Personalized federated learning for buildings energy consumption forecasting

被引:2
作者
Wang, Rui [1 ]
Bai, Ling [1 ]
Rayhana, Rakiba [1 ]
Liu, Zheng [1 ]
机构
[1] Univ British Columbia, Sch Engn, Kelowna, BC V1V 1V7, Canada
关键词
Federated learning; Building energy management; Deep learning; Mixture-of-experts;
D O I
10.1016/j.enbuild.2024.114762
中图分类号
TU [建筑科学];
学科分类号
0813 ;
摘要
Buildings' energy consumption forecasting is critical for energy saving and building maintenance. However, most studies only focus on centralized learning of one dataset, which ignores the data privacy and data shortage issue. Meanwhile, the difference in energy data distributions from many buildings causes difficulties in training a good machine learning model. Although these two challenges of data privacy and data heterogeneity could be resolved through personalized federated learning algorithms to some degree, there is still a lack of investigation into applying these algorithms to building energy data analytics. Besides using existing personalized federated learning algorithms, we design a new deep learning model through a mixture of experts to support personalization for heterogeneous data distribution. This new design is the first trial to tackle the data heterogeneity through ensemble architecture in federated load forecasting. Extensive experiments are conducted to evaluate the effectiveness of our proposed model with different training algorithms. The results show that our proposed method outperforms other state-of-the-art models in energy forecasting accuracies by 10% to 40% across the buildings' energy data from university campuses.
引用
收藏
页数:14
相关论文
共 50 条
[41]   Hierarchical Personalized Federated Learning for User Modeling [J].
Wu, Jinze ;
Liu, Qi ;
Huang, Zhenya ;
Ning, Yuting ;
Wang, Hao ;
Chen, Enhong ;
Yi, Jinfeng ;
Zhou, Bowen .
PROCEEDINGS OF THE WORLD WIDE WEB CONFERENCE 2021 (WWW 2021), 2021, :957-968
[42]   Personalized Graph Federated Learning With Differential Privacy [J].
Gauthier F. ;
Gogineni V.C. ;
Werner S. ;
Huang Y.-F. ;
Kuh A. .
IEEE Transactions on Signal and Information Processing over Networks, 2023, 9 :736-749
[43]   Multiserver Federated Learning for Building Energy Prediction With Wind Speed [J].
Wang, Rui ;
Fan, Yanshuo ;
Yun, Honguang ;
Rayhana, Rakiba ;
Bai, Ling ;
Liu, Zheng .
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, 2025, 74
[44]   Robust Personalized Federated Learning with Sparse Penalization [J].
Liu, Weidong ;
Mao, Xiaojun ;
Zhang, Xiaofei ;
Zhang, Xin .
JOURNAL OF THE AMERICAN STATISTICAL ASSOCIATION, 2024, :266-277
[45]   Federated Learning for Personalized Image Aesthetics Assessment [J].
Xiong, Zhiwei ;
Yu, Han ;
Shen, Zhiqi .
2023 IEEE INTERNATIONAL CONFERENCE ON MULTIMEDIA AND EXPO, ICME, 2023, :336-341
[46]   Energy Consumption Forecasting in Korea Using Machine Learning Algorithms [J].
Shin, Sun-Youn ;
Woo, Han-Gyun .
ENERGIES, 2022, 15 (13)
[47]   Residential energy consumption forecasting using deep learning models [J].
Ramos, Paulo Vitor B. ;
Villela, Saulo Moraes ;
Silva, Walquiria N. ;
Dias, Bruno H. .
APPLIED ENERGY, 2023, 350
[48]   Energy-Aware Edge Association for Cluster-Based Personalized Federated Learning [J].
Li, Yixuan ;
Qin, Xiaoqi ;
Chen, Hao ;
Han, Kaifeng ;
Zhang, Ping .
IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, 2022, 71 (06) :6756-6761
[49]   Residential Load Forecasting Using Modified Federated Learning Algorithm [J].
Park, Keon-Jun ;
Son, Sung-Yong .
IEEE ACCESS, 2023, 11 :40675-40691
[50]   A solar forecasting framework based on federated learning and distributed computing [J].
Wen, Haoran ;
Du, Yang ;
Lim, Eng Gee ;
Wen, Huiqing ;
Yan, Ke ;
Li, Xingshuo ;
Jiang, Lin .
BUILDING AND ENVIRONMENT, 2022, 225