冰蓄冷空调系统XGBoost-LSTNet负荷预测模型  

XGBoost-LSTNet for Ice Storage and Cooling Air Conditioning System Load Forecasting Model

作　　者：魏东 何建林 闫畔 WEI Dong;HE Jian-lin;YAN Pan(School of Environment and Energy Engineering,Beijing University of Civil Engineering and Architecture,Beijing 100044,China;Beijing Key Laboratory of Intelligent Processing for Building Big Data,Beijing 100044,China)

机构地区：[1]北京建筑大学电气与信息工程学院,北京100044 [2]北京市建筑大数据智能处理重点实验室,北京100044

出　　处：《计算机仿真》2025年第2期501-508,共8页Computer Simulation

基　　金：北京市属高校高水平创新团队建设计划项目(IDHT20190506);住房城乡建设部科学技术项目(研究开发项目)(No.2019-K-149);北京建筑大学高级主讲教师培育计划(GJZJ20220803)。

摘　　要：建筑冷负荷预测是冰蓄冷空调系统实时控制和优化调度的基础。针对冰蓄冷空调系统冷负荷模型特征选择工作量大、长期负荷预测精度难以提升等问题,提出基于XGBoost(eXtreme Gradient Boosting)-LSTNet(Long-and Short-term Timeseries network)的冰蓄冷空调系统负荷预测方法。利用XGBoost算法选择重要特征,以改善模型精度与泛化能力;基于LSTNet算法中卷积模块对负荷数据中的细粒度特征进行有效提取,利用循环及循环跳跃模块对多元时间序列的短/长期趋势进行有效记忆,以满足系统短期实时控制需求,避免出现冷源供不应求的情况;融入自回归模块提取负荷线性分量,以增强预测结果的稳定性与准确性。针对北京某大厦冰蓄冷空调系统的负荷预测仿真表明,所提出的XGBoost-LSTNet模型相较于其它模型,具有更高的预测精度和泛化能力。Building cold load prediction is the basis for real-time control and optimal scheduling of ice storage and cooling air conditioning systems.To address the problems of large workload in selecting cold load model features and difficulty in improving the long-term load prediction accuracy of ice storage air conditioning system,we propose a load prediction method based on XGBoost(eXtreme Gradient Boosting)-LSTNet(Long-and Short-term Time-series network)for ice storage air conditioning system.The XGBoost algorithm is used to select the important features to improve the model accuracy and generalization ability;based on the convolution module in the LSTNet algorithm,finegrained features are effectively extracted from load data.,The loop and loop jump module is used to effectively memorize the short/long-term trends of multivariate time series to meet the short-term real-time control requirements of the system and to avoid the situation of oversupply of cold sources;the self-regression module to extract the linear components of the load is integrated to enhance the stability and accuracy of the prediction results.The simulation experiments of load prediction for an ice storage air conditioning system in a building in Beijing show that the proposed XGBoost-LSTNet model has higher prediction accuracy and generalization ability compared with other models.

关 键 词：冰蓄冷空调 负荷预测 负荷预测模型 

分 类 号：TU831[建筑科学—供热、供燃气、通风及空调工程] TP391[自动化与计算机技术—计算机应用技术]