Proposed Neural Network with FFT Transfer Function to Estimate Loranz Dynamical Map

The aim of this paper is to design a feed forward artificial neural network (Ann) to estimate three dimensional Loranz dynamical map by selecting an appropriate network, transfer function and node weights to get Loranz dynamical map estimation. The proposed network side by side with using Fast Fourier Transform (FFT) as transfer function is used. For different cases of the system, Deterministic, chaotic and noisy, the experimental results of proposed algorithm will compared empirically, by means of the mean square error (MSE) with the results of the same network but with traditional transfer functions, Logsig and Tagsig. The performance of proposed algorithm is best from others in all cases from Both sides, speed and accuracy.