Prediction of S-CO<sub>2</sub> flow resistance characteristics based on Bayesian-regularization BP neural network

WU Quan; ZHU Meng; CHEN Lei; QING Haoran; ZHOU Lingang; XU Kai; SU Sheng; HU Song; LIU Hui; XIANG Jun

doi:10.19666/j.rlfd.202112259

English

您当前的位置：

首页 >

文章列表页 >

Prediction of S-CO₂ flow resistance characteristics based on Bayesian-regularization BP neural network

Thermal Power Generation Vol. 51, Issue 6, Pages: 50-58(2022)

更新时间：2025-09-28

- Prediction of S-CO₂ flow resistance characteristics based on Bayesian-regularization BP neural network
- 作者机构：
  
  1. 国家能源集团新能源技术研究院有限公司
  2. 华中科技大学煤燃烧国家重点实验室
- 作者简介：
- 基金信息：
- DOI：10.19666/j.rlfd.202112259
  CLC： TP183;TM611.3
- Published：2022
- 稿件说明：
移动端阅览
WU Quan, ZHU Meng, CHEN Lei, et al. Prediction of S-CO₂ flow resistance characteristics based on Bayesian-regularization BP neural network[J]. Thermal Power Generation, 2022, 51(6): 50-58.
DOI：

WU Quan, ZHU Meng, CHEN Lei, et al. Prediction of S-CO₂ flow resistance characteristics based on Bayesian-regularization BP neural network[J]. Thermal Power Generation, 2022, 51(6): 50-58. DOI： 10.19666/j.rlfd.202112259.

摘要

超临界二氧化碳（S-CO_2）流动摩擦阻力特性尚不明晰，现有阻力预测公式泛化能力不强。对S-CO_2在垂直上升光滑圆管中进行湍流流动的阻力特性进行了实验研究，系统分析了不同质量流速、压力和热流密度对S-CO_2流动摩擦阻力的影响，实验参数范围为质量流速750～2 200 kg/(m2·s)，流体压力10～20 MPa，热流密度200～340 kW/m2以及主流温度60～500℃（远离拟临界点）。实验结果表明：S-CO_2流动摩擦阻力随热流密度的增加而减小，这有利于S-CO_2发电的大规模工程应用；S-CO_2流动摩擦阻力随压力的增加而减小，主要原因在于压力增加导致CO_2流体密度增加。根据实验数据，构建了贝叶斯正则化BP神经网络模型，实现了对S-CO_2流动摩擦阻力的有效预测。预测值与实验值的相关系数R=0.998 6，超过98%的预测值均在10%的误差范围内，表明该模型拟合精度高，具有较好的泛化能力。

Abstract

The flow resistance characteristics of supercritical carbon dioxide(S-CO_2) are still unclear

and the generalization ability of existing resistance prediction formula is poor. The resistance characteristics of S-CO_2turbulent flow in a vertically ascending smooth circular tube were experimentally studied

and the effects of mass flow rate

pressure and heat flux on friction resistance of S-CO_2 were systematically analyzed. The experimental parameters' range are: 750～2 200 kg/(m2·s) of mass flow rate

10～20 MPa of fluid pressure

200～340 kW/m2 of heat flux and 60～500 ℃ of bulk temperature(away from the pseudo critical point). The experimental results show that

the friction resistance of S-CO_2 decreases with the increase of heat flux

which is beneficial to the engineering application of S-CO_2 power generation. The flow friction resistance of S-CO_2 decreases with the increase of pressure

which is mainly because the fluid density increases with the pressure. Furthermore

a Bayesianregularization BP neural network model was established to predict the flow friction resistance of S-CO_2 effectively.The correlation coefficient R between the predicted values and the experimental values is 0.998 6

and more than 98% of the predicted values are within the error range of 10%

indicating that the model has good generalization ability and high fitting accuracy.

关键词

Keywords

references

Views

213

下载量

CSCD

Alert me when the article has been cited

提交

Publicity Resources

STUDY ON APPLICATION OF PREDICTIVE CONTROL BASED ON SUPERVISORY INFORMATION SYSTEM

STUDY ON APPLICATION OF INTELLIGENT COMBUSTION-OPTIMIZING TECHNOLOGY ON UTILITY BOILERS

PREDICTION MODEL FOR POLLUTANTS SO₂ AND NO_x FROM LARGE-SCALE COAL-FIRED POWER PLANT BASED ON NEURAL NETWORK

A MODEL FOR PREDICTING VACUUM IN THE CONDENSER BASED ON ELMAN NEURAL NETWORK BY USING PARTICLE SWARM OPTIMIZATION ALGORITHM

STUDY ON PREDICITION OF TUBE-WALL TEMPERATURE IN SUPERHEATER AND REHEATER OF BOILER BASED ON NEURAL NETWORK

Related Author

LUO Yi

ZHANG Yuan-yuan

ZHANG Yi

DING Yan-jun North China Electric Power University

WU Zhi-qun1

HUANG Ting-hui1

HU Hong-hua1

L Wen-liang21.Xi'an Thermal Power Research Institute Co Ltd

Related Institution

North China Electric Power University

Tsinghua University

Xi'an Thermal Power Research Institute Co., Ltd

China Power Investment Qinghe Power Generation Co., Ltd

Zhejiang University Energy Engineering Design and Research Institute

AI问答

⁰

Prediction of S-CO2 flow resistance characteristics based on Bayesian-regularization BP neural network

DOI：10.19666/j.rlfd.202112259

摘要

Abstract

关键词

Keywords

references

Prediction of S-CO₂ flow resistance characteristics based on Bayesian-regularization BP neural network