A framework of a multi-agent system for detection and diagnosis of fault on shipboard power systems

A Navy ship power system is a highly complex network that consists of various generators, backup power, power cables, AC and DC loads, and various converters to generate, transmit and supply power at different voltage levels. Several conditions may develop during normal missions that give rise to short circuit current flows at a bus or node, within equipment, or in a current-carrying conductor. These typically result in high-impact contingencies on the electrical subsystems that adversely affect the normal operation of the ship.;A special class of faults resulting in intermittent fault current, termed 'arc' fault current, can occur. These types of faults are not easily detectable and may lead to fires or other unwanted conditions that rapidly degrade the operation of the ship. There is, therefore, urgent need to rapidly detect and effectively isolate these faults automatically since these can lead to forced unwanted shut-down of subsystems and consequently cause permanently damaged equipment and devices. The current techniques for arcing fault detection are limited in their detection speed and do not have autonomous control when attempting fault isolation.;The approach implemented in this work bridges the gap between existing work and the much needed arc fault control. A multi-agent based system (MAS) framework is introduced and it provides a tool for effective detection and diagnosis of faults speedily and also provides fault management via scheduling to maintain ship system operation. Various components of the ship are represented by several agents that work and communicate with each other to maintain the serviceability of the ship. The control schemes in the MAS framework are based on optimization methods and rule-based techniques. A prototype framework for implementation has been developed and tested in this research.;The benefits of this work, when implemented on a large scale, include a more efficient and secure ship system with intelligent control systems that possess added flexibility due to the distributed methodology.