，而分子泵是真空抽气系统中的关键设备。传统的分子泵维护由于缺少对设备故障的诊断及预测，可能出现由于分子泵故障导致托卡马克装置发生严重安全事故。本项目依托EAST托卡马克装置实验平台，面向未来聚变装置建设，采用先进的机器学习算法为基础，通过对托卡马克装置中分子泵关键运行数据的分析，研究以数据为驱动的分子泵故障诊断及预测系统的模型构建，故障诊断及预测系统的实施架构，关键技术与系统实现方法，并通过多种语言混合编程，将其集成到现有的EAST真空控制系统中，从而实现对分子泵故障的诊断与预测。我们希望通过本项目的开展，探索并构建托卡马克装置中分子泵故障诊断与预测系统，从而实现对关键设备故障的提前预判。为托卡马克装置中分子泵的维护，建立智能化预防性的主动维护机制，进而提高真空抽气系统和托卡马克装置的安全性。

Vacuum pumping system is one of the indispensable core subsystem of Tokamak device, and molecular pump is the key equipment of vacuum pumping system, directly affecting the running safety of the Tokamak device. Due to the lack of diagnosis and prediction system, the molecular pump fault may even lead to a serious safety accident in the Tokamak device in the traditional molecular pump maintenance. This project relies on the experimental platform of EAST device and future fusion device construction, adopts advanced machine learning algorithm, through the analysis of the key operation data of molecular pump in the Tokamak device, this project investigates the model construction, architecture, key technologies and system implementation method of the molecular pump fault diagnosis and prediction system, and integrate the system into the EAST vacuum control system via mixed language programming, in order to realize the diagnosis and prediction for molecular pump fault. Through this project, we hope to explore and build a molecular pump fault diagnosis and prediction system in Tokamak device, thus realize the prediction for the key equipment fault. To building an intelligent and preventive maintenance mechanism for the maintenance of molecular pumps in Tokamak device, improve the safety of the vacuum pump system and Tokamak device.