It　is　a　challenge　to　control　the　transient　air　fuel　ratio　of　gasoline　engines　accurately.　In　this　work,　the　traditional　PI　controller　was　used　to　control　the　transient　air　fuel　ratio　by　using　the　estimated　signal.　To　verify　the　validity　of　the　control　strategy,　a　single　cylinder　gasoline　engine　model　was　built　with　GT　(Grand　Touring)　-Power.　Based　on　this,　the　simulation　model　for　controlling　the　air　fuel　ratio　of　the　gasoline　engine　was　built,　using　GT-Power/Simulink.　The　neural　network　was　programmed　with　S-functions.　The　simulation　results　showed　a　fair　self-adaptability　of　this　control　strategy,　which　could　effectively　avoid　enormous　calibration　experiments　that　are　needed　in　the　transient　air　fuel　ratio　control　at　present.