基于多保真度数据和迁移学习的凝汽器系统特性建模方法

张逸佳; 任少君; 朱保宇; 翁琪航; 卫子涵; 司风琪

doi:10.19666/j.rlfd.202501029

English

您当前的位置：

首页 >

文章列表页 >

基于多保真度数据和迁移学习的凝汽器系统特性建模方法

热力发电 2025年54卷第11期页码：107-116

热能科学研究 | 更新时间：2026-08-11

- 基于多保真度数据和迁移学习的凝汽器系统特性建模方法
- Modeling method for characterizing condenser system based on multi-fidelity data and transfer learning
- 作者机构：
  
  东南大学能源热转换及过程测控教育部重点实验室
- 作者简介：
- 基金信息：
  
  国家自然科学基金项目（52306230）;国家重点研发计划项目（2022YFB4100700）~~
- DOI：10.19666/j.rlfd.202501029
  中图分类号： TM621
- 纸质出版：2025
- 稿件说明：
移动端阅览
[1]张逸佳,任少君,朱保宇,等.基于多保真度数据和迁移学习的凝汽器系统特性建模方法[J],2025,54(11):107-116.
[1]张逸佳,任少君,朱保宇,等.基于多保真度数据和迁移学习的凝汽器系统特性建模方法[J],2025,54(11):107-116. DOI： 10.19666/j.rlfd.202501029.

DOI：10.19666/j.rlfd.202501029.

摘要

数据驱动模型有效性取决于训练样本是否完备，对于样本覆盖范围外的工况，模型泛化能力较差。为构建适应机组宽负荷变化的凝汽器系统特性模型，训练样本需包括较大变化范围的发电负荷和环境温度。然而，新建机组因投运时间短，无法提供完备数据集。因此，针对不完备数据集提出了一种基于多保真度数据和迁移学习的凝汽器系统特性建模方法。该方法首先采用同类型系统的完备数据集构建预训练模型，在预训练模型上，新增线性和非线性校准网络，通过新建机组的不完备数据对校准网络进行更新，使预训练模型迁移适配不完备数据集的特征空间。最后通过某1 000 MW超临界机组凝汽器系统验证该方法有效性，验证结果表明，在训练样本不充分的情况下，该方法能准确预测凝汽器压力和循环水出口温度等参数，决定系数R

平均值为0.95，明显优于传统单一数据集数据驱动模型的R

平均值0.81。

Abstract

The effectiveness of a data-driven model relies on the completeness of its training samples. For operating conditions beyond the scope of the training samples

the model's generalization ability is compromised. Therefore

to develop a condenser model that can adapt to the wide load variation of the unit

it is essential for the training samples to involve a diverse range of power generation loads and ambient temperatures. However

achieving this complete dataset is difficult for newly-commissioned units because of their short operation time. To address these challenges

a method for characterizing

condensing units using multi-fidelity data and transfer learning is proposed

even with incomplete data. In this method

a pre-trained model is firstly built based on the comprehensive operational dataset collected from a similar unit. On the basis of the pre-trained model

additional linear and nonlinear calibration networks are introduced. The calibration networks are updated through the incomplete data of newly constructed units

enabling the transfer of the pre-training model to the feature space that is adapted to the incomplete dataset. The effectiveness of this method is validated through the condenser of a 1 000 MW supercritical unit. The results indicate that

even with limited training samples

the method accurately predicts parameters such as condenser pressure and circulating water outlet temperature

with an average R

of 0.95

significantly outperforming the conventional data-driven model based on a single data set

of which the average R

is only 0.81.

关键词

Keywords

references

杜小泽,金衍胜,姜剑波,等.火电厂直接空冷凝汽器传热性能实验研究[J].工程热物理学报, 2009, 30(1):99-101.DU Xiaoze, JIN Yansheng, JIANG Jianbo, et al.Experimental study on heat transfer performance of direct air-cooled condenser in power plant[J]. Journal of Engineering Thermophysics, 2009, 30(1):99-101.

JIANG J. Fault diagnosis of condenser based on wavelet neural network[J]. Industrial Engineering Journal, 2010,13(4):104-107.

ZHONG D, ZENG H, MENG J, et al. Numerical simulation of power plant condenser performance and roles for tube arrangement[J]. Journal of Engineering Thermophysics, 2014, 35(1):123-127.

薛倩.凝汽器强化传热研究[J].当代化工, 2020, 49(9):2037-2040.XUE Qian. Study on enhanced heat transfer of condenser in power plant[J]. Contemporary Chemical Industry, 2020,49(9):2037-2040.

WANG J, WANG H, BI W. Research of energy consumption of evaporative condenser in ammonia refrigeration system[J]. Cryogenics&Superconductivity,2011, 31(10):74-77.

WANG Q, LIU B, DONG X, et al. Performance characteristics of microchannel condenser under different heat loads[J]. Cryogenics, 2014(5):58-62.

HAJIDAVALLOO E, EGHTEDARI H. Performance improvement of air-cooled refrigeration system by using evaporatively cooled air condenser[J]. International Journal of Refrigeration, 2010, 33(5):982-988.

张远绪.基于PSO-DDSAEN的凝汽器故障诊断研究[D].青岛:青岛科技大学, 2019:1.ZHANG Yuanxu. Research on condenser fault diagnosis based on PSO-DDSAEN[D]. Qingdao:Qingdao University of Science and Technology, 2019:1.

冯金金,章颢,李健平,等. 1 000 MW燃煤机组双压凝汽器冷端系统运行优化[J].科技和产业, 2024, 24(21):198-204.FENG Jinjin, ZHANG Hao, LI Jianping, et al. Operation optimization of the cold end system with dual-pressure condenser of 1 000 MW coal-fired unit[J]. Science Technology and Industry, 2024, 24(21):198-204.

HAFFEJEE R A, LAUBSCHER R. Development of a thermofluid network modeling methodology for doublerow air-cooled condensers[J]. Thermal Science and Engineering Progress, 2020, 19:100646.

朱自强.不同凝汽器冷却方式的燃气蒸汽联合循环系统热经济性研究[D].北京:华北电力大学, 2022:1.ZHU Ziqiang. Study on thermoeconomic performance of gas-steam combined cycle system with different condenser cooling modes[D]. Beijing:North China Electric Power University, 2022:1.

代雨辰.基于机器学习的凝汽器设备故障分类方法研究[D].南京:东南大学, 2022:1.DAI Yuchen. Research on fault classification method of condenser based on machine learning[D]. Nanjing:Southeast University, 2022:1.

WU D, ZHOU D, ZHANG J, et al. Multimode process monitoring based on fault dependent variable selection and moving window-negative log likelihood probability[J]. Computers&Chemical Engineering, 2020,136:106787.

WU H, ZHAO J. Self-adaptive deep learning for multimode process monitoring[J]. Computers&Chemical Engineering, 2020, 141(20):107024.

ZAMZAM A S, FU X, SIDIROPOULOS N D. Datadriven learning-based optimization for distribution system state estimation[J]. IEEE Transactions on Power Systems,2019, 34(6):4796-4805.

高建强,王艳.基于BP神经网络的直接空冷凝汽器换热性能预测[J].动力工程学报, 2013, 33(6):443-447.GAO Jianqiang, WANG Yan. BP neural network prediction on heat-transfer performance of direct aircooled condensers[J]. Journal of Chinese Society of Power Engineering, 2013, 33(6):443-447.

LI X, WANG N, WANG L, et al. A data-driven model for the air-cooling condenser of thermal power plants based on data reconciliation and support vector regression[J].Applied Thermal Engineering Design Processes Equipment Economics, 2018, 129:1496-1507.

LAI C, BARALDI P, ZIO E. Physics-informed deep autoencoder for fault detection in new-design systems[J].Mechanical Systems&Signal Processing, 2024, 215:111420.

王力光,贠勇博,朱保宇,等.基于迁移学习的湿法烟气脱硫系统出口SO2浓度预测研究[J].发电设备,2021, 35(6):420-425.WANG Liguang, YUN Yongbo, ZHU Baoyu, et al.Research on the prediction of SO2 concentration at the outlet of wet flue gas desulfurization system based on transfer learning[J]. Power Equipment, 2021, 35(6):420-425.

BENJAMIN P, KAREN W, MAX G. Survey of multifidelity methods in uncertainty propagation,inference, and optimization[J]. SIAM Review, 2018,60(3):550-591.

TORZONI G, MANZONI R, MARIANI S. A multifidelity surrogate model for structural health monitoring exploiting model order reduction and artificial neural networks[J]. Mechanical Systems and Signal Processing,2023, 197:110376.

任少君,朱保宇,翁琪航,等.基于数据增强和模型迁移的生物质气化产物分布预测方法[J].中国电机工程学报, 2024, 44(18):7309-7321.REN Shaojun, ZHU Baoyu, WENG Qihang, et al.Prediction method of biomass gasification product distribution based on data augmentation and model migration[J]. Proceedings of the CSEE, 2024, 44(18):7309-7321.

GENG X, LIU P, HU T, et al. Multi-fidelity optimization of a quiet propeller based on deep deterministic policy gradient and transfer learning[J]. Aerospace Science and Technology, 2023, 137:108288.

YE X, NI Y Q, SAJJADI M, et al. Physics-guided, datarefined modeling of granular material-filled particle dampers by deep transfer learning[J]. Mechanical Systems&Signal Processing, 2022, 180:109437.

GISELLE M, PARK C, KIM N H, et al. Review of multifidelity models[J]. Advances in Computational Science and Engineering, 2017, 1(4):351-400.

浏览量

下载量

CSCD

文章被引用时，请邮件提醒。

提交

关联资源

基于TrAdaBoost迁移学习的跨仪器LIBS煤质定量分析模型研究

凝汽器管束入口蒸汽流场对汽轮机排汽压力的影响

135MW间接空冷机组双速循环水泵设计与经济运行

活动收球网新型传动方案设计

减少凝汽器铜管泄漏的措施