Fault detection, isolation and reconfiguration for autonomous aircraft
Thesis (MEng)--Stellenbosch University, 2016.
ENGLISH ABSTRACT: This research is a continuation of a series of projects done at the Electronic Systems Laboratory (ESL) that deal with the design and development of Fault Tolerant Control Systems. The focus of this research is the design and development of a fault diagnostic system that allows an autonomous aircraft to automatically diagnose faults and reconfigure itself to maintain stable flight conditions. The motivation behind this research is to have the developed system be easily reconfigurable making it possible to use it in other UAVs at the ESL. The Meraka Modular UAV, developed at the Council for Scientific and Industrial Research (CSIR), was chosen as a test-bed for the Fault Detection, Isolation and Reconfiguration (FDIR) System. A nonlinear mathematical model of the flight dynamics of the Meraka Modular UAV is derived, linearised and discretised for control and FDIR system purposes. A Failure Mode and Effects Analysis is performed in order to characterise and identify critical subsystem components as a means of guiding the fault modelling and identification process. The Hybrid Diagnosis Engine (HyDE), developed at NASA’s Ames Research Center, and an Approximate Input Reconstruction Algorithm are used together to give the Meraka Modular UAV fault diagnostic capabilities. An expandable FDIR architecture is developed to facilitate the addition of more FDIR methods in an effort to allow the fault diagnostic system’s functionality to be enhanced easily. The architecture was integrated based on the design of a simplified Integrated Vehicle Health Management (IVHM) System. This system consists of a Diagnostic Agent built into the ESL’s custom Ground Station, the Meraka Modular UAV, and a Diagnostic System that runs on a Model B+ Raspberry Pi. A high fidelity software in the loop simulation environment was used to test the integrated fault diagnostic system to ensure that it will function as expected in the field when used by the Meraka Modular UAV and Ground Station Operator to perform diagnoses. The fault diagnostic system is tested extensively by deliberately failing the UAV’s actuators during simulated flight. The performance of the system was then verified by using flight test data collected by the Meraka Modular UAV during flight missions. When compared with the associated FDIR research, the FDI performance of the fault diagnostic system is found to be sensitive to the use of filters and relatively agnostic to actuator excitation. For the Meraka Modular UAV in particular, it is found that the fault diagnostic system performs as expected in the presence of disturbances and noise and improvements can be made by incorporating additional points of observation.
AFRIKAANSE OPSOMMING: Geen opsomming beskikbaar