Estimation of Stream Temperature in Degirmendere River (Trabzon-Turkey) Using Artificial Neural Network Model
Abstract
Artificial Neural Networks (ANN) is a modeling technique with training which takes the working system of the brain as basis. Learning in ANNs is realized with the renewal of the connection gaps. ANNs make possible to solve the non-defined problems through the learning ability. In this study, we have estimated various environmental factors using an artificial model of the brain, known as Artificial Neural Network (ANN). Here, we has developed and tested an ANN model to predict stream temperature of Degirmendere in Black Sea, using local water temperature, air temperature, and stream temperature. In the structure of the ANN used in the model, the number of the hidden neuron was determined as 16, Sum-Squared Error was determined as 0.005 and the number of the iteration has been determined as 40000. As a result of the regression analysis realized between the model outputs and measurement results obtained in the study, the value of r = 0.92 was calculated. When the other literature studies which had done before have been examined, in the light of the model outputs and statistical evaluations, and regarding the complex and nonlinear structure of the study environment, it was seen that the ANN modeling technique can be utilized in the timely prediction of the temperatures of the stream waters.
Estimation of Stream Temperature in Degirmendere River (Trabzon-Turkey) Using Artificial Neural Network Model
Abstract
Artificial Neural Networks (ANN) is a modeling technique with training which takes the working system of the brain as basis. Learning in ANNs is realized with the renewal of the connection gaps. ANNs make possible to solve the non-defined problems through the learning ability. In this study, we have estimated various environmental factors using an artificial model of the brain, known as Artificial Neural Network (ANN). Here, we has developed and tested an ANN model to predict stream temperature of Degirmendere in Black Sea, using local water temperature, air temperature, and stream temperature. In the structure of the ANN used in the model, the number of the hidden neuron was determined as 16, Sum-Squared Error was determined as 0.005 and the number of the iteration has been determined as 40000. As a result of the regression analysis realized between the model outputs and measurement results obtained in the study, the value of r = 0.92 was calculated. When the other literature studies which had done before have been examined, in the light of the model outputs and statistical evaluations, and regarding the complex and nonlinear structure of the study environment, it was seen that the ANN modeling technique can be utilized in the timely prediction of the temperatures of the stream waters.
Details
| Primary Language | Turkish |
|---|---|
| Journal Section | Articles |
| Authors | |
| Publication Date | April 1, 2009 |
| Published in Issue | Year 2009 Volume: 9 Issue: 2 |