The　accurate　instantaneous　energy　consumption　estimation　for　electric　buses　(EBs)　is　helpful　for　drivers　to　schedule　and　use　EBs　more　reasonably,　which　is　also　essential　for　the　realization　of　electrification,　intelligence,　and　cyber-physical　integration　of　public　transportation　systems.　In　this　paper,　real-world　driving　data　collected　from　ten　electric　buses　in　Beijing　city　are　obtained,　and　the　influencing　factors　of　energy　consumption　from　multiple　sources　such　as　driving-related,　vehicle　status,　and　external　environment　are　extracted.　With　the　extracted　key　influencing　factors,　we　propose　a　new　multi-model　fusion　method　to　estimate　the　1　Hz　energy　consumption　of　electric　buses.　The　results　demonstrate　that　the　proposed　fusion　model　has　a　mean　absolute　percentage　error　of　6.21%,　which　is　superior　to　the　single　model.　Finally,　we　use　the　control　variable　method　for　feature　importance　analysis,　which　provides　valuable　insights　into　the　factors　that　affect　the　energy　consumption　of　EBs　in　real-world　scenarios.