Design and simulation of fault tolerant flight control based on a physical approach

This paper presents a study on fault tolerant flight control of a benchmark aircraft model, using a physical modular approach. Reconfiguring control is implemented by making use of adaptive nonlinear dynamic inversion for autopilot control. The weakness of classical nonlinear dynamic inversion, its sensitivity to modeling errors, is circumvented here by making use of a real-time identified physical model of the damaged aircraft. With help of the Boeing 747 benchmark simulation model, including the realistic component as well as the structural failure modes, it is possible to analyze the damage accommodation capabilities of the considered approach. In failure conditions, the damaged aircraft model is identified in real time by the two-step method and this model is applied subsequently to the model-based adaptive nonlinear dynamic inversion routine in a modular structure, which allows flight control reconfiguration on-line. Pseudo Control Hedging is included to guarantee that no unachievable reference commands are given to the system, which would otherwise result in control effector saturation. Various simulated reconfiguration test results are shown for damaged aircraft models. These results indicate satisfactory failure handling capabilities of this fault tolerant control setup, for component as well as structural failures.