Research on Intelligent Identification Method of Power Grid Missing Data Based on Improved Generation Countermeasure Network with Multi-dimensional Feature Analysis

Abstract

AbstractWith the rapid development of data acquisition system in power grid, the data fusion of power grid has become more and more mature. Aiming at the problem of data missing in power grid, this paper proposes an intelligent identification method of power grid missing data based on generation countermeasure network with multi-dimensional feature fitting. Firstly, based on the fluctuation cross-correlation analysis (FCCA), the correlation between missing data and multidimensional features is analyzed, and the feature data set with strong correlation is selected; Secondly, the kernel principal component analysis (KPCA) algorithm is used to map the feature data set into a low-dimensional vector. Finally, the improved generation countermeasure network (WGAN) is used to reconstruct and distinguish the low-dimensional vector, fill in the missing data of power grid and enhance the identification ability of missing data. The simulation results show that the proposed method has higher data filling accuracy than the traditional data filling method.KeywordsFluctuation cross-correlation analysisMultidimensional featuresGenerate countermeasure networkMissing dataNuclear principal component analysisIntelligent identification