A novel scheme to estimate the model order of physical systems

A method to estimate the model order of physical systems is proposed. It is assumed that the region where the poles of a physical system are located is known. This a priori knowledge is generally obtained from the physical laws governing the physical system. Various model orders are chosen using the conventional model selection criteria. For each order, a model of the physical system is identified. If the poles of the identified model are not located in this region, this model is assumed to include extraneous poles and is rejected. This process is continued till the estimated model does not include extraneous poles. The proposed scheme is extensively evaluated on physical systems and the results are compared with those of the conventional schemes such as Akaike Information Criterion (AIC).