电力系统自动化2026，Vol.50Issue(5)：54-62,9.DOI:10.7500/AEPS20241118002
基于LSTM-Transformer的钢铁工业用户调节潜力预测与优化
LSTM-Transformer Based Prediction and Optimization for User Regulation Potential of Iron and Steel Industry
摘要
Abstract
As one of the main bodies of urban electricity consumption,industrial users have complex and variable loads that are highly affected by the user regulation potential.Traditional prediction methods have difficulty accurately estimating the regulation capability of iron and steel industrial users.To balance the uncertainty of load fluctuations and the regular characteristics of power consumption behaviors of iron and steel industry users,a prediction method for the regulation potential of iron and steel industry users based on long short-term memory(LSTM)-Transformer is proposed.This method uses the LSTM network to capture the long-term dependencies of sequences such as adjustable industrial load equipment,maintenance schedules,and user regulation potential samples for feature extraction.Meanwhile,the Transformer module is adopted for position encoding,and a dual-layer multi-head self-attention mechanism is used to capture the relationships among different data attributes and concatenate features,thereby obtaining the regulation potential of industrial users under the influence of multiple factors.The actual operation data of an iron and steel plant in Tianjin,China,are selected to compare the potential values calculated by four models.The experimental results show that,compared with other models,the average error of the proposed model is reduced by approximately 40%,achieving higher accuracy and effectively reflecting the regulation potential of the iron and steel industry users.The proposed model provides strong support for optimal scheduling.关键词
需求响应/钢铁工业/负荷/调节潜力/用电/LSTM-Transformer模型/多头自注意力机制Key words
demand response/iron and steel industry/load/regulation potential/electricity consumption/LSTM-Transformer model/multi-head self-attention mechanism引用本文复制引用
李彬,张雨蒙,周照钒..基于LSTM-Transformer的钢铁工业用户调节潜力预测与优化[J].电力系统自动化,2026,50(5):54-62,9.基金项目
国家电网有限公司科技项目:"需求侧可调节资源池动态构建技术及应用验证"(5108-202218280A-2-389-XG). This work is supported by State Grid Corporation of China(No.5108-202218280A-2-389-XG). （5108-202218280A-2-389-XG）
