A generally applicable, simple and adaptive forecasting method for the short-term heat load of consumers

Research output: Contribution to journalArticleResearchpeer-review

Abstract

Energy management systems aiming for an efficient operation of hybrid energy systems with a high share of different renewable energy sources strongly benefit from short-term forecasts for the heat-load. The forecasting methods available in literature are typically tailor-made, complex and non-adaptive. This work condenses these methods to a generally applicable, simple and adaptive forecasting method for the short-term heat load. From a comprehensive literature review as well as the analysis of measurement data from seven different consumers, varying in size and type, the ambient temperature, the time of the day and the day of the week are deduced to be the most dominating factors influencing the heat load. According to these findings, the forecasting method bases on a linear regression model correlating the heat load with the ambient temperature for each hour of the day, additionally differentiating between working days and weekend days. These models are used to predict the future heat load by using forecasts for the ambient temperature from weather service providers. The model parameters are continuously updated by using historical data for the ambient temperature and the heat load, i.e. the forecasting method is adaptive. Additionally, the current prediction error is used to correct the prediction for the near future. Due to their simplicity, all necessary steps of the forecasting method, the update of the model parameters, the prediction based on linear regression models and the correction, can be implemented and computed with little effort. The final evaluation with measurement data from all seven consumers investigated leads to a Mean Absolute Range Normalized Error (MARNE) of 2.9% on average, and proves the general applicability of the forecasting method. In summary, the forecasting method developed is generally applicable, simple and adaptive, making it suitable for the use in energy management systems aiming for an efficient operation of hybrid energy systems.

LanguageEnglish
Pages73-81
Number of pages9
JournalApplied Energy
Volume241
DOIs
StatusPublished - 1 May 2019

Fingerprint

forecasting method
Thermal load
Energy management systems
prediction
temperature
Linear regression
literature review
energy
Temperature
weather

Keywords

  • Energy management system
  • Forecast
  • Heat demand
  • Heat load
  • Prediction
  • Short-term

ASJC Scopus subject areas

  • Mechanical Engineering
  • Energy(all)
  • Management, Monitoring, Policy and Law
  • Building and Construction

Cite this

A generally applicable, simple and adaptive forecasting method for the short-term heat load of consumers. / Nigitz, Thomas; Gölles, Markus.

In: Applied Energy, Vol. 241, 01.05.2019, p. 73-81.

Research output: Contribution to journalArticleResearchpeer-review

@article{dab4be577a12478f9440e9182fad1646,
title = "A generally applicable, simple and adaptive forecasting method for the short-term heat load of consumers",
abstract = "Energy management systems aiming for an efficient operation of hybrid energy systems with a high share of different renewable energy sources strongly benefit from short-term forecasts for the heat-load. The forecasting methods available in literature are typically tailor-made, complex and non-adaptive. This work condenses these methods to a generally applicable, simple and adaptive forecasting method for the short-term heat load. From a comprehensive literature review as well as the analysis of measurement data from seven different consumers, varying in size and type, the ambient temperature, the time of the day and the day of the week are deduced to be the most dominating factors influencing the heat load. According to these findings, the forecasting method bases on a linear regression model correlating the heat load with the ambient temperature for each hour of the day, additionally differentiating between working days and weekend days. These models are used to predict the future heat load by using forecasts for the ambient temperature from weather service providers. The model parameters are continuously updated by using historical data for the ambient temperature and the heat load, i.e. the forecasting method is adaptive. Additionally, the current prediction error is used to correct the prediction for the near future. Due to their simplicity, all necessary steps of the forecasting method, the update of the model parameters, the prediction based on linear regression models and the correction, can be implemented and computed with little effort. The final evaluation with measurement data from all seven consumers investigated leads to a Mean Absolute Range Normalized Error (MARNE) of 2.9{\%} on average, and proves the general applicability of the forecasting method. In summary, the forecasting method developed is generally applicable, simple and adaptive, making it suitable for the use in energy management systems aiming for an efficient operation of hybrid energy systems.",
keywords = "Energy management system, Forecast, Heat demand, Heat load, Prediction, Short-term",
author = "Thomas Nigitz and Markus G{\"o}lles",
year = "2019",
month = "5",
day = "1",
doi = "10.1016/j.apenergy.2019.03.012",
language = "English",
volume = "241",
pages = "73--81",
journal = "Applied Energy",
issn = "0306-2619",
publisher = "Elsevier B.V.",

}

TY - JOUR

T1 - A generally applicable, simple and adaptive forecasting method for the short-term heat load of consumers

AU - Nigitz, Thomas

AU - Gölles, Markus

PY - 2019/5/1

Y1 - 2019/5/1

N2 - Energy management systems aiming for an efficient operation of hybrid energy systems with a high share of different renewable energy sources strongly benefit from short-term forecasts for the heat-load. The forecasting methods available in literature are typically tailor-made, complex and non-adaptive. This work condenses these methods to a generally applicable, simple and adaptive forecasting method for the short-term heat load. From a comprehensive literature review as well as the analysis of measurement data from seven different consumers, varying in size and type, the ambient temperature, the time of the day and the day of the week are deduced to be the most dominating factors influencing the heat load. According to these findings, the forecasting method bases on a linear regression model correlating the heat load with the ambient temperature for each hour of the day, additionally differentiating between working days and weekend days. These models are used to predict the future heat load by using forecasts for the ambient temperature from weather service providers. The model parameters are continuously updated by using historical data for the ambient temperature and the heat load, i.e. the forecasting method is adaptive. Additionally, the current prediction error is used to correct the prediction for the near future. Due to their simplicity, all necessary steps of the forecasting method, the update of the model parameters, the prediction based on linear regression models and the correction, can be implemented and computed with little effort. The final evaluation with measurement data from all seven consumers investigated leads to a Mean Absolute Range Normalized Error (MARNE) of 2.9% on average, and proves the general applicability of the forecasting method. In summary, the forecasting method developed is generally applicable, simple and adaptive, making it suitable for the use in energy management systems aiming for an efficient operation of hybrid energy systems.

AB - Energy management systems aiming for an efficient operation of hybrid energy systems with a high share of different renewable energy sources strongly benefit from short-term forecasts for the heat-load. The forecasting methods available in literature are typically tailor-made, complex and non-adaptive. This work condenses these methods to a generally applicable, simple and adaptive forecasting method for the short-term heat load. From a comprehensive literature review as well as the analysis of measurement data from seven different consumers, varying in size and type, the ambient temperature, the time of the day and the day of the week are deduced to be the most dominating factors influencing the heat load. According to these findings, the forecasting method bases on a linear regression model correlating the heat load with the ambient temperature for each hour of the day, additionally differentiating between working days and weekend days. These models are used to predict the future heat load by using forecasts for the ambient temperature from weather service providers. The model parameters are continuously updated by using historical data for the ambient temperature and the heat load, i.e. the forecasting method is adaptive. Additionally, the current prediction error is used to correct the prediction for the near future. Due to their simplicity, all necessary steps of the forecasting method, the update of the model parameters, the prediction based on linear regression models and the correction, can be implemented and computed with little effort. The final evaluation with measurement data from all seven consumers investigated leads to a Mean Absolute Range Normalized Error (MARNE) of 2.9% on average, and proves the general applicability of the forecasting method. In summary, the forecasting method developed is generally applicable, simple and adaptive, making it suitable for the use in energy management systems aiming for an efficient operation of hybrid energy systems.

KW - Energy management system

KW - Forecast

KW - Heat demand

KW - Heat load

KW - Prediction

KW - Short-term

UR - http://www.scopus.com/inward/record.url?scp=85062474381&partnerID=8YFLogxK

U2 - 10.1016/j.apenergy.2019.03.012

DO - 10.1016/j.apenergy.2019.03.012

M3 - Article

VL - 241

SP - 73

EP - 81

JO - Applied Energy

T2 - Applied Energy

JF - Applied Energy

SN - 0306-2619

ER -