基于异步个性化联邦学习的多元用户负荷预测方法

熊小勇; 刘大明

doi:10.19725/j.cnki.1007-2322.2024.0207

欢迎来到能源电力期刊网！

English Version 期刊登录

您当前的位置：

首页 >

文章列表页 >

基于异步个性化联邦学习的多元用户负荷预测方法

智能电网 | 更新时间：2026-10-19

- 基于异步个性化联邦学习的多元用户负荷预测方法
- A Multi-user Load Forecasting Method Based on Asynchronous Personalized Federated Learning
- 现代电力 2026年43卷第5期页码：853-864
- 作者机构：
  
  上海电力大学计算机科学与技术学院, 上海市浦东新区,200120
- 作者简介：
  
  熊小勇(1998)，男，硕士研究生，通信作者，研究方向为联邦学习、电力系统负荷预测，E-mail：1363677096@qq.com
  刘大明(1971)，男，博士，副教授，硕士生导师，研究方向为物联网技术和嵌入式系统，E-mail：ldm@shiep.edu.cn
- 基金信息：
  
  上海市科技计划项目(23010501500)。
- DOI：10.19725/j.cnki.1007-2322.2024.0207
  中图分类号： 5
- 纸质出版：2026
- 稿件说明：
移动端阅览
熊小勇, 刘大明. 基于异步个性化联邦学习的多元用户负荷预测方法[J]. 现代电力, 2026,43(5):853-864.

XIONG Xiaoyong, LIU Daming. A Multi-user Load Forecasting Method Based on Asynchronous Personalized Federated Learning[J]. 2026, 43(5): 853-864.
熊小勇, 刘大明. 基于异步个性化联邦学习的多元用户负荷预测方法[J]. 现代电力, 2026,43(5):853-864. DOI： 10.19725/j.cnki.1007-2322.2024.0207.

XIONG Xiaoyong, LIU Daming. A Multi-user Load Forecasting Method Based on Asynchronous Personalized Federated Learning[J]. 2026, 43(5): 853-864. DOI： 10.19725/j.cnki.1007-2322.2024.0207.

摘要

具有实时、高采样率和高精度的智能电表为多元用户用电负荷预测提供了数据基础。目前，用户级负荷预测仍面临数据隐私泄露、数据样本不足和模型泛化能力弱等挑战。传统的联邦学习方法通过各客户端的协同训练，得到具有多方共性的全局模型，但该模型忽视了多元用户在用电模式和数据分布上的差异，且同步聚合容易受设备掉线、网络延迟的影响。针对上述问题，提出异步个性化联邦学习(asynchronous personalized federated learning，Async-PFL)方法，并在此基础上设计用于短期负荷预测的卷积神经网络–门控循环单元(convolutional neural network-gated recurrent unit，CNN-GRU)混合模型。Async-PFL采用模型分层与局部调优的两阶段个性化方法，有效增强预测模型对不同用户异构数据的适应能力，使模型能够更好地契合多元用户的个性化特征。同时，该方法引入异步联邦聚合算法，结合滞后容忍模型分发机制与时间滞后权重函数，显著缓解因节点丢失和网络延迟对全局模型更新造成的影响，从而加快模型的收敛速度。将所提方法在“建筑数据基因组计划2”数据集上进行算例分析。结果表明，与单一模型及标准联邦学习相比，该方法在确保隐私安全的同时，不仅提升了模型精度，还增强了系统鲁棒性。

Abstract

Smart meters with real-time

high-sampling rate

and high-precision capabilities lay a data foundation for multi-user electricity load forecasting. Currently

user-level load forecasting still faces challenges such as data privacy leakage

insufficient data samples

and weak model generalization ability. Traditional federated learning methods derive a global model with common characteristics across multiple clients through collaborative training. However

these models overlook the differences in electricity consumption patterns and data distributions among diverse users. Additionally

synchronous aggregation is susceptible to device disconnection and network latency issues. To address these issues

an asynchronous personalized federated learning (Async-PFL) method is proposed. Based on this

a hybrid CNN-GRU model is designed for short-term load forecasting. The Async-PFL enhances the adaptability of the predictive model to heterogeneous data from different users through a two-stage personalization method based on model layering and local fine-tuning

thereby making the model better suited to the individual characteristics of diverse users. The asynchronous federated aggregation algorithm

combined with a lag-tolerant model distribution and a time-lag weight function

effectively mitigates the negative impact of node loss and network delays on global model updates

thereby accelerating model convergence. The proposed method is experimentally validated on the "Building Data Genome Project 2" dataset. The results demonstrate that the proposed method not only protects privacy but also outperforms single models and standard federated learning in terms of both model accuracy and system robustness.

关键词

Keywords

references

ZHANG Zhigang, KANG Chongqing. Challenges and prospects for constructing the new-type power system towards carbon neutrality future[J]. Proceedings of the CSEE, 2022, 42(08): 2806−2819(in Chinese).

KONG Xiangyu, MA Yuying, AI Qian, et al . Review on modeling of power consumption characteristics and load forecasting for multi-users in new power systems[J ] . Automation of Electric Power Systems, 2023, 47(13): 2−17(in Chinese).

HAN Fujia, WANG Xiaohui, QIAO Ji, et al . Review on Artificial intelligence based load forecasting research for the new-type power system[J ] . Proceedings of the CSEE, 2023, 43(22): 8569−8592(in Chinese).

CHENG Xiaoming, WANG Lei, ZHANG Pengchao, et al . Data characteristics and short-term forecasting of regional power load[J ] . Power System Technology, 2022, 46(03): 1092−1099(in Chinese).

TAN Mao, YUAN Siping, LI Shuaihu, et al . Ultra-short-term industrial power demand forecasting using LSTM based hybrid ensemble learning[J ] . IEEE Transactions on Power Systems, 2020, 35(4): 2937−2948.

LI Jian, DE NG Daiyu, ZHAO Junbo, et al . A novel hybrid short-term load forecasting method of smart grid using MLR and LSTM neural network[J ] . IEEE Transactions on Industrial Informatics, 2021, 17(4): 2443−2452.

WANG Jinsong, CHEN Xuhui, ZHANG Fan, et al . Building load forecasting using deep neural network with efficient feature fusion[J ] . Journal of Modern Power Systems and Clean Energy, 2021, 9(1): 160−169.

ZANG Haixiang, CHEN Yuwei, CHENG Lilin, et al . User level ultra-short-term load forecasting based on multi-scale component feature learning[J ] . Power System Technology, 2024, 48(06): 2584−2592(in Chinese).

BANABILAH S, ALOQAILY M, ALSAYED E, et al . Federated learning review: Fundamentals, enabling technologies, and future applications[J ] . Information Processing Management, 2022, 59(6): 103061.

MCMAHAN B, MOORE E, RAMAGE D, et al . Communication-efficient learning of deep networks from decentralized data[C ] //Artificial Intelligence and Statistics. PMLR, 2017: 1273−1282.

WANG Beibei, ZHU Jing, WANG Jiale, et al . Federated-learning based industry load forecasting framework under privacy protection of meter data[J ] . Automation of Electric Power Systems, 2023, 47(13): 86−93(in Chinese).

QIN D, WANG C, WEN Q, et al . Personalized federated darts for electricity load forecasting of individual buildings[J ] . IEEE Transactions on Smart Grid, 2023, 14(6): 4888−4901.

TAÏK A, CHERKAOUI S. Electrical load forecasting using edge computing and federated learning[C]//ICC 2020-2020 IEEE International Conference on Communications (ICC). IEEE, 2020: 1−6.

PETRANGELI E, TONELLOTTO N, VALLATI C. Performance evaluation of federated learning for residential energy forecasting[J]. IoT, 2022, 3(3): 381−397.

WANG Rui, YUN Honggang, RAYHANA R, et al . An adaptive federated learning system for community building energy load forecasting and anomaly prediction[J ] . Energy and Buildings, 2023, 295: 113215.

ZHU Songyang, ZHANG Ge, JIA Yujing, et al . Resident load forecasting using federal learning based on LSTM model[J ] . Modern Electric Power, 2025, 42(01): 129−136(in Chinese).

TAN A Z, YU H, CUI L, et al . Towards personalized federated learning[J ] . IEEE Transactions on Neural Networks and Learning Systems, 2022, 34(12): 9587−9603.

COLLINS L, HASSANI H, MOKHTARI A, et al . Fedavg with fine tuning: Local updates lead to representation learning[J ] . Advances in Neural Information Processing Systems, 2022, 35: 10572−10586.

DINH C T, TRAN N H, NGUYEN T D. Personalized federated learning with Moreau envelopes[J]. Advances in Neural Information Processing Systems, 2020, 33: 21394−21405.

XU C, QU Y, XIANG Y, et al . Asynchronous federated learning on heterogeneous devices: A survey[J ] . Computer Science Review, 2023, 50: 100595.

ZHOU C, TIAN H, ZHANG H, et al . TEA-fed: time-efficient asynchronous federated learning for edge computing[C ] //Proceedings of the 18th ACM international conference on computing frontiers. 2021: 30−37.

CHEN S, WANG X, ZHOU P, et al . Heterogeneous semi-asynchronous federated learning in internet of things: A multi-armed bandit approach[J ] . IEEE Transactions on Emerging Topics in Computational Intelligence, 2022, 6(5): 1113−1124.

浏览量

289

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

基于K-means聚类与LSTM模型的多能源耦合电力负荷预测

基于长短期记忆网络模型的联邦学习居民负荷预测

面向用户侧不同用电群体的低碳运行用电策略

一种改进Transformer的电力负荷预测方法

基于变分模态分解的卷积神经网络−双向门控循环单元−多元线性回归多频组合短期电力负荷预测