Title
Using evidence accumulation-based clustering and symbolic transformation to group multiple buildings based on electricity usage patterns
RIS ID
140732
Abstract
2020, Springer Nature Singapore Pte Ltd. This paper presents a cluster analysis-based strategy to group multiple buildings based on their electricity usage patterns. In this strategy, an evidence accumulation-based clustering algorithm was first used to cluster the daily electricity usage profiles of all buildings over the course of a whole year. The whole year electricity usage time series data of all buildings were then transformed into symbolic representations based on the clustering result of daily electricity usage profiles and further clustered using an agglomerative hierarchical clustering algorithm to group the buildings. The one-year hourly electricity usage time series data collected from 40 buildings on a campus were used to evaluate the performance of this strategy. The result showed that this strategy can group the buildings effectively so that the electricity usage patterns of the buildings in the same group had high similarity to each other while that among different groups were remarkably different. The results from this study can be potentially used to assist in the planning and operation of building energy systems.
Publication Details
Li, K., Ma, Z., Robinson, D. & Ma, J. (2020). Using evidence accumulation-based clustering and symbolic transformation to group multiple buildings based on electricity usage patterns. Smart Innovation, Systems and Technologies, 163 61-71.