Safety and reliability of air vehicles is of the utmost importance. This is particularly true for large civil transport aircraft where a large number of human lives depend on safety critical design. With the increase in the use of unmanned aerial vehicles (UAVs) in our airspace it is essential that UAV safety is also given attention to prevent devastating failures which could ultimately lead to loss of human lives. While civil aircraft have human operators, the pilot, to counteract any unforeseen faults, autonomous UAVs are only as good as the on board flight computer. Large civil aircraft also have the luxury of weight hence redundant actuators (control surfaces) can be installed and in the event of a faulty set of actuators the redundant actuators can be brought into action to negate the effects of any faults. Again weight is a luxury that UAVs do not have. The main objective of this research is to study the design of a fault tolerant flight controller that can exploit the mathematical redundancies in the flight dynamic equations as opposed to adding hardware redundancies that would result in significant weight increase. This thesis presents new research into fault tolerant control for flight vehicles. Upon examining the flight dynamic equations it can be seen, for example, that an aileron, which is primarily used to perform a roll manoeuvre, can be used to execute a limited pitch moment. Hence a control method is required that moves away from the traditional fixed structure model where control surface roles are clearly defined. For this reason, in this thesis, I have chosen to study the application of model predictive control (MPC) to fault tolerant control systems. MPC is a model based method where a model of the plant forms an integral part of the controller. An optimisation is performed based on model estimations of the plant and the inputs are chosen via an optimisation process. One of the main contributions of this thesis is the development of a nonlinear model predictive controller for fault tolerant flight control. An aircraft is a highly nonlinear system hence if a nonlinear model can be integrated into the control process the cross-coupling effects of the control surface contributions can be easily exploited. An active fault tolerant control system comprises not only of the fault tolerant controller but also a fault detection and isolation subsystem. A common fault detection method is based on parameter estimation using filtering techniques. The solution proposed in this thesis uses an unscented Kalman filter (UKF) for parameter estimation and controller updates. In summary the main contribution of this thesis is the development of a new active fault tolerant flight control system. This new innovative controller exploits the idea of analytical redundancy as opposed to hardware redundancy. It comprises of a nonlinear model predictive based controller using pseudospectral discretisation to solve the nonlinear optimal control problem. Furthermore a UKF is incorporated into the design of the active fault tolerant flight control system.
展开▼
