Model reduction of a parametrically excited drivetrain

The complexity of engineering systems is continuously increasing, resulting in mathematical models that become more and more computationally expensive. Furthermore, in model based design, for example, system parameters are subject of change, and therefore, the system equations have to be evaluated repeatedly. Hence, there is a need for providing reduced models which are as compact as possible, but still reflect the properties of the original model in a satisfactory manner. In this contribution, the reduction of differential equations with time-periodic coefficients, termed as parametrically excited systems, is investigated using the method of Proper Orthogonal Decomposition (POD). A reduced model is set up based on the solution of the original system for a certain parametric combination resonance of the difference type, resulting in an additional stability margin of the trivial solution. It is shown that the POD reduced model approximates the stability behavior of the original system much better than a modally reduced model even if system parameters are subject of change.