Short term electricity load forecasting for institutional buildings
Short term electricity load forecasting for institutional buildings
复制标题
DOI：
10.1016/j.egyr.2019.08.086
复制
发表时间：
2019-11
期刊：
影响因子：
5.2
通讯作者：
Yunsun Kim;Heung-gu Son;Sahm Kim
中科院分区：
文献类型：
--
作者：
Yunsun Kim;Heung-gu Son;Sahm Kim
随着气候变化、技术发展和能源政策导致高峰需求的增加，高峰负荷需求预测在建筑单位部门非常重要。因此，准确的峰值负荷预测是防止停电或能源损失的关键作用。本文对首尔某机构建筑的高峰负荷需求进行了预测研究。数据集收集自由23栋建筑组成的校园区域。采用ARIMA模型、ARIMA- garch模型、多季节指数平滑和人工神经网络模型。我们通过移动窗口模拟和步进预测找到了一个最优模型。此外，包括天气和假日变量对于预测峰值负荷需求至关重要。带外部变量的人工神经网络模型（NARX）对提前1小时至1天的预测效果最好。
Peak load demand forecasting is important in building unit sectors, as climate change, technological development, and energy policies are causing an increase in peak demand. Thus, accurate peak load forecasting is a critical role in preventing a blackout or loss of energy. This paper presents a study forecasting peak load demand for an institutional building in Seoul. The dataset were collected from campus area consisting of 23 buildings. ARIMA models, ARIMA-GARCH models, multiple seasonal exponential smoothing, and ANN models are used. We find an optimal model with moving window simulations and step-ahead forecasts. Also, including weather and holiday variables is crucial to predict peak load demand. The ANN model with external variables (NARX) worked best for 1-h to 1-d ahead forecasting.