Neural Networks Based Aircraft Fault Tolerant Control," presented at the

The purpose of this communication is to deal with the case in which an aerodynamic actuator failure occurs to an aircraft while it has to perform guidance maneuvers. The problem considered deals with the reallocation of redundant actuators to perform the required maneuvers and maintain the structural integrity of the aircraft. A Nonlinear Inverse Control technique is used to generate online nominal moment along the three axis of the aircraft. Then, taking into account all material and structural constraints as well as the redundant effects from other actuators, a Mathematical Programming problem to be solved online is considered. The proposed solution method is based on dynamic neural networks which appears to provide acceptable response times. Simulation results are displayed and new development perspectives are discussed. Nomenclature α = angle of attack, rad β = sideslip angle, rad δp, δq, δr = roll, pitch, yaw actuator deflections, rad p, q, r = roll, pitch, yaw rates, rad/s V = airspeed, m/s I