南京信息工程大学学报2026，Vol.18Issue(3)：321-330,10.DOI:10.13878/j.cnki.jnuist.20250117001
基于联邦学习和DAL策略的电力负荷预测
Electric load forecasting based on federated learning and DAL strategy
摘要
Abstract
Electric load forecasting is a core foundation for power grid planning and operation.However,traditional methods rely on training models with data from a single region,resulting in a significant decline in generalization ca-pability when applied to cross-regional forecasting.To address this issue,this paper proposes a hybrid model that integrates a Temporal Convolutional Network(TCN),Long Short-Term Memory(LSTM),and an attention mecha-nism(TCN-LSTMs-Attention),combined with a Decentralized Aggregation Learning(DAL)strategy.In this framework,multiple sub-models are jointly trained to obtain a global model capable of cross-regional forecasting by sequentially training with data from different regions on the same server.Moreover,the proposed method incorpo-rates a dynamic learning rate halving and parameter resetting mechanism to further accelerate model convergence.Experiments based on the dataset from the 9th China Electrical Engineering Cup Competition demonstrate that,com-pared to models trained independently on regional data,the proposed method improves Mean Squared Error(MSE),Mean Absolute Error(MAE),Root Mean Squared Error(RMSE),and Mean Absolute Percentage Error(MAPE)by 43.0％,29.5％,24.4％,and 35.4％,respectively,in cross-regional forecasting tasks.These results validate the model's robustness and engineering practicality in high-heterogeneity load scenarios.关键词
电力负荷预测/长短期记忆网络/去中心化聚合学习/跨区域预测Key words
electric load forecasting/long short-term memory(LSTM)/decentralized aggregation learning(DAL)/cross-regional forecasting分类
信息技术与安全科学引用本文复制引用
周聪,李明,袁隆发,丁南威,铁瑞君,曾蒸..基于联邦学习和DAL策略的电力负荷预测[J].南京信息工程大学学报,2026,18(3):321-330,10.基金项目
重庆市自然科学基金(CSTB2022NSCQ-MSX1231) （CSTB2022NSCQ-MSX1231）
重庆市高等教育教学改革研究项目(243400) （243400）
国网重庆信通公司项目(SGCQXT00JSJS2400122) （SGCQXT00JSJS2400122）
