A comparison study between artificial neural networks and AR models, applied to Norwegian inflow time series

摘要

Presents the use of artificial neural networks (ANN) and autoregressive models (AR models) as runoff models for application in the Norwegian hydro scheduling system. The main aim of the work is to investigate the performance of the different models applied to inflow time series. Improved performance of the runoff models gives better scheduling and better energy economics. The main contribution of this work is the application of these types of models in identifying daily and weekly inflow in Norwegian hydro electric power systems. The models are applied to single and multivariate systems.