摘要
为了能有效、快速、准确地辨识电网用户的窃电行为,结合粗糙集的约简技术,采用基于全局最近邻的约简离群点检测算法,将用户的功率标幺值作为研究对象,计算对象集中的任意点与其他对象之间的距离和,最终通过加权因子挖掘出离群点对象。通过实例验证,所采用的算法能够准确地辨识出窃电用户,从而为电网工作人员利用用电信息采集系统的数据快速有效地进行窃电辨识提供依据。
In order to identify the act of electricity stealing effectively and quickly, an algorithm is used in this paper for outlying reduction based on global neighbor relation inspired by rough set theory. The power standard of users is used as research object in this approach. Then the sum of the distance between the current point to all the other points is calculated. Finally, a neighbor-based outlier factor is formulated to mining the outlier object. The example validates that the used algorithm can accurately identify electricity stealing users, providing a basis to grid staff for the effectively and quickly indentify the electricity stealing with the data of electricity information acquisition system.
出处
《齐鲁工业大学学报》
2016年第6期46-48,共3页
Journal of Qilu University of Technology
基金
国家自然科学基金(61174018)
江苏省电力公司科技项目
关键词
窃电辨识
约简
离群点算法
距离和
electricity stealing identification
reduction
outlier algorithm
distance sum
