An element-oriented model simplification algorithm based on dynamic similarity

Abstract

This paper presents an element-oriented model simplification algorithm (EO-MSA) in multi-physical domains. It aims to generate a simplified model to release computational loading while maintaining model fidelity. The dynamics of the multi-physical system can be constructed based on a unified bond graph approach. This EO-MSA approach eliminates those bond graph elements which influence the system behaviour insignificantly in the mathematical model. This study proposes two newly-defined indices, dynamic similarity ( DS), and model similarity index ( MSI), to evaluate the similarity of candidates which are order reduced. Ten error-dynamics modules are mathematically constructed and structurally classified for both the DS and the MSI. The reduced model with minimum MSI can be determined using this approach. Quarter-car simulation results demonstrate that this EO-MSA approach is capable of reducing system complexity while maintaining fidelity with satisfactory tolerance.