Data mining approach for islanding detection of synchronous distributed generation in micro grid

Islanding of DGs in micro grid has created many problems in operation and network management. Thus correct and accurate diagnosis of this phenomenon is one of the most important cases in micro grid protection. Passive methods are the most conventional methods in this case. These methods are based on electrical parameters measurement in the point in which DGs connects to the micro grids. Passive methods based on data mining can be used to limit the non-detection zones and decrease the distinguish time. In this paper islanding is detected by measuring the electrical parameters of connection point, selecting effective features and data mining methods. The effectiveness of these methods are verified by simulation a typical network with PSCAD/EMTDC.