Comparison of Nominal and Incremental Dynamic Inversion for Helicopter Flight Control

Abstract

This thesis presents a comparison and analysis of Nominal Dynamic Inversion and Incremental Dynamic Inversion for helicopter flight control. In particular, the use for manned helicopters was examined. The helicopter model used for simulation is that of the "Flying Helicopter Simulator" (FHS) of the German Aerospace Center (DLR). The FHS is a heavily modified EC 135 that is used as a flying test bench. Both controllers were modified such that the actuator dynamics can be accounted for without appending the actuator model to the model of the plant. In case of the incremental controller, a novel approach for including measurement and derivative filters is proposed. It enables the incremental controller to achieve good tracking, despite slow rotor dynamics. Both controllers were tested under realistic conditions for their tracking performance and their robustness against disturbances and other imperfections. They were examined for the sensitivity to measurement noise, time delays, an attenuation of the output signals and modeling errors. The result of the analysis shows a clear advantage of the incremental controller. Both, the tracking performance and the robustness are significantly higher than in case of the nominal controller. Only with respect to measurement noise is the nominal controller superior over the incremental approach. In addition, it was examined how a direct measurement of rotational accelerations compares to the derivation of signals with the help of filters. For this purpose, the controllers were simulated with both, rotational accelerometers and an Unscented Kalman Filter. Latter estimates the rotational accelerations. The comparison shows that a direct measurement of the rotational accelerations results in significantly increased performance and increased robustness.