Detection of Nonlinear Dynamic Systems by Analyzed Neural Fuzzy Based
Mahdi Koohdaragh · 2012
In this paper, a powerful fuzzy neural model is used for dynamic modeling of nonlinear discrete time systems. The presented model is based on NARMAX 1 model. Just the same, for increasing efficiency and accuracy of estimation, genetic algorithms (GA) is used for determining the number of membership functions and also optimization of neural network weights that is the parameters of fuzzy membership functions. The structure of the presented model is based on analysis of FIS 2 . At the end, to show the performance of the presented model, the data related to Box-Jenkins furnace is used to compare other previous methods with the presented method.