Document Type : Research Paper

Authors

• Samrad Jafarian-Namin ¹
• Davood Shishebori ²
• Alireza Goli ³

¹ Industrial Engineering Department, Faculty of Engineering, Yazd University, Yazd, Iran

² Department of Industrial Engineering, Yazd University, Yazd, Iran

³ Department of Industrial Engineering and Future Studies, Faculty of Engineering, University of Isfahan, Iran

Abstract

The temperature has been a highly discussed issue in climate-changing. Predicting it plays an essential role in human affairs and lives. It is a challenging task to provide an accurate prediction of air temperature because of its complex and chaotic nature. This issue has drawn attention to utilizing the advances in modeling capabilities. ARIMA is a popular model for describing the underlying stochastic structure of available data. Artificial neural networks (ANNs) can also be appropriate alternatives. In the literature, forecasting the temperature of Tehran using both techniques has not been presented so far. Therefore, this article focuses on modeling air temperatures in the Tehran metropolis and then forecasting for twelve months by comparing ANN with ARIMA. Particle swarm optimization (PSO) can help deal with complex problems. However, its potential for improving the performance of forecasting methods has been neglected in the literature. Thus, improving the accuracy of ANN using PSO is investigated as well. After evaluations, applying the seasonal ARIMA model is recommended. Moreover, the improved ANN by PSO outperforms the pure ANN in predicting air temperature

Keywords

• Temperature
• Forecasting
• ARIMA
• ANN
• PSO
• Tehran

Main Subjects

• Forecasting, production planning, and control