Reinforcement _recurrent fuzzy rule based system based on brain emotional learning structure to predict the complexity dynamic system

In this study, new approach based on brain emotional Learning process is presented to predict chaotic system more accurate than other learning models. So the main scope of this paper is to reveal the advantages of this learning model that imitate the internal representation of brain emotional learning model to provide a correct response to stimuli to state a purposeful predicting system. The convergence theory is clarified by utilizing the model to predict the complex dynamical Lorenz system. Also the consequence of using this method to forecast such a complex system is compared with obtained results from other related studies that examine other methods such as Locally Linear Model Tree(LOLIMOT) and radial basis function (RBF) Neural network with Orthogonal lest square (OLS ) for predicting the Lorenz chaotic time series. The comparison indicates the superior performance of presented method to make the multi step ahead prediction. Also the effect of noise on the performance of the techniques is also considered. In deed, the learning methods could deal with predicting the future state of complex system with limited training data as well as large data set.