Predicting solar radiation using fuzzy systems

and stochastic process


R. Iqdour*, A. Zeroual

Department of physics, Cadi Ayyad University, Faculty of Sciences, Semlalia,

BP 2390, Marrakesh 40000, Morocco.

* Corresponding author. E-mail:

Received: 30 December 2005; revised version accepted: 02 May 2006



     In this work we use two models for predicting the daily solar radiation data. The first model is based on the fuzzy systems of Takagi-Sugeno. The second one uses a linear Auto-Regressive Moving Average stochastic process (ARMA). These models manipulate the information in different way. Indeed, the ARMA approach supposes that the time series is a signal generated by a linear stochastic process, whereas the fuzzy systems are non linear and suppose that the observed signal result from a determinist process. A comparison between the two used models has performed. The obtained results show that the Takagi-Sugeno fuzzy systems are not only more accurate than the linear models but provide also some qualitative information.


Keywords: Daily solar radiation; Predicting; Stochastic process; Takagi-Sugeno fuzzy systems.

© 2015