You need JavaScript to view this

Short-term forecasts of district heating load and outdoor temperature by use of on-line connected computers; Korttidsprognoser foer fjaerrvaermelast och utetemperatur med on-linekopplade datorer

Abstract

In this report the available methods for forecasting weather and district heating load have been studied. A forecast method based on neural networks has been tested against the more common statistical methods. The accuracy of the weather forecasts from the SMHI (Swedish Meteorological and Hydrological Institute) has been estimated. In connection with these tests, the possibilities of improving the forecasts by using on-line connected computers has been analysed. The most important results from the study are: Energy company staff generally look upon the forecasting of district heating load as a problem of such a magnitude that computer support is needed. At the companies where computer calculated forecasts are in use, their accuracy is regarded as quite satisfactory; The interest in computer produced load forecasts among energy company staff is increasing; At present, a sufficient number of commercial suppliers of weather forecasts as well as load forecasts is available to fulfill the needs of energy companies; Forecasts based on neural networks did not attain any precision improvement in comparison to more traditional statistical methods. There may though be other types of neural networks, not tested in this study, that are possibly capable of improving the forecast precision; Forecasts of outdoor temperature  More>>
Authors:
Malmstroem, B; Ernfors, P; Nilsson, Daniel; Vallgren, H [1] 
  1. Chalmers Tekniska Hoegskola, Goeteborg (Sweden). Institutionen foer Energiteknik
Publication Date:
Oct 01, 1996
Product Type:
Technical Report
Report Number:
SVF-589
Reference Number:
SCA: 290800; 320603; PA: SWD-97:007038; EDB-97:013731; NTS-97:005128; SN: 97001719420
Resource Relation:
Other Information: DN: Figures and tables with text in English; PBD: Oct 1996
Subject:
29 ENERGY PLANNING AND POLICY; 32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATION; DISTRICT HEATING; LOAD MANAGEMENT; ENERGY DEMAND; FORECASTING; WEATHER; METEOROLOGY; AMBIENT TEMPERATURE; ON-LINE SYSTEMS; NEURAL NETWORKS; MATHEMATICAL MODELS; NUMERICAL DATA
OSTI ID:
414558
Research Organizations:
Stiftelsen foer Vaermeteknisk Forskning, Stockholm (Sweden)
Country of Origin:
Sweden
Language:
Swedish
Other Identifying Numbers:
Journal ID: ISSN 0282-3772; Other: ON: DE97722060; TRN: SE9707038
Availability:
OSTI as DE97722060
Submitting Site:
SWD
Size:
74 p.
Announcement Date:

Citation Formats

Malmstroem, B, Ernfors, P, Nilsson, Daniel, and Vallgren, H. Short-term forecasts of district heating load and outdoor temperature by use of on-line connected computers; Korttidsprognoser foer fjaerrvaermelast och utetemperatur med on-linekopplade datorer. Sweden: N. p., 1996. Web.
Malmstroem, B, Ernfors, P, Nilsson, Daniel, & Vallgren, H. Short-term forecasts of district heating load and outdoor temperature by use of on-line connected computers; Korttidsprognoser foer fjaerrvaermelast och utetemperatur med on-linekopplade datorer. Sweden.
Malmstroem, B, Ernfors, P, Nilsson, Daniel, and Vallgren, H. 1996. "Short-term forecasts of district heating load and outdoor temperature by use of on-line connected computers; Korttidsprognoser foer fjaerrvaermelast och utetemperatur med on-linekopplade datorer." Sweden.
@misc{etde_414558,
title = {Short-term forecasts of district heating load and outdoor temperature by use of on-line connected computers; Korttidsprognoser foer fjaerrvaermelast och utetemperatur med on-linekopplade datorer}
author = {Malmstroem, B, Ernfors, P, Nilsson, Daniel, and Vallgren, H}
abstractNote = {In this report the available methods for forecasting weather and district heating load have been studied. A forecast method based on neural networks has been tested against the more common statistical methods. The accuracy of the weather forecasts from the SMHI (Swedish Meteorological and Hydrological Institute) has been estimated. In connection with these tests, the possibilities of improving the forecasts by using on-line connected computers has been analysed. The most important results from the study are: Energy company staff generally look upon the forecasting of district heating load as a problem of such a magnitude that computer support is needed. At the companies where computer calculated forecasts are in use, their accuracy is regarded as quite satisfactory; The interest in computer produced load forecasts among energy company staff is increasing; At present, a sufficient number of commercial suppliers of weather forecasts as well as load forecasts is available to fulfill the needs of energy companies; Forecasts based on neural networks did not attain any precision improvement in comparison to more traditional statistical methods. There may though be other types of neural networks, not tested in this study, that are possibly capable of improving the forecast precision; Forecasts of outdoor temperature and district heating load can be significantly improved through the use of on-line-connected computers supplied with instantaneous measurements of temperature and load. This study shows that a general reduction of the load prediction errors by approximately 15% is attainable. For short time horizons (less than 5 hours), more extensive load prediction error reductions can be reached. For the 1-hour time horizon, the possible reduction amounts to up to 50%. 21 refs, 4 figs, 7 appendices}
place = {Sweden}
year = {1996}
month = {Oct}
}