基于SSA-LSTM的海洋条件下稳压器液位回归预测研究

李东阳; 权紫轩; 张彪; 李江宽; 谭思超; 田瑞峰

doi:10.13832/j.jnpe.2024.050045

基于SSA-LSTM的海洋条件下稳压器液位回归预测研究

doi: 10.13832/j.jnpe.2024.050045

李东阳^{1, 2,},
权紫轩^{1, 2},
张彪³,
李江宽^{1, 2},
谭思超^{1, 2},
田瑞峰^{1, 2}

1.
哈尔滨工程大学黑龙江省核动力装置性能与设备重点实验室，哈尔滨，150001
2.
哈尔滨工程大学核安全与先进核能技术工信部重点实验室，哈尔滨，150001
3.
武汉第二船舶设计研究所，武汉，430062

详细信息

作者简介:
李东阳（1986—），男，副教授，博士生导师，现主要从事反应堆热工水力方面的研究，E-mail: lidy@hrbeu.edu.cn

中图分类号: TL334
计量
- 文章访问数: 37
- HTML全文浏览量: 11
- PDF下载量: 0
- 被引次数: 0
出版历程
- 收稿日期: 2024-05-15
- 修回日期: 2024-06-17
- 网络出版日期: 2025-01-15
- 刊出日期: 2025-04-15

Research on Regression Prediction of Pressurizer Liquid Level under Ocean Conditions Based on SSA-LSTM Neural Network

Li Dongyang^{1, 2
,},
Quan Zixuan^{1, 2},
Zhang Biao³,
Li Jiangkuan^{1, 2},
Tan Sichao^{1, 2},
Tian Ruifeng^{1, 2}

1.
Heilongjiang Provincial Key Laboratory of Nuclear Power System & Equipment , Harbin Engineering University, Harbin, 150001, China
2.
Key Laboratory of Nuclear Safety and Advanced Nuclear Energy Technology, Ministry of Industry and Information Technology, Harbin Engineering University, Harbin, 150001, China
3.
Wuhan Second Ship Design and Research Institute, Wuhan, 430062, China

摘要

摘要: 为保证核动力装置在海洋环境下的安全运行，有必要建立一套计算模型获得稳压器内的实时液位。通过搭建实验系统采集相关数据，采用基于麻雀搜索算法（SSA）优化长短期记忆（LSTM）神经网络（SSA-LSTM），根据测得的压力、运动姿态等关键参数建立液位回归预测模型。研究结果表明，所建立的液位回归预测模型预测精度优秀，明显优于其他传统神经网络。此外，该模型的泛化能力良好，对于新鲜样本的预测精度也较高，将其集成到控制系统中可实时输出稳压器液位，从而提高海洋条件下核动力装置运行的安全性，并为后续核动力装置的智能运维提供参考。
- 海洋条件 /
- 麻雀搜索算法（SSA） /
- 长短期记忆（LSTM）神经网络 /
- 液位回归预测
Abstract: To ensure the safe operation of the nuclear reactor system in the ocean environment, it is necessary to establish a set of computational models to obtain the real-time liquid level in the pressurizer. By building an experimental system to collect relevant data, the Long-Short Term Memory (LSTM) neural network is optimized based on the sparrow search algorithm (SSA), and the liquid level regression prediction model is established according to the measured key parameters such as pressure and motion attitude. The research results show that the prediction accuracy of the established liquid level regression prediction model is excellent, which is obviously better than other traditional neural networks. The model has good generalization ability, and the prediction accuracy of fresh samples is still acceptable. By integrating the model into the control system, the liquid level can be output in real time, which can improve the safety of nuclear power operation under ocean conditions and provide reference for the intelligent operation and maintenance of nuclear power.
- Ocean conditions /
- Sparrow search algorithm (SSA) /
- Long-Short Term Memory (LSTM) neural network /
- Liquid level regression prediction

HTML全文

图 1 实验系统

Figure 1. Experimental System

下载: 全尺寸图片幻灯片

图 2 LSTM神经网络原理图

Figure 2. Schematic Diagram of LSTM Neural Network

下载: 全尺寸图片幻灯片

图 3 SSA-LSTM模型流程图

Figure 3. Flowchart of SSA-LSTM Model

下载: 全尺寸图片幻灯片

图 4 不同工况下训练集的实验数据

Figure 4. Experiment Data of Training Set under Different Operating Conditions

下载: 全尺寸图片幻灯片

图 5 训练过程中的学习曲线

Figure 5. Learning Curves during the Training Process

下载: 全尺寸图片幻灯片

图 6 不同工况下的不同神经网络预测值与真实值对比

Figure 6. Comparison between Predicted Values and True Values of Different Neural Networks under Different Operating Conditions

下载: 全尺寸图片幻灯片

表 1 实验系统各参数的测量精度 %

Table 1. Measurement Accuracy of Experimental System Parameters

参数	仪表不确定度	采集不确定度	B类不确定度
压力	0.025	0.5	0.501
运动姿态	0.330		0.330
水箱质量	0.500	0.5	0.707

下载: 导出CSV

表 2 实验工况设置

Table 2. Experimental Operating Conditions

工况编号	水位状态	横摇条件	纵摇条件
P1	上充	15°/5 s
P2	下泄	15°/7 s
P3	上充	10°/5 s	10°/10 s
P4	下泄	10°/7 s	10°/10 s
T1	下泄	15°/5 s
T2	上充	10°/7 s	10°/10 s
T3	上充	15°/7 s
T4	下泄	10°/5 s	5°（倾斜）
T5	上充	10°/5 s	5°（倾斜）

下载: 导出CSV

表 3 不同数据集下的模型表现

Table 3. Model Performance under Various Datasets

数据集/个	RMSE/mm	R²
2000	6.838±0.663	0.756±0.048
3000	5.352±0.082	0.849±0.005
4000	4.967±0.167	0.871±0.009
5000	4.371±0.119	0.901±0.005
6000	4.368±0.197	0.903±0.009
7000	4.245±0.228	0.905±0.010
8000	4.291±0.112	0.902±0.006

下载: 导出CSV

表 4 不同神经网络下的RMSE数据对比 mm

Table 4. Comparison of RMSE Data under Different Neural Networks

工况编号	LSTM	BP	ELM
T1	4.316	15.211	10.139
T2	2.326	9.386	11.497
T3	4.458	13.238	8.810
T4	13.420	10.795	14.423
T5	2.128	10.053	4.445

下载: 导出CSV

参考文献(16)

[1]	LI D Y, QUAN Z X, ZHANG B, et al. Study on the liquid level characteristics and reconstruction strategy of the in-vessel pressurizer under ocean conditions[J]. Nuclear Engineering and Design, 2024, 418: 112857. doi: 10.1016/j.nucengdes.2023.112857
[2]	WEI T Y, CHEN J R, ZHANG B, et al. Effects of ocean conditions on water level measurement of pressurizer[J]. Annals of Nuclear Energy, 2022, 169: 108913. doi: 10.1016/j.anucene.2021.108913
[3]	LI D Y, YUE J Y, WEI T Y, et al. Optimization of pressure tube-type measuring instrument for floating nuclear power plant[J]. Annals of Nuclear Energy, 2024, 205: 110558. doi: 10.1016/j.anucene.2024.110558
[4]	CONG T L, SU G H, QIU S Z, et al. Applications of ANNs in flow and heat transfer problems in nuclear engineering: a review work[J]. Progress in Nuclear Energy, 2013, 62: 54-71. doi: 10.1016/j.pnucene.2012.09.003
[5]	张恒,吕雪,刘东,等. 核电人工智能应用: 现状、挑战和机遇[J]. 核动力工程,2023, 44(1): 1-8.
[6]	陈涵瀛. 核电站热工水力系统工况预测与诊断方法研究[D]. 哈尔滨: 哈尔滨工程大学,2018.
[7]	MARSEGUERRA M, ZIO E, AVOGADRI P. Model identification by neuro-fuzzy techniques: predicting the water level in a steam generator of a PWR[J]. Progress in Nuclear Energy, 2004, 44(3): 237-252. doi: 10.1016/S0149-1970(04)90012-1
[8]	谭季秋,鄂加强,钟定清. 联合循环余热锅炉汽包水位晃荡信号智能预测[J]. 中南大学学报: 自然科学版,2012, 43(3): 966-971.
[9]	魏天一,张彪,李东阳,等. 海洋条件下小型堆稳压器液位智能预测研究[J]. 哈尔滨工程大学学报,2023, 44(9): 1590-1596. doi: 10.11990/jheu.202201031
[10]	王磊,马驰骋,齐俊艳,等. 基于ARIMA-LSTM的矿区地表沉降预测方法[J]. 计算机工程,2025,51(1): 98-105.
[11]	孙大彬,李磊,田兆斐,等. 混合神经网络的包壳峰值温度预测研究[J]. 哈尔滨工程大学学报,2022, 43(12): 1728-1735.
[12]	刘岳,于静,金秀章. 基于特征优化和改进长短期记忆神经网络的NO_x质量浓度预测[J]. 热力发电,2021, 50(7): 162-169.
[13]	LI Z L, DU J M, ZHU W M, et al. Regression predictive modeling of high-speed motorized spindle using POA-LSTM[J]. Case Studies in Thermal Engineering, 2024, 54: 104053. doi: 10.1016/j.csite.2024.104053
[14]	HOCHREITER S, SCHMIDHUBER J. Long short-term memory[J]. Neural Computation, 1997, 9(8): 1735-1780. doi: 10.1162/neco.1997.9.8.1735
[15]	薛建凯. 一种新型的群智能优化技术的研究与应用——麻雀搜索算法[D]. 上海: 东华大学,2019.
[16]	XUE J K, SHEN B. A novel swarm intelligence optimization approach: sparrow search algorithm[J]. Systems Science & Control Engineering, 2020, 8(1): 22-34.