基于正则化矩阵补全的用户电量缺失值填补研究

Research on User Power Missing Value Filling Based on Regularization Matrix Completion

针对用户电量采集过程中数据缺失的问题,提出一种基于正则化矩阵补全的用户电量缺失值填补方法。首先,基于原始用户电量缺失值进行特性分析,构造初始化矩阵;其次,搭建一个新的近似矩阵,采用低秩矩阵分解策略将近似矩阵拆分为两个潜在矩阵;最后,引入随机梯度最速下降法对目标参数进行优化,求解近似矩阵模型,完成缺失电量数据的填补。算例采用真实电网数据进行仿真分析,结果表明,所提方法能准确补全用户电量缺失值。

Aiming at the problem of data missing in the process of user power collection,a method for user power missing value filling based on regularization matrix completion is proposed.Firstly,the characteristics are analyzed based on the power loss value of the original user,and the initialization matrix is constructed.Secondly,a new approximation matrix is built,and the low rank matrix decomposition strategy is used to split the approximation matrix into two potential matrices.Finally,the stochastic gradient steepest descent method is introduced to optimize the target parameters and solve the approximate matrix model to fill the missing electricity data.The real power grid data are used for simulation analysis.The results show that the proposed method can accurately complete the power loss value of users.