Load forecasting with neural nets: prediction of the hourly load with a time horizon of up to seven days-a common project of EOS, EPFL and ABB

This article discusses the forecasting of load for a period varying from an hour to a week. As usual for the modelling and prediction of nonlinear processes, the use of artificial neural nets appears very promising. In the project described, measured loads for the past seven days, those for the two preceding days, maxima and minima of average temperatures of the previous day, and predicted day temperatures were used as input values, with the day of the week and day of the year as indicators. The load and temperature are modelled separately. An automated method is used for online testing

Publié dans:
Bulletin des schweizerischen elektrotechnischen vereins, 87, 21, 11-16
load forecasting;neural nets;hourly load prediction;EOS;EPFL;ABB;modelling;nonlinear processes;measured loads;seven days ahead prediction;average temperatures;predicted day temperatures;temperature modelling;load modelling;online testing;

 Notice créée le 2007-04-04, modifiée le 2019-03-31

Évaluer ce document:

Rate this document:
(Pas encore évalué)