Analysis of Model Uncertainties Using Inverse Simulation

摘要

Helicopter modelling is a complex task as helicopters consist of many subsystems （nacelle, rotor, engine,...） that are described by coupled differential equations. Controller design requires a model that is accurate over a certain frequency range. Usually, system identification is used to derive the models from flight test data and the model complexity depends on the frequency range of interest. Since system identification is an empirical process and the measured data are approximated by low-order equivalent models, high modelling accuracy is not self-evident. To assist system identification, inverse simulation is applied. This type of simulation computes control inputs needed to obtain the measured output variables. The comparison of measured and inverse simulated control inputs provides an additional tool for model validation. In this paper, inverse simulation is applied to identified matrices of the EC135 ACT/FHS. The results show that the model uncertainties can be subdivided into deterministic and stochastic ones. Finally the deterministic modelling errors are described by generic transfer functions which can be used as disturbance input into the plant for controller design.