A simplified HVAC energy prediction method based on degree-day. (November 2019)

• Record Type: Journal Article
• Authors: Sha, Huajing; Xu, Peng; Hu, Chonghe; Li, Zhiling; Chen, Yongbao; Chen, Zhe
• Is Part Of: Sustainable cities and society. Volume 51(2020)
• Publication Date: 2019-11
• Subjects: Data-driven model -- Energy prediction -- Degree-day; Sustainable urban development -- Periodicals; Sustainable buildings -- Periodicals; Urban ecology (Sociology) -- Periodicals; 307.76
• DOI: 10.1016/j.scs.2019.101698
• Languages: English
• ISSNs: 2210-6707

Abstract

Highlights: A simplified data-driven model of building HVAC energy prediction is proposed. The method proposed by this paper innovatively use degree-day as input feature which is more efficient. Three different machine learning algorithms are compared in this research. The coefficient of variation of root mean squared error (CV-RMSE) of this method is less than 20%. Abstract: A building heating, ventilation, and air-conditioning (HVAC) system consumes large amounts of energy. Energy consumption prediction is an effective strategy for operation optimization and energy management in a building. The energy consumption of an HVAC system in a building is influenced by many factors, such as weather conditions, building usage, and thermal performance. However, it is impractical to consider all factors for predicting energy consumption. In this paper, a simplified data-driven model is proposed for predicting the energy consumption of an HVAC system in a building. A novel feature transformation method is introduced to select the most relevant features. Three input features (i.e., degree-day, day type, and month type) are finally adopted in this model. Compared to models developed in previous studies, this simplified model largely reduces the computation time and is easier to operate. The cross-validated root mean square error of this method for cooling energy prediction is less than 20%, indicating its suitability for use in engineering applications.