Application of Supercontroller to Fighter Aircraft Reconfiguration

Accepted methods of coping with aircraft control system failures and damage involve fault detection, isolation, and control law reconfiguration as a function of determined state. This incurs significant computations, accurate system models, and, with incorrect state determination, degraded performance. The supercontrolter methodology implicitly learns the effect of failures on the aircraft dynamics and efficiently stores the knowledge for real time use. A data base of ideal input/output relationships is created by designing optimal control gains for a range of flight conditions, maneuvers, and failures. An implicit modeling technique determines the underlying, nonlinear model strictly from the observed data. This model is designed as a feedforward polynomial network. This paper discusses a fighter aircraft control system design using the supercontroller methodology. A polynominal network is developed for each control surface to enable the aircraft to perform a range of maneuvers at a single flight condition subject to locked control surfaces. Performance is compared to the baseline aircraft and a reconfigurable aircraft with explicit failure detection and isolation.