Lucas - Electric & Computer Eng.Dep. Tehran University
Ganjavie - Electrical Eng. Dep. Ferdosi University of Mashad
Javidi - Electrical Eng. Dep. Ferdosi University of Mashad

This paper uses approximate analogical reasoning approach,,first proposed by Turksen and Zhong, for load foreasdng. This approximate reasoning approach has neural network structure. Therefore, we refer to it as analogical Fuzzy neural network (ANFNN). Our invstigations into the proposed method demonstrate that the ANFNN has better accuracy in comparison with conventional F neural network (FNNs). furthermore in this study, the ANNFF's rulebase is four times smaller than conventionl FNN's rulebase and consequently the AFNN's training is faster that FNN's training.