The objective of this thesis is to describe the implementation of an innovative agent-based architecture of controllers for stand-alone DC microgrids. The controllers have to regulate voltage to the required level and manage energy flow in the system. In addition, they should maintain a deterministic time frame on the order of a few tens of milliseconds for a system with tens of power electronic converters with no limitation in the number of events which might happen concurrently. Optimal power sharing ensures minimum transmission and distribution loss while enforcing constraints such as generators’ capacity limits. Multiple agents take part in the process to determine optimum power sharing for the converters. The thesis compares system complexity using numerical analysis of different distributed lookup algorithms based on defined metric values for a standalone DC microgrid including 32 converters. The numerical analysis results aid in choosing a publish-subscribe model as the most efficient and scalable solution for developing agent technology for standalone DC microgrids. Application of publish-subscribe agent-based control is presented for real-time coordination of power converters in a defined microgrid. To test the design, a sample DC shipboard microgrid with eight converters is used as a case study. Results of implementing the agent-based publish-subscribe control system using Java Agent DEvelopment Framework (JADE) are illustrated in the thesis. Simulation results affirm the accuracy of numerical analysis results.
展开▼
