In　view　of　the　difficulty　in　controlling　the　transient　air　fuel　ratio　of　electronic　controlled　LPG　engines　accurately,　a　kind　of　air　fuel　ratio　control　strategy　that　combined　the　modified　Elman　neural　network　and　the　traditional　PI　controller　was　put　forward　in　this　paper.　The　Elman　neural　network　was　used　to　estimate　the　air　fuel　ratio　signal　without　transfer　delay.　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　simulation　model　for　controlling　the　air　fuel　ratio　of　the　electronic　controlled　LPG　engine　was　set　up,　using　GT-Power/Simulink.　The　simulation　results　showed　a　fair　self-adaptability　of　this　control　strategy,　which　could　keep　the　controlled　error　of　the　transient　air　fuel　ratio　within　±　5%.