Abstract

Discrete event systems (DES) are dynamical systems equipped with a discrete state set and an event driven state transition structure. An event in a DES occurs instantaneously causing transition from one state to another. DES models have emerged to provide a formal treatment of many man-made systems such as automated
manufacturing systems, computer systems, communication networks and air traffic control systems.
In this thesis, we study fault diagnosis in teams of autonomous systems. In particular, one consider a team of two spacecraft in deep space. The spacecraft cooperate
with each other in leader-follower formation flying. Formation flying demonstrates the capability of spacecraft to react to each other in order to maintain a desired relative distance autonomously without human intervention. In the system considered here, instruments (actuators and sensors) may fail and cause error. Because of the communication delays in deep space, each entity should be able to diagnose the failure and decide how to reconfigure itself.
Basically, fault diagnosis in such systems requires information exchange between the autonomous elements of the team. The exchanged information for example may include position and velocity data. Our goal in the thesis is to propose a method for fault diagnosis with reduced information exchange. One solution is to transmit only discrete event information between autonomous systems. Transmission of discrete event data occurs less frequently than the transmission of continuous streams of data. The discrete event data may include high level supervisory commands issued every now and then and discretized values of continuous data that are transmitted only when a continuous-variable data (such as angle or acceleration)
crosses the threshold. The fault diagnosis scheme proposed in this thesis is an adaptation of hybrid fault diagnosis for distributed autonomous systems.
This system is simulated using MATLAB/SIMULINK Software and DECK Toolbox. We examined different maneuvers for spacecraft and investigated the effect of faults on the overall system and the performance of our designed fault diagnoser.