TY - JOUR
T1 - A hybrid system of data-driven approaches for simulating residential energy demand profiles
AU - Patidar, Sandhya
AU - Jenkins, David P.
AU - Peacock, Andrew
AU - McCallum, Peter
N1 - Funding Information:
This work was supported by the Engineering and Physical Sciences Research Council [Project: ?Community-scale Energy Demand Reduction in India" (CEDRI), grant number EP/R008655/1].
Publisher Copyright:
© 2021 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group.
PY - 2021/5/4
Y1 - 2021/5/4
N2 - This paper presents a novel system of data-driven approaches for simulating the dynamics of electricity demand profiles. Demand profiles of individual dwellings are decomposed into deterministic (e.g. ‘Trends’ and ‘Seasonal’) and stochastic (‘remainder’) components using the STL (a Seasonal-Trend decomposition procedure based on Loess) approach. Stochastic components are modelled using a Hidden Markov Model (HMM) and combined with deterministic components to generate synthetic demand profiles. To simulate extreme (peak) demand, the synthetic profiles were post-processed using a Generalised Pareto (GP) distribution, and a percentile-based bias-correction scheme. All the techniques are systematically coupled into a hybrid system, referred to as ‘STL_HMM_GP’. The STL_HMM_GP system is thoroughly accessed and validated by comparing a range of statistical characteristic of observed and simulated profiles for three case study communities. The potentials of the STL_HMM_GP system is demonstrated for simulating aggregated demand profiles, generated using an accessible small sample of observed individual demand profiles.
AB - This paper presents a novel system of data-driven approaches for simulating the dynamics of electricity demand profiles. Demand profiles of individual dwellings are decomposed into deterministic (e.g. ‘Trends’ and ‘Seasonal’) and stochastic (‘remainder’) components using the STL (a Seasonal-Trend decomposition procedure based on Loess) approach. Stochastic components are modelled using a Hidden Markov Model (HMM) and combined with deterministic components to generate synthetic demand profiles. To simulate extreme (peak) demand, the synthetic profiles were post-processed using a Generalised Pareto (GP) distribution, and a percentile-based bias-correction scheme. All the techniques are systematically coupled into a hybrid system, referred to as ‘STL_HMM_GP’. The STL_HMM_GP system is thoroughly accessed and validated by comparing a range of statistical characteristic of observed and simulated profiles for three case study communities. The potentials of the STL_HMM_GP system is demonstrated for simulating aggregated demand profiles, generated using an accessible small sample of observed individual demand profiles.
KW - Time series decomposition
KW - community energy demand
KW - energy demand simulation
KW - hidden markov model
KW - statistical modelling
UR - http://www.scopus.com/inward/record.url?scp=85105140415&partnerID=8YFLogxK
U2 - 10.1080/19401493.2021.1908427
DO - 10.1080/19401493.2021.1908427
M3 - Article
SN - 1940-1493
VL - 14
SP - 277
EP - 302
JO - Journal of Building Performance Simulation
JF - Journal of Building Performance Simulation
IS - 3
ER -