Research on Monitoring Technology for the Operation Status of Nuclear Safety Equipment in Nuclear Power Plants
沈江飞①② SHEN Jiang-fei；张圣①②ZHANG Sheng；喻昕① YU Xin；王保军① WANG Bao-jun
（①苏州热工研究院有限公司，苏州 215004；②国家核电厂安全及可靠性工程技术研究中心，苏州 215004）
（①Suzhou Nuclear Power Research Institute Co.，Ltd.，Suzhou 215004，China；②National Engineering Research Center for Nuclear Power Plant Safety & Reliability，Suzhou 215004，China）
摘要：核电厂安全相关设备的状态可靠是电厂安全保障的核心要素，然而绝大部分核安全设备都是设计在发生基准事故期间或之后用来确保反应堆安全的，日常生产中并不运行，只能通过定期试验验证设备可用性。本文提出一种基于多元状态估计技术的核安全设备状态智能监测方法，该方法通过建立状态监测模型，可以在设备启停试验过程中实时预测参数并判断设备运行状态，预警设备的早期异常。该方法首先采集设备历史试验过程中的状态优良数据，训练监测模型来挖掘数据的关联特征；然后，在新的定期试验过程中实时采集设备运行数据，并基于训练好的模型推断各参数的实时预测值，当实测值与预测值产生过大偏差时发出预警信息。本文以某电厂某型号核安全水泵为例进行建模并验证，结果表明本文提出的监测模型可以模拟核安全设备在定期试验过程中的数据变化特性，并且可以有效监测和预警核安全设备在定期试验过程中的早期异常。
Abstract: The reliability of the nuclear safety equipment is a core element of nuclear power plants safety assurance. However, most of nuclear safety equipment are designed to ensure the safety of the reactor during or after the occurrence of accidents, do not work during daily production, and can only verify availability through regular testing. This article proposed an intelligent monitoring method for nuclear safety equipment status based on multi-state estimation technique. This method can predict parameter and judge equipment operation status in real time during equipment start-stop testing by establishing monitoring models, warning early abnormalities of the equipment. The model first collects high-quality data from the equipment's historical testing process, and trains a monitoring model to mine the associated features of the data. Then, during the new regular testing process, real-time equipment operation data is collected, and the trained model is used to infer real-time predicted values for various parameters. When there is a significant deviation between the measured values and the predicted values, warning information is issued. This article uses a certain type of nuclear safety pump in a power plant as an example for modeling and verification. The results show that the monitoring model proposed in this article can simulate the data variation characteristics of nuclear safety equipment and can effectively monitor and warning early abnormalities during regular testing.
关键词：核安全设备；数据模型；核电厂；状态监测
Key words: nuclear safety equipment；data-driven model；nuclear power station；on-line monitoring
中图分类号：TM623.9 文献标识码：A 文章编号：1006-4311（2024）28-014-05 doi:10.3969/j.issn.1006-4311.2024.28.005
|
