Based　on　the　global　gradient-less　stochastic　search　method　of　traditional　particle　swarm　optimization　algorithm,　an　improved　particle　swarm　optimization　algorithm　for　finding　mixed-strategy　Nash　equilibrium　is　proposed　in　view　of　the　shortcomings　of　prone　particle　overflow　and　unconstrained　relationship　between　individual　dimensions.　In　the　improved　evolution　equation,　the　position　function　is　improved　to　conform　to　the　concept　and　properties　of　the　combination　of　strategies,　and　the　proportional　factor　is　introduced　to　ensure　that　the　numerical　value　of　particle　iteration　is　mapped　to　the　solution　space　of　normal　form　game.　Through　the　test　of　several　groups　of　classical　examples,　the　results　show　that　the　algorithm　not　only　improves　efficiency　and　accuracy,　but　also　solves　the　problems　of　divergence　of　particle　and　low　efficiency　of　iteration.　The　experiments　show　that　the　algorithm　has　good　practical　performance.　©　Published　under　licence　by　IOP　Publishing　Ltd.