考虑数据分类的建筑电能耗集成预测方法

建筑侧各类可再生能源的应用日益普及,建筑电能耗预测在用能供需平衡、电网稳定运行、尖峰需求响应等方面发挥越来越重要作用。尽管诸多数据驱动模型在能耗预测方面获得广泛应用,当前仍缺乏预测精度高、泛化能力强的短期预测模型。针对该问题,提出一种基于建筑物能耗特点并结合数据挖掘技术的分类集成式能耗预测方法。首先,采用递归特征消除法对数据进行特征筛选,并用模糊C均值聚类算法对训练集数据进行聚类,使用K最邻近法对验证集和测试集数据进行归类；选择5种结合智能优化算法的混合数据驱动模型作为子学习器,分别对每类数据做预测,最后使用多元线性回归法进行结果集成。经3个建筑电力用能案例验证,此集成预测模型精度均优于单个子模型,具有适用不同建筑类型和用能尺度的预测潜力。

An ensemble forecasting method for building electricity consumption considering data classification

The application of various types of renewable energy on the building side is becoming more and more popular. Forecasting of building electricity consumption plays an increasingly important role in the balance of energy supply and demand, stable grid operation,and peak demand response. Although many data-driven models have been widely used in energy consumption prediction, there is still a lack of short-term prediction models with high prediction accuracy and strong generalization ability. In order to solve this problem, a classification and integration energy consumption prediction method based on the characteristics of building energy consumption and combined with data mining technology is proposed. Firstly, the recursive feature elimination method is used to screen the features of the data, and the fuzzy C-means clustering algorithmisused to cluster the training set data, meanwhile, K-nearest neighbor methodis used to classify the validation set and test set data. Then, five hybrid data-driven models combined with intelligent optimization algorithms areselected as sub-learners, and each type of data is predicted respectively. Finally, multiple linear regression method is used to integrate the results. The accuracy of the ensemble prediction model is better than that of single sub-model, and it has potential to predict different building types and energy use scales.