基于CCA-ELM模型的国产LNG出厂价格中短期预测研究——以陕西省为例
Short-and medium-term forecast of domestic LNG ex-factory prices based on CCA-ELM model—Taking Shaanxi province as an example
潘凯 1谢翔 1张曦 1刘定智 1张晗 1张元涛 1邓钰暄 2贺美 2李慧慧 2孙仁金2
作者信息
- 1. 中国石油规划总院
- 2. 中国石油大学[北京]经济管理学院
- 折叠
摘要
考虑供需基本面因素和非基本面因素,构建CCA-ELM模型用于国产LNG出厂价格的预测.供需基本面因素包括LNG的产量、销量、库存、气温以及原料气成本,非基本面影响因素包括原油、汽油、柴油、煤炭等替代能源价格与东北亚天然气现货价格.通过典型相关性分析,研究各个影响因素对价格的作用程度.以10个影响因素的周度数据为研究对象,以LNG出厂价格的历史序列与其影响因素构建CCA-ELM神经网络预测模型.10个影响因素整体与LNG出厂价格的相关性较强,中国LNG出厂价格受能源市场的影响程度较高,受供需基本面的影响程度较低.兼顾LNG出厂价格历史数据与影响因素的CCA-ELM模型有效改进了时间序列神经网络的预测方法,提高了预测精度.
Abstract
This paper constructs a CCA-ELM model for LNG ex-factory price forecast,considering supply and demand fundamentals and non-fundamental factors.Supply and demand fundamentals include LNG production,sales volume,inventory,air temperature,and feedstock gas cost,while such alternative energy prices as crude oil,gasoline,diesel fuel,coal,as well as Northeast Asian natural gas spot prices belongs to non-fundamental influences factors.It conducts the typical correlation analysis to examine the extent to which each influencing factor contributes to the price.A CCA-ELM neural network prediction model is constructed using the weekly data of the 10 influencing factors,the historical series of LNG ex-factory prices and their influencing factors.10 influencing factors are strongly correlated with LNG ex-factory prices,LNG ex-factory prices in China are more influenced by the energy market and less influenced by supply and demand fundamentals,and the CCA-ELM model,which takes into account the historical data of LNG ex-factory price and the influencing factors,effectively improves the prediction method of time series neural networks as well as increases the prediction accuracy.
关键词
LNG出厂价格/影响因素/ELM神经网络/典型相关分析Key words
LNG ex-factory price/influence factor/ELM neural network/canonical correlation analysis引用本文复制引用
基金项目
中国石油天然气集团有限公司院所基金课题(KJ2024-009)
出版年
2024
国家科技期刊平台
NSTL
万方数据