Performance of fuzzy inference system for prediction of daily entrance flow into Bostan dam

Rajabi, G. A. (2012) Performance of fuzzy inference system for prediction of daily entrance flow into Bostan dam. Masters thesis, University of Zabol.

Performance of fuzzy inference system.pdf

Download (82kB) | Preview


On one hand, substantially increasing demand for water consumption driven by population growth, and on the other hand limited water resources makes water scarcity a crucial problem in Iran. This in turn has a deleterious effect on the rivers and inflows at dams' reservoirs. Therefore, stream flow prediction is essential for water quality management, hydropower, irrigation systems and management of dam reservoir utilization. In recent years, use of Artificial intelligence methods for modeling of Hydrological phenomenon's that is including complexity and uncertainly, is considered scholars. In the recent years, much attention has been paid to the use of Artificial intelligence techniques for modeling hydrological phenomena that has high complexity and uncertainly. In this research, the performances of Fuzzy Inference System (FIS), Adaptive Neuro-Fuzzy Inference System (ANFIS) and an integrated model of Wavelet-ANFIS for predicting daily inflow at Boostan dam reservoir located in Golestan province have been comparatively evaluated on the basis of daily data for a 20-year period (1988-2008) including inflow discharge, rainfall and air temperature. The best combination of the model inputs were selected with Gamma Test (GT) based on the minimum Gamma value, including discharge on the previous day, discharge on the previous two days, and temperature on the previous two days and rainfall at the same day. In addition, the training data length of 5840 and the testing data length of 1460 were determined by using M Test. After conducting prediction by using AFNIS and Wavelet-ANFIS models, the performances of these models were evaluated on the basis of statistical criteria of Coefficient of Nash - Sutcliffe (CNS), correlation coefficient (R), Root mean square error (RMSE) and mean absolute error (MAE). ANFIS and Wavelet-ANFIS models are compared due to the performance evaluation criteria, indicate that Wavelet- ANFIS model is best predictive model with CNS=0.922 , R=0.961, RMSE=0.406, and MAE=0.248, compared with ANFIS model with CNS=0.834 , R=0.92, RMSE=0.59, and MAE=0.249. The obtained results indicate higher accuracy of Wavelet-ANFIS model rather than ANFIS model in order to predict daily inflow at Boostan dam reservoir.

Item Type: Thesis (Masters)
Uncontrolled Keywords: prediction; Adaptive Neuro-Fuzzy Inference System (ANFIS); Wavelet transform; Gamma Test; Boostan Dam of Golestan province
Subjects: T Technology > TD Environmental technology. Sanitary engineering
Depositing User: admin admin1 admin2
Date Deposited: 06 Nov 2016 05:54
Last Modified: 06 Nov 2016 05:54

Actions (login required)

View Item View Item