Real time nonlinear ARMA model structure identification

This paper addresses the nonlinear autoregressive moving average (NARMA) identification problem in connection with the choice of the model structure (order) and computation of the time varying system coefficients. We introduce an intelligent method that is based on the reformulation of the problem in the standard state space form and the subsequent implementation of a bank of extended Kalman filters, each fitting a different order model. The problem is reduced then to selecting the true model, using the well known multi-model partitioning theory. Simulations illustrate that the proposed method is selecting the correct model order and identifies the time varying model parameters in real time, while it is insensitive to the noise variations.