Research on Multi-parameter Synchronous Monitoring Technology of Nuclear Power Plant Equipment Status
-
摘要: 核电厂设备稳定的运行状态以及长期的运行数据积累为实现数据驱动的设备状态智能监测建立了良好的数据基础。本文提出一种基于多参数关联关系的设备状态智能监测方法,该方法包括建模、训练和推断3个步骤,建立数据驱动的设备状态智能监测和预警模型。首先识别并分析系统设备监测参数、参数监测内容和关联关系,设计建立监测参数的关联关系模型;其次,采集并筛选设备正常运行历史数据作为训练数据,基于反向传播(BP)前馈神经网络训练关联关系模型;最终,实时采集设备监测参数实测值,并基于模型推断各参数的预测值,监控实测值与预测值的偏差,当偏差超过预定的阀值时发出预警信息。本文以某电厂热交换器和主给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有效同步监测设备参数微小异常变化,预警早期异常,同时保持极低的误报警率。Abstract: The stable operation status and long-term operation data accumulation of nuclear power plant equipment have established a good data foundation for realizing data-driven intelligent monitoring of equipment status. In this paper, an intelligent monitoring method of equipment condition based on multi-parameter correlation is proposed, which includes three steps: modeling, training and inference, and a data-driven intelligent monitoring and early warning model of equipment condition is established. First, the system equipment monitoring parameters, parameter monitoring contents and correlation are identified and analyzed, and the correlation model of monitoring parameters is designed and established. Second, the historical data of normal operation of the equipment are collected and selected as training data, and the correlation model is trained based on BP feed forward neural network; Finally, the measured values of the monitoring parameters of the equipment are collected in real time, and the predicted values of each parameter are inferred based on the model, and the deviation between the measured value and the predicted value is monitored, and an early warning message is given when the deviation exceeds a predetermined threshold. This paper takes the heat exchanger and main feed pump of a power plant as an example to model and verify. The results show that the monitoring model proposed in this paper can effectively monitor the small and abnormal changes of equipment parameters synchronously, give early warning against early abnormalities, and maintain a very low false alarm rate.
-
Key words:
- Data model /
- On-line monitoring /
- Nuclear power plant /
- Neural network
-
图 4 APA201PO工作过程和参数识别
P11—泵入口水压1;P12—泵入口水压2;P13—泵入口水温;P14—泵转速;P15—润滑油油温(冷却后);P16—润滑油油压;P17—泵出口水压;P18—泵出口水流量1;P19—泵出口水流量2;P20—泵出口水温;P21—泵非驱动端水平方向振动位移幅值;P22—泵非驱动端垂直方向振动位移幅值;P23—泵驱动端水平方向振动位移幅值;P24—泵驱动端垂直方向振动位移幅值;P25—泵推力轴承外侧水平方向温度;P26—泵推力轴承外侧垂直方向温度;P27—泵推力轴承内侧水平方向温度;P28—泵推力轴承内侧垂直方向温度;P29—泵非驱动端径向轴承温度;P30—泵驱动端径向轴承温度;P31—润滑油油温(冷却前);P32—SRI冷却水温度;P33—泵非驱动端轴封温度;P34—泵驱动端轴封温度
Figure 4. Working Process and Parameter Identification of APA201PO
-
[1] 刘勤明. 基于状态监测信息的设备在线健康预测及维护优化研究[D]. 上海: 上海交通大学, 2014. [2] RUSAW R. Guideline for online monitoring of nuclear power plants: volume 2[R]. U.S.: EPRI, 2011. [3] 赵倩. 变电站设备监测系统及温度预测算法的研究[D]. 保定: 华北电力大学, 2015. [4] 李景. 基于神经网络的火电厂设备状态实时监测系统设计[D]. 北京: 华北电力大学(北京), 2018. [5] 赵畅畅. 船用机械设备润滑磨损状态监测诊断研究[D]. 广州: 华南理工大学, 2018. [6] 王丹. 图像识别技术在配网设备状态监测中的应用研究[D]. 北京: 华北电力大学(北京), 2018. [7] 王博维,刘爱莲,杜景琦. 基于并行神经网络的水电机组振动状态劣化研究[J]. 电力科学与工程,2018, 34(9): 59-66. doi: 10.3969/j.ISSN.1672-0792.2018.09.010 [8] 朱少民,夏虹,彭彬森,等. 基于PCA的主泵传感器状态监测模型[J]. 核动力工程,2020, 41(3): 170-176. [9] 吴天昊,刘韬,施海宁,等. 基于核主元分析法的核电厂设备状态监测技术研究[J]. 核动力工程,2020, 41(5): 132-137. [10] 何鹏,陈静,李小芬,等. 核电厂堆腔冷却状态监测研究[J]. 核动力工程,2020, 41(5): 94-98. [11] 李翔,简捷,李海,等. 基于国产化PXI模块的松脱部件监测系统软件开发[J]. 核动力工程,2018, 39(3): 171-175.