Informations et ressources scientifiques
sur le développement des zones arides et semi-arides

Titre : The Use Of Different Artificial Neural Network Methods In Long Range Forecasting Of The Water Resources Data

Değişik Yapay Sinir Ağı Metotlarının Su Kaynakları Verisinin Uzun Zaman Aralıklı Tahminlerinde Kullanımı

Auteur : ASLAN İnş. Müh. Erdem

Université de soutenance : İstanbul Teknik Üniversitesi

Grade : Master of Science (MS) 2008

Résumé
In this study, three different Artificial Neural Networks are applied to daily continuous river flow series, daily intermittent river flow series, monthly river flow series and ground water level series of four hydrologic regions from Turkey and in order to investigate the long range forecasting results. Study results are presented by comparing the ANN model long-range forecasts and the observed time series. General Regression Neural Network, Radial Basis Functions Neural Network and Feed Forward Back Propagation Neural Network are used to identify the models. Multi Linear Regression is also employed for comparison with ANN model forecasts. Mean squared error and coefficient of determination are used as the performance comparison criteria. It is seen that long-range forecasting results show variable performances for each type of data series. Nevertheless, General Regression Neural Network is seen to be more efficient compared with other ANN methods considering the performance evaluation criteria. Consequently, GRNN method is found to be the best method because of defining data trend in monthly river flow series and specially ground water level data forecasting.

Présentation

Version intégrale (2,3 Mb)