M.A. Thesis:

temperature forecasting by using artificial neural networks (case study:esfarayen)

Temperature changes and future temperature predictions are so important; therefore, the element of temperature is one of the main factors in the classification of climate and, accordingly, fluctuations and temperature variability are some particular matters . In recent years, domestic and foreign researchers have studied the modeling of air temperature, glaciation, maximum temperature and air pollution in various cities. The present study uses a similar approach to study the results obtained from the air temperature of Esfarayn city using actual data of past years paid. The purpose of this study was to predict the air temperature of Esfarayn city with the least error rate. In this research, after collecting meteorological data, effective factors in predicting the temperature of the air were identified and selected for application in the model. Eventually, using the MATLAB programming environment TOOLBOX, neural networks were formed and research modeling was done. In the next phase, this study identifies the most important features for predicting air temperature from received data with genetic algorithm. The results show that neural network modeling with .genetic algorithm is more efficient and less error rate than other networks

Keywords: Neural Network, Multilayer Perceptron, MLP, Genetic Algorithm, Prediction