Parameter Identification Using the Model Fitting Method

Abstract

Electric powertrains for passenger cars exhibit a different vibration behavior compared to conventional powertrains. These differences are characterized by significantly higher frequencies and, in particular, higher critical frequencies. This characteristic poses new challenges in regard to the testing of electric powertrains on powertrain test benches. Furthermore, the requirements relating to test benches and the methods are also changing in this case. The vibration behavior in particular is worth mentioning here. Vibration measurement technology is already used on powertrain test benches for machine and device under test (for short DUT) monitoring. In order to further develop the existing test bench technology and thus make it fit for the future, it is necessary to know information regarding the energy transfer paths of mechanical vibrations from unbalance excitations of the test bench machines.

Starting from a simple model of a test bench machine as a multi-mass spring-damper system, this paper uses the model fitting method to demonstrate the determination of model parameters from vibration measurement data. The model fitting method is based on the least squares method - it is a minimization problem. The system’s model equations are determined and the experimental observations are measured. The differential equations of motion of the multi-mass spring-damper system are used as the model equations. For the solution of the differential equations, several approaches were compared and finally one approach was implemented.

Based on an existing simplified vibration model of a test bench machine, sought-after model parameters could be identified and plausibilized from vibration measurement data using the model fitting method.