In the context of significant changes in current electricity load characteristics,in order to achieve efficient response to power loads with obvious seasonal peak valley differences,effectively improve the operational economy of distribution transformers,and respond to the call for smart grid construction,this paper proposes a power load forecasting method based on encoder decoder prediction model from the perspective of regulating transformers.Based on this,the applicability analysis and energy-saving evaluation of regulating transformers can be carried out,which can provide important support for energy conservation,loss reduction,and precise investment in smart grids.