Stochastic modelling techniques for generating synthetic energy demand profiles

Sandhya Patidar, David P Jenkins, Sophie Ann Simpson

Research output: Contribution to journalArticle

Abstract

This paper investigates three stochastic modelling procedures for generating N (user specified) synthetic annual electricity demand profiles at one-minute resolution. The paper reviews previous work in the application of Hidden-Markov modelling (HMM) for synthesizing highly stochastic time-series of domestic electricity demand through a sophisticated framework coalescing 480 distinct HMM. The efficiency of a proposed approach for integrating a time-series deseasonalizing technique with a single HMM has been studied in parallel with a compatible stochastic modeling framework of a time-series deseasonalized ARIMA model. Various statistical measures/characteristics of the real and synthetic profiles have been compared for all the three stochastic modelling procedures to identify the most efficient and practically suitable medium for generating synthetic electricity time-series at a fine temporal resolution. Results have been shown for both the individual buildings and the composite (aggregated) profiles of many buildings.
Original languageEnglish
Article number1650014
JournalInternational Journal of Energy and Statistics
Volume4
Issue number3
DOIs
Publication statusPublished - 30 Sep 2016

Fingerprint

Time series
Electricity
Composite materials

Keywords

  • Synthetic time-series
  • Energy
  • Hidden Markov model
  • ARIMA model
  • TIme Series Decomposition

Cite this

@article{b075d55d983840a197c000ac8c2109f1,
title = "Stochastic modelling techniques for generating synthetic energy demand profiles",
abstract = "This paper investigates three stochastic modelling procedures for generating N (user specified) synthetic annual electricity demand profiles at one-minute resolution. The paper reviews previous work in the application of Hidden-Markov modelling (HMM) for synthesizing highly stochastic time-series of domestic electricity demand through a sophisticated framework coalescing 480 distinct HMM. The efficiency of a proposed approach for integrating a time-series deseasonalizing technique with a single HMM has been studied in parallel with a compatible stochastic modeling framework of a time-series deseasonalized ARIMA model. Various statistical measures/characteristics of the real and synthetic profiles have been compared for all the three stochastic modelling procedures to identify the most efficient and practically suitable medium for generating synthetic electricity time-series at a fine temporal resolution. Results have been shown for both the individual buildings and the composite (aggregated) profiles of many buildings.",
keywords = "Synthetic time-series, Energy, Hidden Markov model, ARIMA model, TIme Series Decomposition",
author = "Sandhya Patidar and Jenkins, {David P} and Simpson, {Sophie Ann}",
year = "2016",
month = "9",
day = "30",
doi = "10.1142/S2335680416500149",
language = "English",
volume = "4",
journal = "International Journal of Energy and Statistics",
issn = "2335-6804",
publisher = "World Scientific Publishing",
number = "3",

}

Stochastic modelling techniques for generating synthetic energy demand profiles. / Patidar, Sandhya; Jenkins, David P; Simpson, Sophie Ann.

In: International Journal of Energy and Statistics, Vol. 4, No. 3, 1650014, 30.09.2016.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Stochastic modelling techniques for generating synthetic energy demand profiles

AU - Patidar, Sandhya

AU - Jenkins, David P

AU - Simpson, Sophie Ann

PY - 2016/9/30

Y1 - 2016/9/30

N2 - This paper investigates three stochastic modelling procedures for generating N (user specified) synthetic annual electricity demand profiles at one-minute resolution. The paper reviews previous work in the application of Hidden-Markov modelling (HMM) for synthesizing highly stochastic time-series of domestic electricity demand through a sophisticated framework coalescing 480 distinct HMM. The efficiency of a proposed approach for integrating a time-series deseasonalizing technique with a single HMM has been studied in parallel with a compatible stochastic modeling framework of a time-series deseasonalized ARIMA model. Various statistical measures/characteristics of the real and synthetic profiles have been compared for all the three stochastic modelling procedures to identify the most efficient and practically suitable medium for generating synthetic electricity time-series at a fine temporal resolution. Results have been shown for both the individual buildings and the composite (aggregated) profiles of many buildings.

AB - This paper investigates three stochastic modelling procedures for generating N (user specified) synthetic annual electricity demand profiles at one-minute resolution. The paper reviews previous work in the application of Hidden-Markov modelling (HMM) for synthesizing highly stochastic time-series of domestic electricity demand through a sophisticated framework coalescing 480 distinct HMM. The efficiency of a proposed approach for integrating a time-series deseasonalizing technique with a single HMM has been studied in parallel with a compatible stochastic modeling framework of a time-series deseasonalized ARIMA model. Various statistical measures/characteristics of the real and synthetic profiles have been compared for all the three stochastic modelling procedures to identify the most efficient and practically suitable medium for generating synthetic electricity time-series at a fine temporal resolution. Results have been shown for both the individual buildings and the composite (aggregated) profiles of many buildings.

KW - Synthetic time-series

KW - Energy

KW - Hidden Markov model

KW - ARIMA model

KW - TIme Series Decomposition

U2 - 10.1142/S2335680416500149

DO - 10.1142/S2335680416500149

M3 - Article

VL - 4

JO - International Journal of Energy and Statistics

JF - International Journal of Energy and Statistics

SN - 2335-6804

IS - 3

M1 - 1650014

ER -