Coal energy consumption forecasting research based on ARIMA-LSTM model:A case study of Henan Province

SONG Kunpeng; SONG Yakai

Abstract：Coal is expected to maintain a significant role in Henan Province's modem energy system for the foreseeable future.Conducting a coal consumption demand forecasting study for a large coal producing and consuming province like Henan Province is conducive to the management of geological survey projects,formulation of development and utilisation policies,and adjustment of energy structure at the provincial level.Combined with Granger causality test and grey correlation analysis model,this study constructed an ARIMA-LSTM model based on the ARIMA model by selecting 2 indicators such as value added of industry and total energy consumption with the residuals of the prediction of the ARIMA model as inputs to the LSTM model.Results demonstrate a notable performance improvement with the ARIMA-LSTM model compared to the standalone ARIMA model.The forecasted coal energy consumption for Henan Province in 2023 is approximately 15,130.36 ten thousand tons of standard coal.This indicates a decrease from 2022 but remains high.Due to limitations in publicly available data,this study only focuses on assessing model applicability.It suggests that relevant departments should continue increasing reserves and production.To address this,it is recommended to adjust the energy structure,conduct more timely coal consumption forecasting,and prepare coal reserves early to avoid coal shortages and power restrictions.