Abstract:
In manned control systems, a human operator serves the role of the central integrator of system
operations. In autonomous control systems, the control is more challenging for which the control
scheme should be robust to external disturbances and faults, in addition to performing the
required operations efficiently. In practical situations, Unmanned Aerial Vehicles and other
autonomous control systems can face different kinds of faults and failures during operation. A
robust scheme has to be designed for detecting and identifying these faults before they lead to
permanent failure and damage. This research employs a residual generation scheme based on
full-state observer for Fault Detection and Identification (FDI) with application of Eigenstructure
assignment for disturbance de-coupling. A hypothesis testing scheme, Sequential Probability
Ratio Testing (SPRT), tests the generated residual signals efficiently to make decision about fault
in the system. Multiple Model Switching and Tuning (MMST) controllers based on two different
schemes are brought into action at the reconfiguration stage to achieve tracking control of roll
and yaw angle. Simulations are performed using a lateral directional model of an Aerosonde
UAV to show that the propose FDI scheme is robust and efficient and the proposed
reconfiguration controllers efficiently track the desired reference signal.
