基于GA-BP模型的混合制冷剂天然气液化装置运行参数优化分析  

Operation parameters optimization analysis of mixed refrigerant cycle natural gas liquefaction unit based on GA-BP model

作　　者：刘淼儿[1] 单卫光 朱建鲁[3] 刘金华 李恩道[1] LIU Miaoer;SHAN Weiguang;ZHU Jianlu;LIU Jinhua;LI Endao(CNOOC Gas&Power Group,Beijing 100028,China;North China Municipal Engineering Design&Research Institute Co.,Ltd.,Hefei Branch,Hefei 230061,Anhui,China;College of Pipeline and Civil Engineering,China University of Petroleum(East China),Qingdao 266580,Shandong,China)

机构地区：[1]中海石油气电集团有限责任公司,北京100028 [2]中国市政工程华北设计研究总院有限公司合肥分公司,安徽合肥230061 [3]中国石油大学(华东)储运与建筑工程学院,山东青岛266580

出　　处：《中国海上油气》2023年第2期195-201,共7页China Offshore Oil and Gas

基　　金：国家自然科学基金项目“新型FLNG装置天然气带压液化复杂流动与换热的基础问题(编号:U21B2085)”部分研究成果。

摘　　要：天然气液化装置的制造及施工通常无法与设计参数保持完全一致,这些偏差的累积可能导致装置实际运行状态偏离于设计的最优工况。本文根据混合制冷剂天然气液化装置的实际历史运行数据,通过选择合理的节点,建立了基于BP神经网络的天然气液化装置比功耗预测模型;为了提高模型预测的准确性,利用遗传算法(GA)对神经网络的初始权值进行优化,形成了GA-BP模型。根据GA-BP模型,以天然气液化装置的比功耗最小化为目标,对该装置的主要操作参数进行了优化。结果表明,与原设计最优工况下的实测结果相比,采用优化后参数运行的混合制冷剂天然气液化装置比功耗比减小了4.3%,提高了整个装置的运行效率。本文研究的基于GA-BP模型的优化方法可为天然气液化装置的工艺设计与高效运行提供参考。The manufacture and construction of natural gas liquefaction unit are usually not completely consistent with the design parameters.The accumulation of such errors often leads to the difference between the actual operation and the initial design requirements.According to the historical operation data of mixed refrigerant cycle natural gas liquefaction unit,a prediction model of specific power consumption of natural gas liquefaction unit based on BP neural network was established by selecting reasonable nodes.In order to improve the accuracy of the model prediction,genetic algorithm is used to optimize the initial weights of the neural network,so GA-BP model is obtained.On this basis,the operation parameters are optimized to minimize the specific power consumption of the natural gas liquefaction unit.The results show that the specific power consumption of the optimized mixed refrigerant cycle natural gas liquefaction unit is reduced by 4.3%compared to the measured result under the original design case,and the operation efficiency of the whole unit is improved.The optimization method based on GA-BP model in this paper can provide reference for the process design and efficient operation of natural gas liquefaction unit.

关 键 词：天然气 液化装置 混合制冷剂 BP神经网络 遗传算法 参数优化 

分 类 号：TE646[石油与天然气工程—油气加工工程] TE832