基于自主运行的热管反应堆启堆过程研究

刘余; 黄蒙奇; 彭常宏; 杜政瑀

doi:10.13832/j.jnpe.2023.03.0144

基于自主运行的热管反应堆启堆过程研究

doi: 10.13832/j.jnpe.2023.03.0144

1.
中国核动力研究设计院核反应堆系统设计技术重点实验室，成都，610213
2.
中国科学技术大学核科学技术学院，合肥，230027

基金项目: 四川省杰出青年科技人才项目（2021JDJQ0034）；中国核动力研究设计院重点实验室创新团队项目（ZDSY-CXTD-21-04-001）；四川省科技计划（2019ZDZX0001）

详细信息

作者简介:
刘　余（1983—），男，研究员级高级工程师，现主要从事核反应堆热工与安全分析相关工作，E-mail: liuyu_npic@163.com

中图分类号: TL334
计量
- 文章访问数: 247
- HTML全文浏览量: 73
- PDF下载量: 51
- 被引次数: 0
出版历程
- 收稿日期: 2022-07-28
- 修回日期: 2022-10-20
- 刊出日期: 2023-06-15

Research on Heat Pipe Reactor Startup Process based on Autonomous Operation

1.
Science and Technology on Reactor System Design Technology Laboratory, Nuclear Power Institute of China, Chengdu, 610213, China
2.
School of Nuclear Science and Technology, University of Science and Technology of China, Hefei, 230027, China

摘要

摘要: 热管反应堆（HPR）的应用对无人自主运行技术提出了迫切需求，将自主运行技术应用于HPR，可实现状态感知、趋势预测、策略优化；能够有效避免人因失误；提升HPR技术性能、拓展核动力应用领域。以MegaPower热管堆为研究对象，以HPRTRAN程序为分析工具，针对热管堆运行过程中的重要组成——HPR启堆过程开展基于自主运行的研究，建立了适用于HPR启堆的、由监测诊断层-预测层-决策层组成的自主运行框架，初步开发了HPR自主运行系统。研究结果表明，自主运行系统预测结果准确性较高，决策方案较为科学且具备一定可行性。相关研究成果可为后续全面实现HPR无人值守自主运行奠定基础。
- 自主运行 /
- 热管反应堆（HPR） /
- 启堆过程 /
- 智能算法 /
- 决策
Abstract: The application of the heat pipe reactor (HPR) urgently requires unmanned autonomous operation technology. Applying autonomous operation technology to HPR can realize state sensing, trend prediction and strategy optimization, which can effectively avoid human errors, improve the technical performance of HPR and expand the application fields of nuclear power. In this paper, the MegaPower reactor was used as the research object and the HPRTRAN program was used as the analysis tool to carry out an autonomous operation research based on the HPR start-up process, which is an important component of the HPR operation process, and then an autonomous operation framework consisting of monitoring and diagnosis layer, prediction layer and decision layer was established for the HPR start-up. The research results show that the prediction results of the autonomous operation system are highly accurate and the decision-making scheme is scientific and feasible. The research results can lay a foundation for the subsequent full realization of the unattended autonomous operation of the HPR.
- Autonomous operation /
- Heat pipe reactor (HPR) /
- Reactor startup process /
- Intelligent algorithm /
- Decision-making

HTML全文

图 1 “蛙跳式”启堆控制策略^[8]

Figure 1. Control Strategy for "Leapfrog" Reactor Startup^[8]

下载: 全尺寸图片幻灯片

图 2 启堆过程功率和温度变化曲线

FP—满功率

Figure 2. Power and Temperature Variation Curve During Reactor Startup

下载: 全尺寸图片幻灯片

图 3 HPR启堆过程自主运行系统的层级架构

Figure 3. Hierarchical Architecture of Autonomous Operation System during HPR Startup

下载: 全尺寸图片幻灯片

图 4 泛化误差与偏差、方差的关系

Figure 4. Relationship Between Generalization Error, Bias and Variance

下载: 全尺寸图片幻灯片

图 5 数据集大小对模型评价得分的影响

Figure 5. Influence of Data Set Size on Model Evaluation Score　　　　　　

下载: 全尺寸图片幻灯片

图 6 模型的复杂程度对评价得分的影响

Figure 6. Influence of Model Complexity on Evaluation Score

下载: 全尺寸图片幻灯片

图 7 HPR启堆过程自主运行系统的决策层

max(A,B)—综合比较验证结果的最优决策与训练集中的最优决策后的最优决策结果

Figure 7. Decision -making Layer of Autonomous Operation System during HPR Startup

下载: 全尺寸图片幻灯片

图 8 不同算法的评价得分

Figure 8. Evaluation Score of Different Algorithms

下载: 全尺寸图片幻灯片

图 9 不同目标值的预测误差

Figure 9. Prediction Error for Different Target Values

下载: 全尺寸图片幻灯片

图 10 HPR启堆过程自主运行系统预测值与验证值的误差

Figure 10. Error in Prediction Values and Verification Values of Autonomous Operation System during HPR Startup

下载: 全尺寸图片幻灯片

图 11 热管堆启堆过程自主运行系统决策能力验证

Figure 11. Verification of Decision-making Ability of Autonomous Operation System during HPR Startup

下载: 全尺寸图片幻灯片

表 1 输入参数的取值范围和分布

Table 1. Range and Distribution of the Input Parameters

参数	取值范围	分布
单次旋转控制鼓角度	(0.5°, 5.5°)	每隔0.5°离散取值
旋转速度/[(°)·min⁻¹]	(10, 60)	均匀分布
最小观察时间/s	(1, 20)
温度稳定判据/K	(1, 30)
功率稳定判据	(0.003, 0.1)

下载: 导出CSV

表 2 降低模型泛化误差的方法^[15]

Table 2. Methods to Reduce Generalization Error of Models^[15]　　　　　　　　

参数名	降偏差	降方差
数据集	异常点处理扩大数据集	异常点处理扩大数据集
算法	增加算法复杂度集成学习	降低算法复杂度集成学习

下载: 导出CSV

表 3 集成回归算法比较

Table 3. Comparison of Ensemble Regression Algorithm

算法	基回归器类型	效果	集成方式	多样性处理
Bagging	同质	降方差	加权平均	训练集抽样+属性集抽样
Boosting	同质	降偏差	加权平均	训练集分布调整
Stacking	异质	降方差	再学习	不同基回归器+训练集抽样
Voting	异质	降方差	加权平均	不同基回归器

下载: 导出CSV

参考文献(20)

[1]	余红星,马誉高,张卓华,等. 热管冷却反应堆的兴起和发展[J]. 核动力工程,2019, 40(4): 1-8.
[2]	RAMUHALLI P, CETINER S M. Concepts for autonomous operation of microreactors: ORNL/TM-2019/1305[R]. Oak Ridge: Oak Ridge National Laboratory, 2019.
[3]	STERBENTZ J W, WERNER J E, MCKELLAR M G, et al. Special purpose nuclear reactor (5MW) for reliable power at remote sites assessment report: INL/EXT-16-40741[R]. Idaho Falls: Idaho National Laboratory, 2017.
[4]	MCCLURE P R, POSTON D I, DASARI V R, et al. Design of megawatt power level heat pipe reactors: LA-UR-15-28840[R]. Los Alamos: Los Alamos National Laboratory, 2015.
[5]	MA Y G, TIAN C Q, YU H X, et al. Transient heat pipe failure accident analysis of a megawatt heat pipe cooled reactor[J]. Progress in Nuclear Energy, 2021, 140: 103904. doi: 10.1016/j.pnucene.2021.103904
[6]	MA Y G, LIU J S, YU H X, et al. Coupled irradiation-thermal-mechanical analysis of the solid-state core in a heat pipe cooled reactor[J]. Nuclear Engineering and Technology, 2022, 54(6): 2094-2106. doi: 10.1016/j.net.2022.01.002
[7]	马誉高,杨小燕,刘余,等. MW级热管冷却反应堆反馈特性及启堆过程研究[J]. 原子能科学技术,2021, 55(S2): 213-220.
[8]	钟睿诚,马誉高,邓坚,等. 热管堆多反馈效应下的启堆特性研究[J]. 核动力工程,2021, 42(S2): 104-108. doi: 10.13832/j.jnpe.2021.S2.0104
[9]	郑龙生,田康生. 基于相对计算的雷达探测概率建模方法[J]. 空军雷达学院学报,2005, 19(4): 23-26.
[10]	张洪. 大亚湾核电站堆芯功率分布测量及其处理[J]. 核科学与工程,1997, 17(1): 6-16.
[11]	PAULLADA A, RAJI I D, BENDER E M, et al. Data and its (dis)contents: A survey of dataset development and use in machine learning research[J]. Patterns, 2021, 2(11): 100336. doi: 10.1016/j.patter.2021.100336
[12]	BROWN G, WYATT J L, TIŇO P. Managing diversity in regression ensembles[J]. Journal of Machine Learning Research, 2005, 6: 1621-1650.
[13]	ASH A, SHWARTZ M. R²: a useful measure of model performance when predicting a dichotomous outcome[J]. Statistics in Medicine, 1999, 18(4): 375-384. doi: 10.1002/(SICI)1097-0258(19990228)18:4<375::AID-SIM20>3.0.CO;2-J
[14]	FRIEDMAN J H. On bias, variance, 0/1-loss, and the curse-of-dimensionality[J]. Data Mining and Knowledge Discovery, 1997, 1(1): 55-77. doi: 10.1023/A:1009778005914
[15]	REN Y, ZHANG L, SUGANTHAN P N. Ensemble classification and regression—recent developments, applications and future directions[J]. IEEE Computational Intelligence Magazine, 2016, 11(1): 41-53. doi: 10.1109/MCI.2015.2471235
[16]	SHAHHOSSEINI M, HU G P, PHAM H. Optimizing ensemble weights and hyperparameters of machine learning models for regression problems[J]. Machine Learning with Applications, 2022, 7: 100251.
[17]	周志华. 机器学习[M]. 北京: 清华大学出版社, 2016: 186-188.
[18]	MENDES-MOREIRA J, SOARES C, JORGE A M, et al. Ensemble approaches for regression: A survey[J]. ACM Computing Surveys, 2012, 45(1): 10.
[19]	CETINER S M, MUHLHEIM M D, GULER YIGITOGLU A, et al. Supervisory control system for multi-modular advanced reactors[R]. Oak Ridge: Oak Ridge National Laboratory, 2016.
[20]	HUANG J C, KO K M, SHU M H, et al. Application and comparison of several machine learning algorithms and their integration models in regression problems[J]. Neural Computing and Applications, 2020, 32(10): 5461-5469. doi: 10.1007/s00521-019-04644-5