IAHR Document Library

« Back to Library Homepage « Proceedings of the 31st IAHR World Congress (Seoul, 2005)

A Neural Network Based Operational Reservoir Inflow Forecasting Model for an Alpine Hydropower Plant

Author(s): Dieter Theiner; Markus Zossmayr; Peter Rutschmann

Linked Author(s): Peter Rutschmann

Keywords: Neural network; Reservoir operation; Runoff forecasting; Multi step forecast

Abstract: The authors used artificial neural network (ANN) models for a three step forecasting of mean daily inflow into a power plant reservoir (42 hm3) in a High Alpine catchment (150 km2) in Northern Italy. Forecasting software uses data from two meteorological stations (precipitation, daily temperature range, snow heights) and data from meteorological forecasts. Besides, the model uses observed inflow from the preceding days and also additional information like date, amount of precipitation fallen at days with mean temperatures below 0°C, or averaged temperatures over a longer time frame. Best results are obtained with three separated multilayer feedforward architectures. Each model has just one output node and predicts one time step.


Year: 2005

Copyright © 2024 International Association for Hydro-Environment Engineering and Research. All rights reserved. | Terms and Conditions