摘要
文章详细阐述了异常量测数据筛选,提出了基于深度学习的电力调度流异常识别方法,结合算例分析展开研究,结果表明,异常情况下所有节点均保持了较好的滤波效果,平均绝对百分比误差与均方根误差均处于较低的水平,具有较高的滤波准确性与稳定性。
This paper elaborates on the screening of abnormal measurement data,and proposes an abnormal identification method for power dispatching stream based on deep learning.Combined with the analysis of examples,the results show that all nodes maintain a good filtering effect under abnormal conditions,and the Mean Absolute Percentage Error and Root Mean Squared Error are at a low level,with high filtering accuracy and stability.
作者
赵玉兴
ZHAO Yuxing(State Grid Xizang Changdu Power Supply Company,Changdu 854000,Xizang,China)
出处
《光源与照明》
2025年第8期224-226,共3页
Lamps & Lighting
关键词
电力调度流数据
数据异常
识别方法
卷积神经网络
状态估计
power dispatching stream data
data anomaly
identification method
convolutional neural network
state estimation
