Abstract: To learn the deep relationship implicit in load data and improve the accuracy of load prediction, this paper presents a new Seq2seq framework based on dual attention and a bidirectional ...