Incorporating machine learning with building network analysis to predict multi-building energy use

Abstract

Predicting energy use of campuses or city district buildings has recently gained more attention due to dynamic large-scale building energy demands. This data enlightens public's awareness of energy use and informs building energy policy. Understanding the correlation of energy use patterns between buildings is a key issue to analyzing multi-building energy use. Moreover, how to apply this inter-building relationship to multi-building energy prediction, using significantly less amount of building energy data, is still an open question. To solve such problems, this study proposed an interdisciplinary research method to predict multi-building energy use by integrating a social network (SN) analysis with an Artificial Neural Network (ANN) technique. The SN method was used to identify reference buildings and determine correlations between reference buildings and non-reference buildings.The ANN technique was applied to learn correlations and historical building energy use, and then used to predict multi-building energy use. To validate the SN-ANN method, 17 buildings in the Southeast ， Corresponding author. tel.: +86 (25) 8379-5689; xxdseu@126.com (Xiaodong Xu) ， Corresponding author. tel.: +1 (510) 486-7082; thong@lbl.gov (Tianzhen Hong) University campus, located in Nanjing, China, were studied. These buildings have three years of actual monthly electricity use data, and were grouped into four types: office, educational, laboratory, and residential. The results showed the integrated SN-ANN method achieved an accuracy of 90.28% for the predicted energy use for all building groups. Finally, this study provides insights into advancing the interdisciplinary research on multi-building energy use prediction.

Identifying key determinants for building energy analysis from urban building datasets

“…This study provides insights regarding urban energy analysis, sustainable urban design, and urban energy planning. First, urban energy analysis usually includes three main branches [52]-big data mining that supports urban building energy policymaking [53,54], data-driven urban building energy modeling (classification, clustering, and prediction) [55][56][57], and urban-scale building energy simulation [58]. Recent studies have been focused more on the building energy usage dataset or building physical dataset for the simulation.…”

Forecasting district-scale energy dynamics through integrating building network and long short-term memory learning algorithm

“…Caputo et al used four archetypes to characterize the energy performance in a neighborhood built environment [33]. Holistic building energy consumption data can be used for defining reference buildings by investigating the closeness of building groups [34]. Deb and Lee [35] determined the critical variables that influencing energy consumption with a cluster analysis on a small sample of 56 office buildings to represent a large building dataset [36].…”

Design and Simulation of an Energy Homeostaticity System for Electric and Thermal Power Management in a Building with Smart Microgrid

“…Other authors have proposed similar works showing simulation and energy management system (EMS) design, such as [27]. For example, reference [28] present an interesting paper for predicting multi-building energy use at a campus or city district scale in Chine involving 17 buildings. In short, they assert that predicting energy use in a cluster of buildings is possible and can lead to better building management policies and guidelines, by noting some key interactions among them.…”

“…This imitative has, according to the authors, recently gained much attention; and more researchers are beginning to define reference buildings, as a metric standard against which other similar buildings can be compared and monitored, so as to achieve certain efficiency standards and metrics. The study focuses on inter-impact amongst building groups [28].…”