MDRAE: An Attention Mechanism-Based Autoencoder for Missing Data Recovery in Smart Grids

Smart grids have rapidly developed in recent years, and a large amount of data has been collected and become the basis for decision-making of power grid operation. To ensure the safe and stable operation of the power grid, how to improve the data quality of the collected data has become an important research topic. Therefore, an attention mechanism-based autoencoder is proposed in this paper for missing data recovery to improve data quality in smart grids. The proposed method named missing data recovery autoencoder (MDRAE) contains two parts of an encoder and a decoder, and the attention mechanism is integrated into the encoder part to better extract the key information of the data. The effectiveness of the proposed method is verified by conducting experiments on two datasets including industrial load data and residential load data. The experimental results show that the model performance of the proposed method is better than that of the two benchmarks, and the proposed method can recover the missing data accurately.