Consecutive Missing Data Recovery Method Based on Long-Short Term Memory Network

Abstract

This paper describes a consecutive missing data recovery method based on long-short term memory (LSTM) network. The supposed method is fully data-driven and does not depend on system topology and parameters. It exploits the deep learning technique to address missing phasor measurement unit (PMU) data, utilizing the characteristics of LSTM suitable for processing and predicting time series. Simulation results show that, under various PMU missing conditions, the proposed method can maintain a competitively high accuracy.