A.G. Parlos

Comparison of the traditional and the neural networks approaches in a stochastic nonlinear system identification

The comparison between the neural networks and traditional approaches as a nonlinear system identification method is investigated in the aspects of the models performance. Two neural networks models which are of the state space and the input/output model structures are considered as neural networks models. In the traditional methods an autoregressive exogeneous input model and a nonlinear autoregressive exogeneous input model are considered.