Short-term Prediction of Power Consumption for Large-scale Public Buildings based on Regression Algorithm

Abstract

Energy consumption monitoring and regulation on large-scale public buildings are always an indispensable part in building energy conservation. With the mounting establishment of the building sub-metering platform, massive amounts of historical power consumption data are provided. In this paper, different building types of six large-scale public buildings in Shanghai are selected, with their sub-metering data deeply analyzed. The concept of CDHs/HDHs (cooling/heating degree hours) is introduced and weekly prediction models of total building power consumption are proposed by the way of multiple linear regression algorithm which is relatively simple and easy to understand. The prediction models are validated to have great accuracy and general applicability in the paper, offering reliable instructions to the building facility manager and relevant competent authorities in terms of decision making and policy implementation.