智能控制在SGTR规程中的应用研究

Investigation on Application of Intelligent Control in SGTR Procedure

核电厂发生蒸汽发生器传热管破裂(SGTR)事故后,操纵员需按相应规程进行操作。但在事故序列处理中,一回路冷却剂的降温、降压效果大多依靠操纵员的运行经验。针对该问题,本文提出一种基于长短期记忆神经网络(LSTM)和自适应比例-积分-微分(PID)的SGTR规程优化方法,并给出了实现流程。利用某1000 MW模拟机开展了SGTR事故模拟实验,基于模拟实验数据对本文方法进行了测试。仿真结果表明,与操纵员人工操作相比,本文提出的智能操作方法具有更好的降温、降压效果,本文方法可为核电厂事故规程的智能化操作提供一种新思路。

After the occurrence of steam generator heat transfer tube rupture(SGTR) accident in nuclear power plant,operators shall follow the corresponding procedures.However,in accident sequence processing,the cooling and depressurizing effects of primary circuit coolant mostly depend on the operating experience of the operator.To solve this problem,this paper proposes a SGTR operating procedure optimization method based on LSTM and adaptive proportional-integraldifferential(PID),and gives the implementation process.The SGTR accident simulation experiment was carried out by a 1000 MW simulator.Based on the simulation experiment data,the proposed method was tested.Simulation results show that,compared with manual operation,the proposed intelligent operation method has better cooling and depressurization effects.The proposed method can provide a new idea for intelligent operation of accident procedures in nuclear power plants.