Development of a Neural-Network (NNET) Based Engine Model for a Two-Stroke Engine of a Rotary Wing UAV

In general, control laws developed for stabilizing rotary wing platforms are dependent on the constant rotor speed assumption that is maintained by a dedicated rotor speed controller, known as governor. Likewise, textbooks on the helicopter theory and most of the technical publications present existing theories and develop new ones under the constant rotor speed assumption, relying on a perfectly working governor. Exceptions to this assumption appear in the literature where the use of rotor transcends the conventional form, such as tiltrotor aircrafts or slowed rotor concepts. In these examples rotor speed degree of freedom had to be included for the stability and performance analysis. In the literature necessity of the rotor speed degree of freedom is shown even for a conventional helicopter in which case the governor did not work well enough to consider the rotor speed constant in the simulations. In case of small size rotary wing UAVs, maintaining a constant rotor speed may not be that much possible or in another perspective the amount of rotor speed variation might have significant effects on the flying qualities of a small size RW-UAV whereas such variations may not pose a problem for the most conventional helicopters. In this regard this study proposes Neural-Network Based engine modeling for a RW-UAV in order to account for the rotor speed degree of freedom in flight dynamics analyses. A test case study for the implementation is also presented.