现代电力2025，Vol.42Issue(1)：129-136,8.DOI:10.19725/j.cnki.1007-2322.2022.0449
基于长短期记忆网络模型的联邦学习居民负荷预测
Resident Load Forecasting Using Federal Learning Based on LSTM Model
摘要
Abstract
Residential electricity consumption constitutes over 15%of the total social electricity consumption,a significant number of users spread across various locations.Accurate pre-diction of the residential load is conducive to integrating de-mand-side resources and meeting the response requirements on the demand side.Most of the existing residential load forecast-ing methods are centralized,requiring extensive data exchange between the server and the client.This results in increased com-munication costs and information security problems.Based on the federal learning framework,the long-term and short-term memory network is utilized to study the residential load fore-casting method.The simulation results indicate that the residen-tial load forecasting based on federal learning outperforms cent-ralized learning in real calculation efficiency.In addition,we compared our algorithm with the FedAvg,FedAdagrad and FedYogi federal learning strategies,and found that the FedAd-agrad federal learning strategy with adaptive optimization func-tion achieves higher accuracy and better convergence in resid-ential load forecasting.关键词
居民用户/集中式/联邦学习/负荷预测/长短期记忆网络Key words
resident users/centralized/federal learning/load forecasting/long and short term memory network(LSTM)分类
信息技术与安全科学引用本文复制引用
朱嵩阳,张歌,贾愉靖,白晓清..基于长短期记忆网络模型的联邦学习居民负荷预测[J].现代电力,2025,42(1):129-136,8.基金项目
国家自然科学基金项目(51967001). Project Supported by National Natural Science Foundation of China(51967001). （51967001）
