1. Introduction;
Most Artificial Intelligent Models such as; Neural Networks(NNs), Wavenets, Fuzzy logic(FL), Fuzzy Controller …, have important applications in real life, but, also have some drawbacks (Jain L.,1997). The trying to resolve ineffective point(s) in some previous methods is to complete its false programming but not as alternatives on it. Such techniques named “Hybrid Systems”. Hybridizing techniques is compiling two (more) techniques models that enhances with other .
A hybrid methods that compiling soft computing methods with wavelet theory have probability to composing two brain s abilities; ability to select an appropriate resolution to the problem description and to somewhat tolerant of imprecision(Synergy,2000). i.e. Computationally, NNs are approaching significantly than others in matching complicated, recognition, vague, or incomplete patterns (Daniel Klerfors,1998). So it has advantage in signal processing, pattern recognition, classifications and predications. In spite of that it have some drawbacks such; generalization, convergence, local minima, to be not efficient in numerous applications .So many attempts to resolve these drawbacks for example; adding a momentum term into adjustment equation to increase the converging speed and to avoid a local minima as in reference(Zurada J.,1997), or using a wavelet transform (WT) in neural networks in feature extraction, or to reduce the dimension of the inputs.