Diformer: A dynamic self-differential transformer for new energy power autoregressive prediction

Chengjie Zhou, Chao Che, Pengfei Wang, Qiang Zhang

Abstract: Highlights•Introduction of the proposed diformer for the new energy power prediction.•Development of a dynamic self-differential transformer that can learn differential properties of time-series data.•Designing the trade loss to balance the periodicity and volatility ratio of time series.•Achieving compelling results on real new energy power data.