摘要
现有的预测方法很少独立分析能源消耗的周期性特征。本文提出了一个短期办公建筑能耗预测模型(DLnet),以解决周期性能耗数据利用效率低下的问题。首先,利用STL对能耗数据的周期成分进行分解,通过网格搜索算法寻找能耗数据的最优周期;然后,根据最优周期构建周期块;再根据周期块的数据形状构建时间序列块数据;之后,利用长短期记忆(LSTM)对时间序列块数据和周期块数据进行训练和学习;最后,通过线性回归将时间序列块数据和周期块数据的预测结果进行融合。事实证明,所提出的模型的4个预测精度指标分别比LSTM模型高7%,21%,25%和26%。
The existing forecasting methods rarely analyze the periodic characteristics of energy consumption independently.A short-term office building energy consumption prediction model(DLnet)is proposed to solve the problem of low efficiency in the utilization of periodic energy consumption data.Firstly,the period component of the energy consumption data is decomposed using STL,and the optimal period of the energy consumption data is searched by grid searching algorithm.Secondly,periodic block is constructed according to optimal period.Then,time-series block data is constructed according to the data shape of the Periodic block.Next,time-series block data and the Periodic block data are trained and learned using LSTM.Finally,the prediction results of the time-series block data and the Periodic block data are fused by linear regression.The fact demonstrates that the four prediction precision indicators of the proposed model are 7%,21%,25%,and 26% higher than those of the long short-term memory(LSTM)model.
作者
廖雪超
黄相
LIAO Xuechao;HUANG Xiang(School of Computer Science and Technology,Wuhan University of Science and Technology,Wuhan 430065,China;Key Laboratory of Intelligent Information Processing and Real-time Industrial Systems,Wuhan,430065,China)
出处
《传感器与微系统》
CSCD
北大核心
2024年第10期46-49,54,共5页
Transducer and Microsystem Technologies
基金
国家自然科学基金资助项目(62273264)。
关键词
时序块
周期块
最佳周期
STL
长短期记忆
time-series block
periodic block
optimal period
seasonal-trend decomposition procedures based on Loess
long short-term memory
