Based　on　the　current　mainstream　time-series　model,　this　paper　proposes　a　network　model　that　can　be　used　for　short-term　power　load　forecasting.　For　time-series　prediction,　this　paper　proposes　a　hybrid　network　that　integrates　a　Deep　Cross　&　Temporal　convolutional　network　(DCN)　and　a　Temporal　convolutional　network　(TCN),　termed　as　Deep　Cross　&　Temporal　convolutional　network　(DTCN).　This　model　can　automatically　extract　the　effective　information　contained　in　the　data　set,　and　automatically　perform　feature　extraction　and　feature　fusion.　The　validity　of　the　model　is　verified　by　＇Australian　Electric　Load　Data＇.　The　comparison　of　the　results　of　different　tests　in　multiple　periods　shows　that　the　model　can　predict　accurate　power　load　forecasting　results.　It　has　a　clear　structure,　high　generalization,　and　strong　data　tolerance.　©　2022　IEEE.