An Algorithm for the Selection of Physical System Model Order Based on Desired State Accuracy and Computational Efficiency

Abstract

In recent years, algorithms have been developed to help automate the production of dynamic system models, The focus of this development has been to create procedures and algorithms that use modeling metrics for generating minimum complexity models with physically meaningful parameters of dynamic systems (proper models). Existing algorithms, however, do not explicitly address the accuracy of the system variables, and validity of system models. The objective of this paper is to integrate a previously published accuracy and validation algorithm with an energy based (activity index) reduction algorithm to quantify model accuracy as a function of model complexity. In addition, a new metric for selecting the best model based on model accuracy and computational efficiency is introduced. A model of a Class VI International 4700 series delivery truck is used as an illustrative example. The results demonstrate that the proposed model assessment procedure not only provides the means for identifying the proper model based on accuracy and computational constraints, but also provides further evidence that practical systematic tools for improving the modeling process can be developed.