基于XGBoost-WOA-BiLSTM-Attention的公共建筑暖通空调能耗预测研究  

PREDICTION OF HVAC ENERGY CONSUMPTION IN PUBLIC BUILDINGS BASED ON XGBOOST-WOA-BILSTM-ATTENTION

作　　者：于水[1] 罗宇晨 安瑞 李思尧 陈志杰 YU Shui;LUO Yu-chen;AN Rui;LI Si-yao;CHEN Zhi-jie(School of Municipal and Environmental Engineering,Shenyang Jianzhu University,110168,Shenyang,China)

机构地区：[1]沈阳建筑大学市政与环境工程学院,沈阳110168

出　　处：《建筑技术》2024年第17期2071-2075,共5页Architecture Technology

基　　金：国家自然科学基金项目(52078308);辽宁省“兴辽英才计划”项目(XLYC2007003);辽宁省教育厅项目(lnzd202003)。

摘　　要：为在双碳目标下实现节能减排,降低能源成本,提出一种基于BiLSTM的公共建筑暖通空调能耗预测模型。在BiLSTM模型基础上,使用XGBoost算法对输入特征进行选择,剔除冗余特征,得到最佳模型输入特征;然后利用WOA优化算法对添加了Attention机制的BiLSTM模型中的6个超参数进行优化,将得到的最优参数代入BiLSTM-Attention神经网络中进行预测,并与BiLSTM模型、BiLSTM-Attention模型和WOA-BiLSTM-Attention模型进行对比。结果表明,所提出的XGBoost-WOA-BiLSTM-Attention模型的RMSE、MAE、R2分别为0.0106、0.006、0.9991,优于其他模型,且相对于持续模型在均方根误差RMSE上提升了98%,为降低公共建筑暖通空调能耗研究提供了参考。To achieve energy conservation and emission reduction under the dual carbon target,and reduce energy costs,a BiLSTM based energy consumption prediction model for HVAC systems in public buildings is proposed.Based on the BiLSTM model,XGBoost algorithm is used to select input features,eliminate redundant features,and obtain the optimal model input features;then,the WOA optimization algorithm was used to optimize the six hyperparameters in the BiLSTM model with the added Attention mechanism.The obtained optimal parameters were substituted into the BiLSTM Attention neural network for prediction,and compared with the BiLSTM model,BiLSTM Attention model,and WOA BiLSTM Attention model.The results show that the RMSE,MAE,and R2 of the proposed XGBoost-WOA BiLSTM Attention model are 0.0106,0.006,and 0.9991,respectively,which are superior to other models.Compared with the continuous model,the root mean square error(RMSE)of the proposed model has increased by 98%,providing some guidance for reducing the energy consumption of HVAC systems in public buildings.

关 键 词：HVAC能耗 XGBoost WOA优化 Attention机制 BiLSTM 

分 类 号：TU74[建筑科学—建筑技术科学]