부산대학교 도서관

A distributed power system, including several microgrids, is always vulnerable to attacks which can result in malfunctioning. In order to protect the system from such attacks, the problem of resilient consensus in microgrids has been discussed in this paper. A network of microgrids as a multiagent system is considered in the presence of misbehaving agents that might make updates arbitrarily. By using the quantization algorithm, each agent will take integer values under directed communication links. The normal agents aim to reach consensus while the malicious ones become isolated. This can be achieved by the use of a special approach of mean subsequence reduced algorithms (MSR) in which all normal agents ignore extreme values received by their neighbors before updating their states. The agents are studied under synchronous and asynchronous updates. Moreover, a controller is suggested that can make the system reach consensus even though, the normal agents only can access their neighbors' values with time-delays. Finally, numerical examples and simulation results are provided.