MODULAR MODEL-REDUCTION FOR INTERCONNECTED SYSTEMS

Simplifying the model of a large scale system, which arises from a large number of interconnected subsystems, poses many difficulties. A direct approach requires that detailed models of each subsystem be interconnected to produce a high order composite model which is then simplified. Due to the intermediate use of a very high order composite model, this method is impractical for large systems. Unfortunately interconnecting simplified subsystem models can produce a model with erroneous dominant time behaviour. We detail the high frequency interactions which give rise to this discrepancy. Furthermore, a computationally feasible modular approach to this model reduction problem is discussed. © 1990.