苏州科技大学学报（自然科学版）2025，Vol.42Issue(3)：70-77,8.DOI:10.12084/j.issn.2096-3289.2025.03.009
基于CNN-LSTM-Attention的空调负荷预测
Air conditioning load forecasting based on CNN-LSTM-Attention
摘要
Abstract
To achieve the dual-carbon goals and reduce carbon emissions,air conditioning load forecasting has become a critical task for optimizing energy management and conserving energy.Single neural network models often struggle with the complexity of processing high-dimensional time series data.This paper proposes an inte-grated model based on Convolutional Neural Networks(CNN),Long Short-Term Memory networks(LSTM),and an attention mechanism.The model first employs CNN to extract key features from the time series,reducing data noise and redundancy.Subsequently,LSTM is utilized to capture the temporal dependencies within the load data,and an attention mechanism is introduced to enhance the model's focus on critical time steps,thereby improving prediction accuracy.Comparative experiments with traditional algorithms,such as BP and LSTM neural networks,demonstrate the effectiveness of the proposed CNN-LSTM-Attention model in air conditioning load forecasting.The results show that the model outperforms other network models in prediction accuracy,providing precise load forecasting support for energy conservation and emission reduction.关键词
空调负荷预测/CNN网络/LSTM网络/注意力机制Key words
air conditioning load forecasting/CNN network/LSTM network/attention mechanism分类
通用工业技术引用本文复制引用
范见雪,陈鑫..基于CNN-LSTM-Attention的空调负荷预测[J].苏州科技大学学报（自然科学版）,2025,42(3):70-77,8.基金项目
国家自然科学基金青年项目(62203317) （62203317）
江苏省自然科学基金青年项目(BK20210862) （BK20210862）
江苏省高等学校自然科学研究面上项目(21KJD120001) （21KJD120001）
