Instant　delivery　is　an　important　part　of　urban　logistics　distribution,　which　realizes　point-to-point　distribution　between　merchants　and　customers.　During　the　peak　period　of　orders,　instant　delivery　is　a　large-scale　variable　NP-hard　combinatorial　optimization　problem,　which　increases　the　difficulty　and　complexity　of　scheduling　greatly.　To　solve　the　large-scale　vehicle　routing　problem　of　instant　delivery　in　peak　periods,　a　knowledge-driven　ant　colony　optimization　(KDACO)　algorithm　is　proposed　in　this　paper.　First,　the　knowledge　base　is　established　to　guide　evolutionary　search,　including　the　knowledge　of　order　priority　and　the　feature　knowledge　of　feasible　schemes.　Second,　the　pheromone　supplementation　strategy　is　designed　based　on　the　knowledge　of　order　priority,　enhancing　the　guiding　ability　of　the　pheromone　table.　Third,　the　adaptive　evolutionary　operator　is　designed　based　on　the　feature　knowledge　of　feasible　schemes,　improving　the　optimization　efficiency　of　the　algorithm.　Finally,　numerical　experiments　on　extensive　classical　datasets　show　that　the　proposed　KDACO　can　obtain　superior　performance　to　other　state-of-the-art　algorithms　in　the　instant　delivery　peak　period.　&　COPY;　2023　Elsevier　B.V.　All　rights　reserved.