Browsing by Subject "fault diagnosis"

Now showing 1 - 1 of 1
  • Thumbnail Image
    Item
    Fault-Tolerant Flight Control Using One Aerodynamic Control Surface
    (2018-06) Venkataraman, Raghu
    Small unmanned aircraft systems (UAS) have recently found increasing civilian and commercial applications. On-board fault management is one of several technical challenges facing their widespread use. The aerodynamic control surfaces of a fixed-wing UAS perform the safety-critical functions of stabilizing and controlling the aircraft. Failures in one or more of these surfaces, or the actuators controlling them, may be managed by repurposing the other control surfaces and/or propulsive devices. A natural question arises in this context: What is the minimum number of control surfaces required to adequately control a handicapped aircraft? The answer, in general, depends on the control surface layout of the aircraft under consideration. For some aircraft, however, the answer is one. If the UAS is equipped with only two control surfaces, such as the one considered in this thesis, then this limiting case is reached with a single control surface failure. This thesis demonstrates, via multiple flight tests, the autonomous landing of a UAS using only one aerodynamic control surface and the throttle. In seeking to arrive at these demonstrations, this thesis makes advances in the areas of model-based fault diagnosis and fault-tolerant control. Specifically, a new convex method is developed for synthesizing robust output estimators for continuous-time, uncertain, gridded, linear parameter-varying systems. This method is subsequently used to design the fault diagnosis algorithm. The detection time requirement of this algorithm is established using concepts from loss-of-control. The fault-tolerant controller is designed to operate the single control surface for lateral control and the throttle for total energy control. The fault diagnosis algorithm and the fault-tolerant controller are both designed using a model of the aircraft. This model is first developed using physics-based first-principles and then updated using system identification experiments. Since this aircraft does not have a rudder, the identification of the lateral-directional dynamics requires some novelty.