基于堆叠集成学习的住宅单位面积能耗预测模型

Journal: Engineering and Management Science DOI: 10.12238/ems.v8i2.18463

陈子涵

同济大学经济与管理学院

Abstract

为解决住宅建筑能耗预测中单一模型稳定性不足的问题,本文提出了一种基于堆叠集成学习的住宅单位面积能耗预测方法。以美国住宅能源消费调查(RECS)2020年数据为研究对象,通过系统化的特征工程,采用Spearman相关性与LightGBM-SHAP的两阶段特征选择策略,构建了包含LightGBM、XGBoost、CatBoost和神经网络的异质基学习器组合,并以支持向量回归等模型作为元学习器进行二次融合。实验结果表明,以GBDT为元学习器的堆叠集成模型在源域测试中达到R2=0.6531、MAE=0.5450,较最优单一模型分别提升3.3%和1.9%,证明了堆叠集成策略在提升预测精度和稳健性方面的有效性。本研究为住宅能源管理和节能决策提供了具有工程适用性的技术路径。

Keywords

住宅能耗预测;单位面积能耗;堆叠集成学习

References

[1] YE Y,ZUO W,WANG G. A comprehensive review of energy-related data for U.S. commercial buildings [J]. Energy Build,2019,186:126-37.
[2] KUMAR MOHAPATRA S,MISHRA S,TRIPATHY H K,et al. A sustainable data-driven energy consumption assessment model for building infrastructures in resource constraint environment [J]. Sustainable Energy Technologies and Assessments,2022,53:102697.
[3] WANG Z,WANG Y,SRINIVASAN R S. A novel ensemble learning approach to support building energy use prediction [J]. Energy Build,2018,159:109-22.
[4] UIDHIR T M,ROGAN F,COLLINS M,et al. Improving energy savings from a residential retrofit policy:A new model to inform better retrofit decisions [J]. Energy Build,2020,209:109656.
[5] DAI Z,HUANG W. Improving energy management practices through accurate building energy consumption prediction:analyzing the performance of LightGBM,RF,and XGBoost models with advanced optimization strategies [J]. Electrical Engineering,2025.

Copyright © 2026 陈子涵

Creative Commons License
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License