Predicting building energy consumption during holiday periods
- Resource Type
- Conference
- Authors
- Qiao, Qingyao; Yunusa-Kaltungo, Akilu; Edwards, Rodger
- Source
- 2021 IEEE PES/IAS PowerAfrica PowerAfrica, 2021 IEEE PES/IAS. :1-5 Aug, 2021
- Subject
- Components, Circuits, Devices and Systems
Engineering Profession
General Topics for Engineers
Nuclear Engineering
Power, Energy and Industry Applications
Robotics and Control Systems
Energy consumption
Sensitivity
Input variables
Buildings
Training data
Predictive models
Data models
building
holidays effects
long-term energy consumption
prediction
Prophet model
- Language
Predicting sudden changes in energy consumption within a short time period remains a challenging task for long-term building energy consumption Prediction. In order to better predict long-term energy consumption during holiday periods, this paper proposes a novel Prophet model to adequately capture the energy usage patterns of a classroom room building during Christmas periods under several data scenarios. The results showed that the incorporation of additional weather information as often advocated by several earlier studies failed to improve the prediction accuracy. Although the extension of the training data size can significantly improve the prediction outcomes under certain scenarios, it failed to capture the sudden drop in energy consumption when holiday effects were incorporated. The best performance was achieved when the model was fed with 2-year training data as well as integrating holiday effects.