文章检索
首页» 过刊浏览» 2023» Vol.8» Issue(1) 102-111     DOI : 10.3969/j.issn.2096-1693.2023.01.007
最新目录| | 过刊浏览| 高级检索
基于物理模型驱动的机器学习方法预测超临界二氧化碳管道最大泄漏速率
王一新, 陆诗建, 李卫东, 滕霖
1 福州大学石油化工学院,福州 350108 2 中国矿业大学碳中和研究院,徐州 221008 3 重庆大学产业技术研究院,重庆 401329
A physical model driven machine learning for predicting maximum leakage rate in supercritical CO2 release
WANG Yixin, LU Shijian, LI Weidong, TENG Lin
1 College of Chemical Engineering, Fuzhou University, Fuzhou 350116, China 2 Carbon Neutrality Institute, China University of Mining and Technology, Xuzhou 221008, China 3 Chongqing University Industrial Technology Research Institute, Chongqing 401329, China

全文:   HTML (1 KB) 
文章导读  
摘要  碳捕集与封存 (CCS)项目中涉及的大规模 CO2 适合采用超临界管道输送。然而超临界 CO2 管道泄漏过程 伴随着复杂相变,因此对其最大泄漏速率进行准确预测是目前的研究难点。鉴于传统物理模型方法存在建模复 杂、假设过多、计算耗时等缺点,研究提出通过机器学习方法预测超临界CO2 管道最大泄漏速率,分别采用粒 子群算法优化的支持向量机 (PSO-SVM)和简化处理的卷积神经网络 (CNN)对等熵阻塞泄漏模型所生成的泄漏特 征数据进行学习,并测试了机器学习模型的预测准确率和泛化能力。研究结果表明:①物理模型、 PSO-SVM、 CNN的预测结果与实验数据的平均误差为 28.82%;②两种机器学习模型预测精度相差不大, CNN的训练时间 远短于 PSO-SVM,但 PSO-SVM的泛化能力强于 CNN,因此, SVM适用于小样本数据精确预测,而CNN更适 用于对大数据的学习和预测。本研究成果为超临界 CO2 管道最大泄漏速率预测提供了一种高效的新方法。
服务
把本文推荐给朋友
加入我的书架
加入引用管理器
关键词 : 机器学习;超临界二氧化碳;管道;泄漏;卷积神经网络;支持向量机
Abstract

Supercritical CO2 pipelines are suitable to transport the large-scale CO2 involved in Carbon capture and storage projects. The leakage process of supercritical CO2 pipelines is accompanied by complex phase changes. Therefore, it is difficult to predict the maximum leakage rate accurately at present. In view of the shortcomings of traditional physical model methods such as complex modeling, too many assumptions and time-consuming calculations, a way of predicting the maximum leakage rate of supercritical CO2 pipelines by machine learning method was proposed. It used to simply convolutional neural networks (CNN) and support vector machine improved by particle swarm optimization (PSO-SVM) respectively to study the leakage feature data generated by the isentropic choked flow leakage model. The prediction accuracy and generalization ability of the trained machine learning model were tested. The results show that: First, the average error between experimental data and prediction results of physical model, PSO-SVM, CNN is 28.82%. Second, the prediction accuracy of the two machine learning models shows little difference, the training time of CNN is much shorter than that of PSO-SVM, but the generalization ability of PSOSVM is stronger than that of CNN. Therefore, SVM is suitable for accurate prediction of small sample data, while CNN is more suitable for learning and prediction of large sample data. This study provides a new efficient method for predicting the maximum leakage rate of supercritical CO2 pipelines.

Key words: machine learning; supercritical CO2; pipeline; leakage; convolutional neural networks; support vector machine
收稿日期: 2023-02-28     
PACS:    
基金资助:重庆市自然科学基金(CYY202010102001) 和福州大学科研启动基金(GXRC-20041) 联合资助
通讯作者: tenglin@fzu.edu.cn
引用本文:   
王一新, 陆诗建, 李卫东, 滕霖. 基于物理模型驱动的机器学习方法预测超临界二氧化碳管道最大泄漏速率. 石油科学通报, 2023, 01: 102-111 WANG Yixin, LU Shijian, LI Weidong, TENG Lin. A physical model driven machine learning for predicting maximum leakage rate in supercritical CO2 release. Petroleum Science Bulletin, 2023, 01: 102-111
链接本文:  
版权所有 2016 《石油科学通报》杂志社