Redesign and evaluation of an equation-based model reduction method in OpenModelica

Abstract

Even in times of a continuously growing computational power and miniaturisation of computers and control devices, the demand for model reduction techniques persists. Detailed models of physical systems typically result in rather complex dynamic models. In the design process, when the design space has been narrowed down to a concrete design, it is often desirable to replace these detailed models by simpler ones. The reduced models are expected to run multitudes faster while preserving important properties of the original model required for control design and validation purposes. For evaluation purposes, the reduction method is applied to a simple sample model (two mass spring damper). The reduced model generated by the reduction algorithm is compared with the results of a theoretically derived reduced model. In the conclusions, deficits of the current implementation are discussed. Proposals for further improvements of the methodology for extended control design use cases are given in the outlook.