Pilot-induced oscillation detection and mitigation

The aim of this thesis is to develop a real time PIO detection and mitigation system that consists of a detector based on short time Fourier transform(STFT) and autoregressive model(ARX) with exogenous inputs, together with an adaptive controller based mitigation system. The system not only detects the traditional PIO characteristics but also focuses on the trend of pilot behaviour by calculating the rate of change in the open loop crossover frequency. In the detection system, a sliding windowed STFT method was applied to identify the frequency and phase characteristics of the system via processing the signal of pilot input and aircraft state. An ARX model was also applied to get the rate of change of the crossover frequency. After detection, a PIO cue was shown on the primary flight display. A scheduled gain controller was coupled to provide PIO mitigation by varying stick input gain. Compensatory and tracking tests for the evaluation of this system were performed using a quasi-linear Boeing-747 aircraft model including nonlinear command gearing and actuator rate-limiting. Bandwidth and Gibson criteria were used to design PIO prone control laws for system evaluation experiments. Results from PIO tests conducted on desktop PCs were presented. These were analyzed and compared with those obtained from implementing the Real-time Oscillation Verifier module available in literature.