Adaptive Neural Control for a Tolerant Fault System

Fault tolerant control of complex systems has received considerable attention in the latest years and it has been considered to a variety of systems with high critical and stringent requirements of dependability, including industrial plants, Unmanned Air Vehicles (UAVs) and medical equipments, among others. In this work, adaptive and robust methods are combined to yield a control scheme able to deal with uncertainties, parameters variations and some classes of faults on the system. The key idea is to add to the robust controller output a second control signal generated by an adaptive artificial neural network capable of compensating a wide range of perturbations, uncertainties, and faults, so that adequate performance characteristics of the main control loop is maintained. As an example, a controller is designed for a four rotor UAV model.