A Building Energy Consumption Prediction Method Based on Random Forest and ARMA
A Building Energy Consumption Prediction Method Based on Random Forest and ARMA
复制标题
DOI：
10.1109/cac.2018.8623540
复制
发表时间：
2018-11
期刊：
影响因子：
--
通讯作者：
Beiyan Jiang;Zhijin Cheng;Qianting Hao;Nan Ma
中科院分区：
文献类型：
--
作者：
Beiyan Jiang;Zhijin Cheng;Qianting Hao;Nan Ma
针对建筑能耗优化问题，提出了一种基于随机森林（RF）和自回归滑动平均（阿尔马）算法的建筑能耗预测方法。考虑到建筑能耗受设备使用、人员信息、气候条件等因素的影响，引入随机森林方法，基于历史数据建立预测模型。按运行特点分为工作日模式和非工作日模式。为了解决临时工作引起的工作模式和非工作模式之间的切换问题，引入阿尔马模型作为基准，对所提出的模型进行改进。一些案例研究已经进行了测试所提出的方法。结果表明，该方法在稳定和暂时条件下都能提供较好的预测。
To address the issue of building energy consumption optimization, a novel prediction method based on Random Forest (RF) and Auto Regressive Moving Average (ARMA) algorithm is presented in this paper. Considering building energy consumption was affected by the factors involving equipment usage, personnel information, and climate conditions, random forest method was introduced to build forecasting models based on the historical data. It includes working-day mode and non-working day mode according to operating characteristics. To address the switching challenge between working mode and non-working mode due to temporary work, ARMA model is introduced to serve as the benchmark to improve the presented model. Some case studies have been carried out to test the proposed method. The results indicate that it could provide good predictions under both stable and temporary conditions.