PDF下载

[1]杜渐,郑坚钦,夏玉恒,等.耦合混油发展机理与数据修正的成品油管道混油浓度预测[J].油气储运,2024,43(07):1-18.
　Du Jian,Zheng Jianqin,Xia Yuheng,et al.Coupling mechanism constraints and data correction for predicting mixed oil concentration distribution[J].Oil & Gas Storage and Transportation,2024,43(07):1-18.

点击复制

耦合混油发展机理与数据修正的成品油管道混油浓度预测

《油气储运》[ISSN:1000-8241/CN:13-1093/TE] 卷: 43 期数: 2024年07期页码: 1-18 栏目: 出版日期: 2024-07-25

Title:: Coupling mechanism constraints and data correction for predicting mixed oil concentration distribution

作者:: 杜渐; 郑坚钦; 夏玉恒; 张秀玲; 徐宁; 廖绮; 涂仁福; 梁永图

Author(s):: Du Jian; Zheng Jianqin; Xia Yuheng; Zhang Xiuling; Xu Ning; Liao Qi; Tu Renfu; Liang Yongtu

关键词:: 成品油管道; 混油浓度预测; 物理信息神经网络; 基本控制方程; 滚动预测

Keywords:: Multi-product pipeline; Mixed oil concentration prediction; Physics-informed neural network; Governing equation; Rolling prediction

分类号:: TE832

文献标志码:: A

摘要:: [目的]顺序输送成品油管道不可避免会在相邻油品交界处产生混油段，影响站场调度计划及分输方案制定，准确预测混油段浓度分布是提高站场混油控制水平、降低混油处理能耗、防止油品质量事故的重要依据。然而，多维混油数值模型计算时间过长、难以应用于长输管道，传统数据驱动方法忽略混油发展机理过程，结果可能违背物理原理，准确度及可解释性差。[方法]通过分析混油发展机理，深入挖掘混油发展遵循的基本控制方程及初始边界条件，通过自动微分方式与深度学习模型耦合以构造物理指导的耦合损失函数，将模型预测结果约束至混油发展所对应的物理解空间内；再利用短管道初始数值解，基于管道递推滚动预测和数据修正方法，预测长输管道混油浓度分布。[结果]数值案例表明，所建模型相较传统数据驱动方法准确度更高，MAPE降低了91%，对数据依赖程度更低，RMSE波动程度随数据量变化降低了60%；相较FLUENT数值仿真方法，计算成本降低至原先的12%；现场工程化应用实例表明，相较Taylor模型及改进Taylor模型，所提出的工程化应用框架的绝对误差分别降低了71%、58%，可实现长输管道混油浓度分布高效求解。[结论]耦合混油发展机理与数据修正的成品油管道混油浓度预测方法可准确快速预测长输管道混油浓度分布，指导站场混油接收方案制定，有效提高混油控制智能化水平。

Abstract:: The multi-product pipelines with sequential transportation inevitably form mixed oil sections between adjacent oil products. Mixed oil affects the scheduling plan and distribution plan of the station. Accurately predicting the mixed oil concentration distribution is an important basis for improving the control level of mixed oil in the station, reducing energy consumption in mixed oil treatment, and preventing oil quality accidents However, the multi-dimensional mixed oil numerical models require long calculation time and is difficult to be applied to long-distance pipelines. Traditional data-driven methods ignore the development mechanism of mixed oil, and the results may violate physical principles, resulting in poor accuracy and interpretability. Therefore, this paper analyzes the mechanism of mixed oil concentration development, delves into the basic control equations and initial boundary conditions followed by mixed oil development, and constructs a physically guided coupling loss function by coupling with deep learning models through automatic differentiation. The model prediction results are constrained to the corresponding physical space of mixed oil development. Afterward, the initial numerical solution of the short pipeline was used to predict the mixed oil concentration distribution in long-distance pipelines based on pipeline recursive rolling prediction and data correction methods. Numerical cases indicate that the proposed model has higher accuracy than traditional data-driven methods, and the relative error is reduced by 91%, Besides, the proposed model achieves a reduction in dependency on available data, and the degree of performance fluctuation has decreased by 60% with changes in data volume. Compared to numerical calculations, the calculation time of the proposed model has been reduced to 12% of the original time. Industrial pipeline cases suggest that the engineering application framework proposed in this paper reduces the error by 71% and 58% compared with the Taylor model and its improved model, and can achieve efficient solutions for mixed oil concentration distribution in long-distance pipelines. Therefore, by coupling the inherent mechanism of coupled mixed oil development to the data model, utilizing a small number of short pipeline numerical results, and through pipeline recursive rolling prediction and data correction, the distribution of mixed oil concentration in long-distance pipelines can be accurately and quickly predicted, guiding the formulation of mixed oil reception plans in stations and effectively improving the intelligent level of mixed oil control.

相似文献/References:

[1]徐广丽,张国忠,赵仕浩.管道低洼处积水排除实验[J].油气储运,2011,30(05):369.[doi:10.6047/j.issn.1000-8241.2011.05.014]
　Xu Guangli,Zhang Guozhong,Zhao Shihao.An experiment on dewatering for lower location of pipeline[J].Oil & Gas Storage and Transportation,2011,30(07):369.[doi:10.6047/j.issn.1000-8241.2011.05.014]
[2]戴福俊.成品油顺序输送混油的检测与切割[J].油气储运,2011,30(04):255.[doi:10.6047/j.issn.1000-8241.2011.04.004]
　Dai Fujun.Detection and cutting technique of oil contamination in batch transportation process[J].Oil & Gas Storage and Transportation,2011,30(07):255.[doi:10.6047/j.issn.1000-8241.2011.04.004]
[3]郭祎,张道全,刘鹏,等.西部成品油管道末站混油切割改进措施[J].油气储运,2011,30(07):520.[doi:10.6047/j.issn.1000-8241.2011.07.011]
　Guo Yi,Zhang Daoquan,Liu Peng,et al.Improvement of mixed oil cutting of Western Oil Product Pipeline Terminal[J].Oil & Gas Storage and Transportation,2011,30(07):520.[doi:10.6047/j.issn.1000-8241.2011.07.011]
[4]张艳彩,李珍,李绍文.成品油管道工程经济的评价问题[J].油气储运,2011,30(07):550.[doi:10.6047/j.issn.1000-8241.2011.07.022]
　Zhang Yancai,Li Zhen,Li Shaowen.Issues of economic assessment in products oil pipeline project[J].Oil & Gas Storage and Transportation,2011,30(07):550.[doi:10.6047/j.issn.1000-8241.2011.07.022]
[5]李会朵,梁静华,朱峰,等.成品油混油掺混实验与临界比例计算[J].油气储运,2011,30(03):180.[doi:10.6047/j.issn.1000-8241.2011.03.006]
　Li Huiduo,Liang Jinghua,Zhu Feng,et al.Mixing experiment and critical proportion calculation for contamination of products[J].Oil & Gas Storage and Transportation,2011,30(07):180.[doi:10.6047/j.issn.1000-8241.2011.03.006]
[6]牛迎战,刘甫,林猛,等.成品油长输管道盗油点抢修方法的应用[J].油气储运,2011,30(02):150.[doi:10.6047/j.issn.1000-8241.2011.02.019]
　Niu Yingzhan,Liu Fu,Lin Meng,et al.Application of the emergency repair method in leak point of products pipeline[J].Oil & Gas Storage and Transportation,2011,30(07):150.[doi:10.6047/j.issn.1000-8241.2011.02.019]
[7]林猛,牛迎战,张玲.EP减阻剂在抚顺－鲅鱼圈成品油管道的应用[J].油气储运,2011,30(05):350.[doi:10.6047/j.issn.1000-8241.2011.05.008]
　Lin Meng,Niu Yingzhan,Zhang Ling.Application of EP DRA in Fushun-Bayuquan Products Pipeline[J].Oil & Gas Storage and Transportation,2011,30(07):350.[doi:10.6047/j.issn.1000-8241.2011.05.008]
[8]朱峰,梁静华,李会朵,等.成品油混油切割与掺混试验[J].油气储运,2011,30(05):364.[doi:10.6047/j.issn.1000-8241.2001.05.013]
　Zhu Feng,Liang Jinghua,Li Huiduo,et al.Testing on contamination cut and blending for products[J].Oil & Gas Storage and Transportation,2011,30(07):364.[doi:10.6047/j.issn.1000-8241.2001.05.013]
[9]宋飞,朱峰,潘红梅,等.兰成渝管道杂质来源分析[J].油气储运,2010,29(5):381.[doi:10.6047/j.issn.1000-8241.2010.05.021]
　Song Fei,Zhu Feng,Pan Hongmei.Analysis on lmpurity Origin of Lanzhou-Chengdu-Chongqing Oil Product Pipeline[J].Oil & Gas Storage and Transportation,2010,29(07):381.[doi:10.6047/j.issn.1000-8241.2010.05.021]
[10]马红娜,李彦娥,武征.成品油管道泄漏的环境风险评价[J].油气储运,2011,30(11):801.[doi:10.6047/j.issn.1000-8241.2011.11.001]
　Ma Hongna,Li Yane,Wu Zheng.Environmental risk evaluation of leaked products pipeline[J].Oil & Gas Storage and Transportation,2011,30(07):801.[doi:10.6047/j.issn.1000-8241.2011.11.001]
[11]杜渐,郑坚钦,夏玉恒,等.耦合混油发展机理与数据修正的成品油管道混油浓度预测[J].油气储运,2024,43(07):796.[doi:10.6047/j.issn.1000-8241.2024.07.009]
　DU Jian,ZHENG Jianqin,XIA Yuheng,et al.Prediction of mixed oil concentration in product oil pipeline coupling oil mixing mechanism and data correction[J].Oil & Gas Storage and Transportation,2024,43(07):796.[doi:10.6047/j.issn.1000-8241.2024.07.009]

更新日期/Last Update: 2024-04-12