Research on Intelligent Monitoring and Warning Algorithms for Unexpected Reactor Shutdown Events in Nuclear Power Plants
-
摘要: 当前核电机组运行时的异常情况发现主要依赖于核电厂数字化仪控系统（DCS）的阈值报警信息，缺乏对趋势的分析。本文通过事件逻辑建立变量间的逻辑关系，并基于此利用自关联神经网络（AANN）建模对关联变量进行异常检测，最后利用经验模态分解（EMD）趋势提取算法与自适应滑动窗口霍尔特线性趋势（HOLT）模型对异常变量进行预测。能够提前对停堆停机事件进行预警，使核电厂运维人员能够更早地发现并解决问题，提高核电运行安全性。利用仿真数据与机组真实异常数据进行测试实验，得到真实数据实验结果的均方误差（MSE）为0.1，拟合优度（R2）为0.99，并且可至少提前1 h对停机动作进行预警，验证了所提出的AANN-HOLT预警算法的准确性与提前预警的能力。
-
关键词:
- 智能监测 /
- 时间序列 /
- 趋势预测；停堆停机预警
Abstract: The detection of abnormal conditions during the operation of nuclear power plant units mainly relies on threshold alarm information from the Digital Control System (DCS), with a lack of trend analysis. This paper investigates the establishment of logical relationships between variables through event logic, and based on this, employs an Auto-Associative Neural Network (AANN) model for anomaly detection of correlated variables. Finally, it uses the Empirical Mode Decomposition (EMD) trend extraction algorithm and the Adaptive Sliding Window Holt Linear Trend (HOLT) model to predict abnormal variables. This approach can provide early warnings for shutdown and reactor trip events, enabling plant operators to detect and resolve issues earlier, thus improving the operational safety of nuclear power plants. Testing experiments were conducted using both simulated data and actual unit anomaly data. The results from real data experiments show a Mean Squared Error (MSE) of 0.1 and a Goodness of Fit (R2) of 0.99, with at least 1 hour of advance warning before shutdown actions. This confirms the accuracy and early warning capabilities of the proposed AANN-HOLT warning algorithm. -
图 4 1%一环路冷端破口基于线性模型的测点数据真实值与预测值对比图
Figure 4. Comparison Chart of Actual and Predicted Values Based on a Linear Model for a 1% Cold End Break in the First Loop
图 5 1%一环路冷端破口基于HOLT模型的测点数据真实值与预测值对比图
Figure 5. Comparison Chart of Actual and Predicted Values Based on the HOLT Model for a 1% Cold End Break in the First Loop
图 6 1%一环路冷端破口基于GRU模型的测点数据真实值与预测值对比图
Figure 6. Comparison Chart of Actual and Predicted Values Based on the GRU Model for a 1% Cold End Break in the First Loop
图 7 2%一环路冷端破口基于HOLT模型的测点数据真实值与预测值对比图
Figure 7. Comparison Chart of Actual and Predicted Values Based on the HOLT Model for a 2% Cold End Break in the First Loop
图 8 2%一环路冷端破口基于HOLT模型的测点数据停堆触发倒计时预测图
Figure 8. Prediction of Shutdown Trigger Countdown Based on the HOLT Model for a 2% Cold End Break in the First Loop
图 9 RCP泵跳闸基于HOLT模型的测点数据真实值与预测值对比图
Figure 9. Comparison Chart of Actual and Predicted Values of RCP Pump Trip Point Data Based on the HOLT Model
图 10 RCP泵跳闸基于HOLT模型的测点数据停堆触发倒计时预测图
Figure 10. Prediction of Shutdown Trigger Countdown of RCP Pump Trip Point Data Based on the HOLT Model
图 11 基于AANN-HOLT模型的机组数据验证结果
Figure 11. Unit Data Validation Results Based on the AANN-HOLT Model
表 1 3种模型针对稳压器压力测点的MSE和R2对比
Table 1. Comparison of MSE and R2 of Three Models for Pressure Measurement Points of Pressurizers
压力测点 线性模型 HOLT模型 GRU模型 MSE R2 MSE R2 MSE R2 RCP005MP 0.0011 0.9965 0.0004 0.9984 0.0627 −1.4738 RCP005MP（含噪声） 0.0058 0.9826 0.0048 0.9844 0.0637 −1.2542 RCP006MP 0.0011 0.9965 0.0002 0.9991 0.0583 −1.2994 RCP006MP（含噪声） 0.0062 0.9837 0.0046 0.9850 0.9533 −2.4050 RCP013MP 0.0011 0.9966 0.0003 0.999 0.0224 0.1153 RCP013MP（含噪声） 0.0059 0.9850 0.0048 0.9861 0.1405 −1.6469 
下载: 导出CSV
表 2 稳压器压力测点测试MSE和R2
Table 2. MSE and R2 of Pressure Measurement Point Test
压力测点 MSE R2 RCP005MP 0.00012 0.9996 RCP005MP（含噪声） 0.0069 0.9767 RCP006MP 0.00014 0.9995 RCP006MP（含噪声） 0.0063 0.9823 RCP013MP 0.00012 0.9996 RCP013MP（含噪声） 0.0089 0.9730
下载: 导出CSV
表 3 二环路流量测点的测试MSE和R2对比
Table 3. MSE and R2 of Secondary Loop Flow Measurement Point Test
流量测点 MSE R2 RCP040MD 0.1853 0.9954 RCP040MD（含噪声） 0.5469 0.9737 RCP041MD 0.1513 0.9971 RCP041MD（含噪声） 0.5493 0.9565 RCP042MD 0.2106 0.9941 RCP042MD（含噪声） 0.5260 0.9680
下载: 导出CSV
表 4 基于AANN-HOLT模型的机组数据测试MSE和R2以及提前预警时间
Table 4. Testing MSE and R2 and Early Warning Time of Unit Data Based on the AANN-HOLT Model
液位测点 MSE R2 开始预警时刻 提前预警时间 GFR001MN 0.1197 0.9999 2020-07-19 11:32:23 1 h 8 min 31 s GFR002MN 0.1124 0.9999 2020-07-19 11:32:33 1 h 8 min 21 s GFR003MN 0.1179 0.9999 2020-07-19 11:32:31 1 h 8 min 23 s
下载: 导出CSV
-
[1] 郑明光,张琴舜,徐济,等. 核电厂先进控制室报警系统[J]. 核动力工程,2001, 22(4): 354-359,384. doi: 10.3969/j.issn.0258-0926.2001.04.015 [2] 许勇,蔡云泽,宋林. 基于数据驱动的核电设备状态评估研究综述[J]. 上海交通大学学报,2022, 56(3): 267-278. [3] STEPHEN B, WEST G M, GALLOWAY S, et al. The use of Hidden Markov Models for anomaly detection in nuclear core condition monitoring[J]. IEEE Transactions on Nuclear Science, 2009, 56(2): 453-461. doi: 10.1109/TNS.2008.2011904 [4] 周柯,王晓明,李肖博,等. 基于ARIMA-BP组合模型的智能变电站遥测数据趋势性分析预警技术研究[J]. 哈尔滨理工大学学报,2023, 28(1): 97-104. [5] 安吉振,乔世超,陈衡,等. 基于多元状态估计和向量相似度的电站磨煤机故障智能预警研究[J]. 热力发电,2022, 51(12): 64-71. [6] 黄宇斐,石新发,贺石中,等. 一种基于主成分分析与支持向量机的风电齿轮箱故障诊断方法[J]. 热能动力工程,2022, 37(10): 175-181. [7] NGUYEN H P, LIU J, ZIO E. A long-term prediction approach based on long short-term memory neural networks with automatic parameter optimization by Tree-structured Parzen Estimator and applied to time-series data of NPP steam generators[J]. Applied Soft Computing, 2020, 89: 106116. doi: 10.1016/j.asoc.2020.106116 [8] 赵志宏,李晴,杨绍普,等. 基于BiLSTM与注意力机制的剩余使用寿命预测研究[J]. 振动与冲击,2022, 41(6): 44-50,196. [9] 朱少民,夏虹,吕新知,等. 基于ARIMA和LSTM组合模型的核电厂主泵状态预测[J]. 核动力工程,2022, 43(2): 246-253. [10] 宋威,丁一,赵凯,等. 基于EMD-LSTM的风机故障停机发生时间预测[J]. 计算机仿真,2023, 40(12): 113-118. doi: 10.3969/j.issn.1006-9348.2023.12.020 [11] HUANG N E, SHEN Z, LONG S R, et al. The empirical mode decomposition and the Hilbert spectrum for nonlinear and non-stationary time series analysis[J]. Proceedings of the Royal Society of London. Series A: Mathematical, Physical and Engineering Sciences, 1998, 454(1971): 903-995. doi: 10.1098/rspa.1998.0193 [12] CHO K, VAN MERRIËNBOER B, GULCEHRE C, et al. Learning phrase representations using RNN encoder–decoder for statistical machine translation[C]//Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP). Doha: Association for Computational Linguistics, 2014: 1724-1734. -
计量
- 文章访问数: 33
- HTML全文浏览量: 13
- PDF下载量: 0
- 被引次数: 0
