In　recent　years,　the　unexpected　factors　leading　to　frequent　accidents　which　causing　heavy　casualties　and　economic　losses.　Therefore,　how　to　make　use　of　computer　technology　to　simulate　the　evacuation　process　has　an　important　practical　significance.　In　this　paper,　we　propose　a　new　model　for　evacuation.　Based　on　the　traditional　particle　swarm　algorithm,　we　revise　the　inertia　weight　of　the　algorithm　to　balance　the　global　and　local　search　ability　according　to　the　aggregation　degree　of　the　population.　Taking　the　interaction　between　individual　and　others,　individual　and　obstacles　or　wall　into　account,　we　introduce　the　social　force　model　to　simulate　this　phenomenon　and　use　it　as　a　collision　resolution　strategy.　Then,　we　combine　the　improved　PSO-based　model　with　modified　Social　Force　Model　to　form　the　velocity　updating　rule.　In　the　end,　through　simulation　and　analyzation,　we　testify　the　modified　model　is　very　effective.　©　2019　IEEE.