摘要
选取秦皇岛港口动力煤价格作为研究对象,搜集10年间煤价数据并分析其影响因素,确定煤炭产量、港口库存、运输成本、火力发电量及社会用电量为主要影响因素;分别建立ARIMA(2,1,2)模型和RF(随机森林)模型并优化,通过加权平均法得到ARIMA和RF模型权重,建立ARIMA-RF组合模型。该模型较深度神经网络模型(DNN)、支持向量回归模型(SVR)、ARIMA模型、RF模型预测的煤价准确度更高,可准确预测动力煤价格走势,为调控能源消费强度、深化能源体制机制改革政策制定提供参考。
This paper selects the thermal coal price of Qinhuangdao port as the research object,collects coal price data over the past decade and analyzes its influencing factors,the main influencing factors are determined to be coal production,port inventory,transportation costs,thermal power generation,and social electricity consumption.The ARIMA(2,1,2)model and RF model are established and optimized respectively.The weights of ARIMA and RF models are obtained through the weighted average method and the combined model of ARIMA and RF(Random Forest)is established.Compared to deep neural network models(DNN),support vector regression models(SVR),ARIMA models,and RF models,this model has higher accuracy in predicting coal prices.This model can accurately predict the thermal coal price,proving reference for regulating energy consumption intensity,deepening energy system and mechanism reform,and policy-making.
作者
季凌雲
陆伟
卓辉
李佐健
Ji Lingyun;Lu Wei;Zhuo Hui;Li Zuojian(College of Safety Science and Engineering,Anhui University of Science and Technology,Huainan 232001,China)
出处
《煤炭经济研究》
2023年第4期28-37,共10页
Coal Economic Research
基金
安徽省高校优秀科研创新团队项目(2022AH010051)
国家自然科学基金(51974178)
安徽省重点研究与开发计划项目(2022m07020006)。
