Uncertainties in modeling helicopter dynamics: A framework applied to system identification results

Abstract

Helicopters are heavily coupled, unstable and their modeling is often a difficult nut to crack. This often makes control design for the helicopter difficult. However, a goal for the near future for the EC135 Active Control Technology / Flying Helicopter Simulator (ACT/FHS) is high-performance control to assess the in-flight simulation capability. This stimulates the requirements for the feedback control design which is based on linear models derived by system identification. The core problem for this type of design is often a mismatch of the closed-loop behavior to measured flight test data. This paper focuses on the question how the design can be assisted to improve the plant's model, so that the gap between design and flight can be reduced. The framework relies on linear models derived by system identification. These linear models are stitched together to describe not only the operating point but also the flight envelope of interest. The linear models are also extended by non-physical models that describe the uncertainties. The model of the plant finally consists of two parts: the physics-based, stitched model and the non-physical uncertainties. Compared to the classical techniques system identification and nonlinear modeling, the framework proposed in this paper can be applied quickly and indicates possible model improvements. The paper shows the application of the whole framework to ACT/FHS flight test results.