Type of Document Dissertation Author Huang, Kai Author's Email Address firstname.lastname@example.org URN etd-07142007-231545 Title Shipboard Power System Reconfiguration Using Multi Agent System Degree Doctor of Philosophy Department Mechanical Engineering, Department of Advisory Committee
Advisor Name Title David Cartes Committee Chair Emmanuel Collins Committee Member Justin Schwartz Committee Member Lois Hawkes Committee Member Keywords
- Decentralized Control
- Shipboard Power System
- Multi Agent
- Real Time Digital Simulator
Date of Defense 2007-07-06 Availability unrestricted AbstractThe reconfiguration process for Shipboard Power System reroutes the electric power in the power system in order to achieve certain objectives, such as maximizing service restoration, minimizing power loss, optimizing power dispatch, etc. The reconfiguration process can improve the survivability and reliability of the power system.
Various methods have been applied to the reconfiguration process of power systems. But most of the proposed reconfiguration methodologies are centralized. In a Shipboard Power Systemís centralized reconfiguration approach, a single point of failure may happen if the system lacks redundancy. The Multi Agent System (MAS) technology is recently applied to the applications in power systems. The MAS is composed of agents that are intelligent entities with the capability of problem solving. However, current MAS applications on power system reconfiguration methodologies are topology dependent. Additionally, No MAS based reconfiguration methodology has been proposed for mesh structured power system reconfiguration.
In this dissertation, a completely decentralized MAS based reconfiguration methodology is proposed for Shipboard Power Systems. In this approach, an MAS is proposed for the reasoning of the reconfiguration. In the MAS, each agent is associated with one electric component in the power system. If two electric components in the power system have a connection with each other, two corresponding agents in the MAS are defined as the neighboring agents of each other. The agents in the MAS are restricted to communicating only with their neighboring agents. In the MAS, there is no central controller or coordinator. Each agent works autonomously and independently based on the information it receives from its neighboring agents and the corresponding electric component. Thus, the MAS works in a completely decentralized manner.
In a completely decentralized MAS with loop(s), a message that starts from an agent in the MAS may reach the agent from which it initially originates through the loop. This looping problem may cause a redundant information accumulation (RIA) problem in an MAS. The RIA problem is like the positive feedback loop in a traditional control system and leads to incorrect message flow in the MAS. In this dissertation, a decentralized spanning tree algorithm is proposed. By applying the decentralized spanning tree algorithm, the RIA problem can be avoided in an MAS with loop(s). The proposed decentralized spanning tree algorithm is topology independent and can be applied to an MAS with any topology.
Based on the decentralized spanning tree algorithm, a completely decentralized MAS based reconfiguration methodology is proposed for the mesh structured power system. The objective of the reconfiguration is to maximize the service restoration in the power system. The agents in the MAS make reconfiguration decisions independently and autonomously based on the information from the environment. The loads in the power system are restored based on the priorities of the loads.
The proposed reconfiguration methodology in this dissertation is tested and validated on a shipboard power system (SPS). The SPS is a special power system with some unique characteristics. In the dissertation, a simulation platform is developed for testing and validating the proposed reconfiguration algorithm. In the simulation platform, the SPS is simulated on the real time digital simulator (RTDS). The agents in the MAS are developed in Java agent development framework (JADE). The hardware for developing the agents is the iPAQ, which is a pocket PC developed by HP. An FPGA interface is developed in a simulation platform as the middleware that can set up the communication between RTDS and iPAQs. The illustration and simulation results show the proposed reconfiguration methodology is effective and promising.
Filename Size Approximate Download Time (Hours:Minutes:Seconds)
28.8 Modem 56K Modem ISDN (64 Kb) ISDN (128 Kb) Higher-speed Access Dissertation_Kai_Huang.pdf 1.15 Mb 00:05:20 00:02:44 00:02:24 00:01:12 00:00:06