The　short-term　forecasting　of　passenger　flow　on　the　metro　platform　is　the　decision-making　basis　and　technical　support　for　the　operation　and　management　of　metro.　In　this　paper,　we　developed　an　improved　Kalman　filter　model　to　forecast　short-term　(15　min)　passenger　fluctuations　after　analyzing　the　characteristic　of　metro　platform.　The　model　illustration　was　conducted　on　the　island,　side,　regular,　and　transfer　metro　platform　in　Beijing,　respectively.　Compared　with　the　traditional　Kalman　filter　model,　the　results　showed　that　the　average　absolute　error　of　the　model　was　0.299,　the　mean　square　error　was　34.094,　and　the　equal　coefficient　was　0.923,　indicating　that　the　proposed　model　could　effectively　predict　the　short-term　passenger　on　the　metro　platform.　Compared　with　the　traditional　Kalman　filter　method,　the　model　presented　in　this　paper　can　improve　the　real-time　prediction　accuracy　and　reduce　the　average　absolute　error　by　0.448.　These　insights　will　help　build　more　prosperous　and　sustainable　metro　systems.　©　ASCE.