Neural Network Compensation Strategy for Preventing Pilot-Induced Oscillations

We propose to develop an automated Pilot Induced Oscillation (PIO) detection and compensation system. This will use artificial neural networks to increase speed and the envelope of operation. PIO is caused when a pilot overcompensates during a "high gain" task such as landing, takeoff, refueling, dropping stores, or tracking. It can also be caused by inefficiencies in the inner loop controllers, such as time delays and phase and rate limits. PIO represents a significant danger to aircraft and has resulted in a number of high profile accidents, such as the crash of the YF-22. Some of the most experienced pilots have had PIOs. Because of their proven ability of fast and accurate pattern recognition and new research into the stability of artificial neural network based control systems, neural networks can be used to develop a system superior to current systems such as the Space Shuttle PIO filter and the JAS-39 system.