摘要
目前部分供电公司对台区的线损管理仍采取人工方式,通过对比供售电量并按照预定的阈值对台区进行线损率异常判断。该方法计算精度较低并且不能充分考虑不同台区的线损水平差异,有一定的局限性。为此,提出了基于边缘计算模式的台区线损率计算方法,结合边缘计算模式以及相关技术,将配电台区作为边缘计算节点,采用近邻传播(AP)算法对台区电气特征参数进行分类,并通过LM算法优化的BP神经网络模型对低压台区线损率进行计算,判断该台区是否存在异常。
At present,some power supply companies still adopt a manual method for the line loss management of the station area.By comparing the supply and sales of electricity and making an abnormal judgment of the line loss rate of the station area according to a predetermined threshold,this method has low calculation accuracy and cannot fully consider the different line loss levels.Therefore,a line loss rate calculation method for the station area based on the edge calculation mode is proposed.This method combines the edge calculation mode and related technologies,makes the power distribution station area as the edge calculation node,and uses the Affinity Propagation to classify the electrical characteristics of the station area.The BP neural network model optimized by the LM algorithm is used to calculate the line loss rate of the low-voltage station area to determine whether there is abnormality in the station area.
作者
宋耐超
史雷敏
王小雷
宋珂
惠杰
王槿
邢毅川
SONG Naichao;SHI Leimin;WANG Xiaolei;SONG Ke;HUI Jie;WANG Jin;XING Yichuan(Zhumadian Electric Power Supply Company,State Grid Henan Electric Power Co.,Ltd.,Zhumadian,Henan 463000,China;Xuchang Electric Power Supply Company,State Grid Henan Electric Power Co.,Ltd.,Xuchang,Henan 461000,China;Shanghai University of Electric Power,Shanghai 200090,China)
出处
《上海电力大学学报》
CAS
2020年第5期471-476,510,共7页
Journal of Shanghai University of Electric Power
基金
国网河南省电力公司科技项目(521750200017)。
关键词
边缘计算
近邻传播算法
BP神经网络
台区线损率
edge caculation
affinity propagation
BP neural network
transformer line loss
