Pilot Induced Oscillations (PIO) are described as pilot-aircraft dynamic couplings which can lead to instability in an otherwise stable system. Previous and ongoing research has attempted to explain, predict, and avoid such oscillations. In contrast to other research, this effort backs away from pilot models and PIO avoidance and focuses on the characteristics of the pilot before, during, and after a PIO. Often, PIO''s can be explained by limit cycles occurring in a non-linear system where the non-linearities cause a sustained, constant amplitude oscillation. The primary instigators in such a PIO are usually a non-linear element (i.e. rate limit saturation) and a trigger event (i.e. pilot mode switching or increased pilot gain). By performing analysis in the frequency domain, determining such oscillations becomes easier. Using spectrograms and power spectral density functions, the frequency content of a signal in the pilot-aircraft system can also be investigated.
An F-14 flight test was recently performed where the hydraulic system was modified to determine the feasibility of trying to recover the aircraft (land on carrier) during such an extreme hydraulic failure. During testing, a severe PIO occurred because of the tight tracking task used during aerial refueling. While performing spectrograms and power spectral analysis, an increase in power concentration at the PIO frequency was observed.
With a linear approximation of the F-14 aircraft dynamics, a closed-loop system containing the aircraft, actuator, and pilot dynamics is developed so that limit cycle analysis can be performed. With stable limit cycle solutions found possible, a pilot-in-the-loop simulation is performed to verify the pilot model used in limit cycle analysis. Using the flight test data, limit cycle analysis, and pilot-in-the-loop simulation, a connection between variation in pilot behavior and PIO predicted by the increase in power concentration is investigated.
The resulting connection showed that an increase in pilot gain along with a transition from observing pitch attitude to pitch rate are the possible trigger events causing the PIO. The use of spectrograms as a PIO predictor is shown to be possible, provided the necessary calculations can be completed in real-time.
