Date of Graduation


Document Type


Degree Type



Statler College of Engineering and Mineral Resources


Lane Department of Computer Science and Electrical Engineering

Committee Chair

Ali Feliachi


The objective of this thesis is to develop an agent based reconfiguration algorithm for power system application with termination detection and to provide communication architecture for these agents. The underlying maximum flow algorithm based on graph theory has proven the concept of autonomous distributed reconfiguration using multiagent systems. The new rule-based algorithm takes into account the requirement of bidirectional power flows, and special power routing for DC devices. Even though the algorithm is self-stabilizing and agents become idle after finite time, implementation of the solution within the actual power system requires that all agents be in an idle state. The process of detecting the idle state of all agents in a distributed environment is called termination detection. After surveying various methods for termination detection, a modified method that is suitable for reconfiguration algorithm is developed. The method uses wave algorithm with tree topology and provides the ability of detecting termination to each of the agents. These reconfiguration agents can perform satisfactorily only if they can communicate and share information with other agents in the system. Thus, it is extremely important to have a uniform agent communication language and messaging scheme among all the agents. With the Controller Area Network as hardware backbone for communication, a new message structure and communication protocol is developed based on CANOpen and standards specified by Foundation for Intelligent Physical Agents. The communication architecture provides easy interface between reconfiguration agents and other agent based systems. The designed algorithms are applied to the reconfiguration of power system on all electric ships. The system is simulated on multiple computers and includes a command and control center to monitor and supervise the entire operation of the system. The integrated system is tested for specific scenarios and statistics are recorded to analyze the algorithm performance and robustness. The evaluation of the system for correctness of reconfiguration solutions and the time required to complete the reconfiguration show promising results.