极端学习机算法下甲醇价格的短期与长期预测研究  

On Short-term and Long-term Prediction of Methanol Price Based on Extreme Learning Machine Algorithm

作　　者：裴钦 雷昭[2] Pei Qin;Lei Zhao(School of Finance and Economics,Wanjian University of Technology,Ma’anshan,Anhui 243031,China;School of Chemistry and Chemical Engineering,Anhui University of Technology,Ma’anshan,Anhui 243032,China)

机构地区：[1]皖江工学院财经学院,安徽马鞍山243031 [2]安徽工业大学化工学院,安徽马鞍山243032

出　　处：《黑龙江工业学院学报（综合版）》2022年第1期127-132,共6页Journal of Heilongjiang University of Technology(Comprehensive Edition)

基　　金：安徽高校人文社会科学研究重点项目“基于熵协同原理的甲醇价格预测模型研究”(项目编号:SK2019A1120)。

摘　　要：基于甲醇价格影响因素的分析,采用极端学习机算法,提出甲醇价格的短期与长期预测等问题,通过构建短期和长期预测模型,分析甲醇价格和进口量、煤炭价格、出口量、进口均价、月产量、出口均价之间的关系,发现甲醇价格与甲醇进口量呈负相关,与出口量、进口均价、出口均价呈线性相关;与其他影响因素相比,与煤炭价格、月产量相关性相对较低。短期预测表明,甲醇价格实际值和预测值之间绝对误差在44元左右,相比于甲醇平均价格2750元/吨,预测精度较好。长期预测表明,甲醇价格实际值和预测值之间平均相对误差约为1.6%,预测精度较好。Based on the analysis of influencing factors of methanol price,using extreme learning machine algorithm,this paper puts forward the short-term and long-term prediction of methanol price.By constructing short-term and long-term prediction models,the relationship between methanol price and import volume,coal price,export volume,average import price,monthly output and average export price was analyzed.The research shows that methanol price is negatively correlated with methanol import volume,and linearly correlated with export volume,average import price and average export price.Compared with other influencing factors,it has a relatively low correlation with monthly output and coal price.The short-term prediction shows that the absolute error between the actual value and the predicted value of methanol price is about 44 yuan.Compared with the average methanol price of 2750 yuan/ton,the prediction accuracy is higher.The long-term prediction shows that the average relative error between the actual value and the predicted value of methanol price is about 1.6%,and the prediction accuracy is higher.

关 键 词：甲醇 预测模型 极端学习机算法 精度 短期和长期 

分 类 号：F426[经济管理—产业经济] F767