电厂燃煤库存ARIMAX-LSTM组合预测方法研究  

作　　者：郭桦 付则开 

机构地区：[1]国家能源集团航运有限公司,北京100080

出　　处：《科技创新与应用》2025年第9期33-36,共4页Technology Innovation and Application

基　　金：国家自然科学基金(62173160)。

摘　　要：为确保燃煤电厂能源稳定供应和控制进煤维护成本,辅助管理人员了解电煤库存情况并采取相应措施,需要对电厂燃煤库存进行短期精准预测。该文提出ARIMAX-LSTM组合预测模型,ARIMAX模型用于预测原始电煤库存时间序列中的线性成分,LSTM模型用于直接预测原始数据中的非线性成分以及ARIMAX模型预测结果与原序列之间的差值非线性成分,最后对预测结果进行误差补偿。以某电厂实际库存数据对该模型进行验证,电厂燃煤10 d库存预测的平均相对误差为10.93%,比其中单一模型具有更高的预测精度,对电煤库存管理具有更强的指导意义。In order to ensure a stable supply of energy to coal-fired power plants and control the maintenance costs of coal-fed coal,auxiliary managers understand the thermal coal inventory situation and take corresponding measures,which requires short-term accurate predictions of coal-fired power plants.This paper proposes an ARIMAX-LSTM combined prediction model.The ARIMAX model is used to predict the linear component in the original thermal coal inventory time series,and the LSTM model is used to directly predict the nonlinear component in the original data and the nonlinear component of the difference between the ARIMAX model prediction results and the original series.Finally,the prediction results are compensated for errors.The model was verified with the actual inventory data of a power plant.The average relative error of the 10-day inventory forecast of coal in the power plant was 10.93%,which has higher prediction accuracy than the single model and has stronger guiding significance for thermal coal inventory management.

关 键 词：组合预测 ARIMAX模型 LSTM模型 电煤库存 误差补偿 

分 类 号：TM621[电气工程—电力系统及自动化]