Research on Measurement Data Missing Reconstruction of Distribution Network Based on Improved LSGAN
A modified LSGAN missing data reconstruction model is proposed to address the issues of low efficiency and poor performance in current power measurement missing data reconstruction.On the basis of analyzing the shortcomings of existing generative adversarial networks,an improved least squares generative adversarial network is designed to fully learn the internal connections between data.In order to provide a more stable training,faster computational convergence,and higher data quality for the network,the objective function in the traditional GAN model was changed from the cross entropy loss function to the least squares loss function,and a new distance metric was adopted.In the experimental stage,compared with GAN,CGAN,and LSGAN models,the proposed improved LSGAN model has the best overall performance indicators.The experimental re-sults validate the practicality and excellent performance of the proposed model,which can provide some reference for the devel-opment of missing data reconstruction in power measurement.
distribution networkmissing datadata reconstructiongenerate adversarial networksleast square
NETL
NSTL
万方数据
