Two approaches for synthesizing scalable residential energy consumption data

Xiufeng Liu, Nadeem Iftikhar, Huan Huo, Li Rongling, Per Sieverts Nielsen

Publikation: Bidrag til tidsskriftTidsskriftsartikelForskningpeer review

Abstract

Research on integrated systems, simulations, and demand-side management algorithms in the energy sector requires scalable detailed energy consumption data. However, due to privacy issues, it is often difficult to obtain sufficiently large data sets. This paper proposes two distinctly different methods for synthesizing fine-grained energy consumption data of residential households, namely a regression-based and a probability-based method. The proposed methods each use a supervised machine learning method, which trains the models with an empirical data set and then generate large data sets based on the models. This paper describes the data generation process, the optimization techniques, and the parallel data generation in a cluster. This paper evaluates the proposed methods and compares consumption profiles with empirical data in detail, including patterns, statistical information and data generation performance in the cluster. The results demonstrate the effectiveness of the proposed data generation methods and their efficiency in generating large-scale data sets.
OriginalsprogEngelsk
TidsskriftFuture Generation Computer Systems
Vol/bind95
Sider (fra-til)586-600
Antal sider15
ISSN0167-739X
DOI
StatusUdgivet - jun. 2019

Emneord

  • Byggeri, miljø og energi

Fingeraftryk

Dyk ned i forskningsemnerne om 'Two approaches for synthesizing scalable residential energy consumption data'. Sammen danner de et unikt fingeraftryk.

Citationsformater