摘要
为了助力国家石油天然气管网集团有限公司“五化一创”(标准化设计、集约化采购、机械化施工、数字化交付、智能化运营、创新引领)工作的顺利开展,采用文献调查法对近十年来天然气资源消费需求预测技术的原理架构、应用实例进行了横向对比与研究综述。研究结果表明:①在当前阶段应用到天然气需求预测领域的常见方法中,以经验预测类、市场调查类、时间序列预测类、相关关系预测类、组合模型预测法为主;②各类方法由于运行机制与数据来源的差异,适用于不同的项目背景;③经验预测法、市场调查法更适用于发展趋势的大致判断,时间序列预测法与相关关系预测法能够更好地揭示模型内各元素间的耦合规律,组合模型预测法能满足多场景预测需求。结论认为:各类预测模型存在自身的优势与不足,组合模型预测方法的实现有利于涵盖更多的输入信息,减小单一算法建模带来的预测误差,并在数字化技术的引领下进一步提升拟合优度,为管理者的合理决策与优化规划提供更多参考。
In order to help PipeChina smoothly implement the"Five Optimizations and One Innovation"(including normalized design,intensive procurement,mechanized construction,digitized delivery,intelligent operation and innovative leadership),both principal framework and practices were studied for advanced technologies to predict the consumption demand on natural-gas resources in the recent decade by means of literature survey.Results show that(i)at present,the commonly used prediction methods mainly include four kinds of prediction,like empiricism,market investigation,time series,correlation,and combination model;(ii)these methods are available for different project background due to their difference in running mechanisms and data source;and(iii)both empirical and market investigation prediction methods are more suitable for judging rough development trend,while both time series and correlation prediction ones can better reveal coupling laws among model elements.And combination model is able to meet the need of multi-scenario prediction.In conclusion,prediction models have their own advantages and disadvantages,and the combination model is conducive to capturing further input information,reducing prediction errors caused by single modeling,and further improving fitting superiority under the guidance of digital technologies,so as to provide additional reference with rational decision-making and optimal planning.
作者
张月
张书勇
ZHANG Yue;ZHANG Shuyong(PipeChina Engineering Technology Innovation Co.,Ltd.,Tianjin 300450,China)
出处
《天然气技术与经济》
2024年第5期43-50,共8页
Natural Gas Technology and Economy
关键词
天然气
资源市场
消费需求
预测方法
对标分析
组合模型
Natural gas
Resource market
Consumption demand
Prediction method
Benchmarking analysis
Combination model