Study on the distribution characteristics and uncertainty of multiple energy load patterns for building group to enhance demand side management

The energy consumption data of buildings contain plenty of knowledge that can be used to enhance demand side management (DSM). However, existing research rarely conducts comprehensive analysis on the multiple energy characteristics for the same building group simultaneously. Additionally, DSM requires the load side to meet the flexibility demands at multiple time scales. To address these problems, the paper presents a strategy with two-stage clustering to recognize multiple energy load patterns of higher education buildings at different time scales. In this work, the low frequency component (LFC) and high frequency component (HFC) are obtained from daily load curve by wavelet decomposition, which are clustered by two-stage clustering to construct typical patterns of load at different time scales. Based on these patterns, a series of exploratory studies on their characteristics are launched. The influence characteristic of factors on multiple energy load patterns are revealed. The results show that different functional buildings have obvious differences in the distribution of their electrical LFC patterns on working days and nonworking days, and the shapes of the dominant patterns of power load and cooling load are more consistent with the change of the ambient temperature throughout the day. In addition, the uncertainty of the load pattern is quantified by entropy, which is used to reflect the order degree of the load pattern change over time for each building. The results indicate that probability of a pattern remaining unchanged is higher than transitioning to other patterns, and the uncertainty of HFC pattern for power load is generally higher than that of LFC. The results obtained from this study could be potentially used to enhance DSM.