基于数据挖掘的分布式系统孤岛检测方法

Application of Data Mining in Island Dectection of Distributed Generation

分布式发电系统并网运行时处于孤岛状态影响电力系统安全正常运行,反孤岛设备必须在可以接受的时限内把孤岛检测出来。该文主要采用数据挖掘技术中的C4.5决策树来作为分布式发电系统的孤岛检测方法。首先离线建立精确的系统运行模型,然后用该模型建立C4.5决策树,最后采用建好的C4.5决策树来进行在线的孤岛检测。在整个孤岛检测过程中C4.5决策树有能力进行自完善,而且可以最小化检测区域。文中使用Matlab仿真验证了C4.5决策树进行孤岛检测的可行性。

Distributed generation system operating in connection with grid under the island state will impact the normal operation of power system security.Anti-islanding unit must detect out island under the acceptable time limit.This paper uses C4.5 decision tree in the data mining technology as a distributed generation system islanding detection method.First,setting up a accurate system off-line model,which is used to build the C4.5 decision tree to detect the island on line.During the island detecting process,the C4.5 decision tree have the ability to self-improve,and can minimize the detection area.Simulation result using C4.5 decision tree for islanding detection is proved to be feasible.