On the Use of Nonlinear Autoregressive Moving Average Models for Simulation and System Identification

Efficient simulation models of nonlinear systems are useful for analyzing vibrational systems and designing and implementing control systems. Such models can also be used to produce estimates of the physical system parameters. NARMAX models can be used to simulate a wide range of nonlinear systems. Development of such models requires identification of the system nonlinearities and model structure and estimates of the system model parameters. Computation time is directly related to the number of terms which increases dramatically with the model order. The apparent significance of a term changes with its placement in the model structure which makes it difficult to determine its true significance. The benefits and drawbacks of this modelling procedure are demonstrated through examples.