基于气象与移动数据的楼宇短期负荷预测

Short-Term Load Forecasting of Buildings Based on Meteorological and Mobile Data

针对用电负荷预测问题,提出了一种基于气象数据及移动人群数据的楼宇短期负荷建模方法,使用上海某商场移动数据及该地天气数据作为特征参数,对于样本特征使用模糊C均值(fuzzy C-means,FCM)预处理,具有相似特征的时刻的气象与移动人群数据分别建立梯度提升决策树(gradient lifting decision tree,GBDT)模型。实验表明,引入移动数据对于小样本楼宇短期负荷建模精度有所提升。

Short-term load forecasting of building power consumption has important reference value for energy system evaluation, control and dispatching strategy. Aiming at the problem of power load forecasting, this paper presents a short-term load modeling method based on meteorological data and mobile population. The mobile data of a shopping mall in Shanghai and the weather data of the area are used as feature parameters. For the sample features, the fuzzy C-means(FCM) is used to preprocess, and the weather and movement at the time of similar characteristics are used. Gradient lifting decision tree(GBDT) models were established for moving population data. Experiments show that the introduction of mobile data can improve the accuracy of short-term load modeling for small sample buildings.