工业数据驱动技术在大型复杂天然气管网运行中的应用  

Application of industrial data-driven technology in the operation of large-scale complex natural gas pipeline network

作　　者：唐善华 杨金辉 徐春野[1] 杨毅 周晓莹 郭永华 刁洪涛 张子涛 TANG Shanhua;YANG Jinhui;XU Chunye;YANG Yi;ZHOU Xiaoying;GUO Yonghua;DIAO Hongtao;ZHANG Zitao(PipeChina Oil&Gas Control Center,Beijing 100013,China;School of Energy&Resource,China University of Geosciences,Beijing 100083,China)

机构地区：[1]国家石油天然气管网集团有限公司油气调控中心,北京市100013 [2]中国地质大学(北京)能源学院

出　　处：《天然气工业》2021年第9期135-141,共7页Natural Gas Industry

摘　　要：大型复杂天然气管网运行数据分析是一类典型的工业大数据驱动技术的应用场景。工业大数据实时性高、数据量大、价值密度低、数据源异构性强,通用的纯数据驱动技术手段无法解决天然气管网运行业务中的特定问题。为了给大型复杂天然气管网的运行提供决策支持,国家石油天然气管网集团有限公司(以下简称国家管网)油气调控中心作为全国天然气管网的运行中枢,在大型数据应用的探索中将先进数据处理技术、深度学习算法与天然气管网运行实际深度结合,基于业务逻辑定制应用功能,充分利用已有的自动化数据采集和关键设备远程控制技术,在天然气管网运行业务中实现了数据驱动技术的应用。研究结果表明:①在数据采集层,通过前置性数据治理算法实现了多源时序数据结构化采集,建立起字段对称的关系型数据库,实现了自动化数据采集、高效的数据管理;②在数据分析层,根据深刻的业务逻辑个性化应用机器学习算法,紧密结合运行需求开发出一套可实现短周期消费量预测、实时进出平衡分析、运行方案推荐的智能分析工具;③在数据服务应用层,基于Echarts开源可视化图表库,对多层次维度管网运行数据实现交互式的可视化展示,打造数据应用平台提取关键信息,为运行决策提供有力支持。Analyzing the operation data of large-scale complex natural gas pipeline networks is a typical application scenario of industrial big data-driven technology.Industrial big data is characterized by strong real-time,large data volume,low value density and strong data source heterogeneity,and the universal pure data-driven technology fails to solve the particular problems in the operation of natural gas pipeline networks.In order to provide decision-making support for the operation of large-scale complicated natural gas pipeline networks,PipeChina Oil&Gas Control Center,as the operation hub of national natural gas pipeline networks,combines advanced data processing technology and deep learning algorithm with the actual operation of natural gas pipeline networks to explore the application of big data,customize the application functions based on business logic,and make full use of its existing automatic data acquisition technologies and remote control technologies of key equipment,so as to realize the application of data-driven technology in the operation of natural gas pipeline networks.And the following research results are obtained.First,in the data acquisition layer,the structured acquisition of multisource temporal data is realized by means of pre-data governance algorithm,a relational database with symmetric field is established,and automatic data acquisition and efficient data management is realized.Second,in the data analysis layer,the personalized application of machine learning algorithm is realized according to profound business logic,and an intelligent analysis tool that can realize short-cycle consumption prediction,real-time input and output balance analysis and operation scheme proposal is developed based on the operation demand.Third,in the data service application layer,interactive visual display of operation data of multi-level and multi-dimension pipeline network is realized based on the Echarts open-source visual chart base,a data application platform is established and the key information is extra

关 键 词：数据驱动 数字化转型 工业数据 天然气管网 调控运行 机器学习 数据可视化 数据治理 

分 类 号：TE46[石油与天然气工程—油气田开发工程]