Data-Driven Approach for Fixed-Frame Rotorcraft Mast-Moment Estimation

Abstract

In contrast to the fixed frame, measuring rotor states in the rotating frame of a helicopter rotor is complex and costly and therefore usually unaffordable for small production rotorcraft. In this paper, a data-driven approach to linear observer tuning is proposed and illustrated for predicting the rotor mast bending moments using fixed-frame rigid-body measurements only. The method is demonstrated on a research helicopter, where flight test data is used to identify three possible design models, optimize their respective observer performances, and select the most promising model structures for final application. The resulting observers are assessed in the time and frequency domains using sweep and turbulence data. The results show that data-driven observer tuning allows the effective adaptation of the observer to the reference data. To extend the observer to a larger flight envelope, improved models and a larger database would be required.