Model　predictive　flux　control　(MPFC)　is　an　alternative　control　solution　to　the　classic　model　predictive　torque　control　(MPTC)　for　driving　induction　motors　(IMs)　without　the　complexity　of　tuning　the　weighting　factors.　However,　when　applied　in　motor　drives　fed　by　a　two-level　converter,　it　results　in　higher　flux　and　torque　ripples.　To　improve　its　performance,　this　article　proposes　a　virtual　three-level　MPFC　(V3-MPFC)　method.　The　virtual　vector　method　increases　the　total　number　of　voltage　vectors　crucial　for　improving　the　closed-loop　performances　of　the　model　predictive　control.　To　reduce　the　computational　burden,　direct　torque　control　and　novel　deadbeat　(DB)　control　strategies　are　proposed.　Compared　with　existing　methods,　the　proposed　DB-based　method　reduces　V3-MPFC　to　a　suboptimization　problem　and　still　ensures　that　the　suboptimal　solution　is　equivalent　to　the　global　optimal　solution　provided　by　V3-MPFC.　In　addition,　the　redundant　states　are　exploited　to　lower　the　switching　frequency　(SF)　without　extending　the　cost　function.　The　simulation　and　experiment　results　show　that,　compared　with　existing　methods,　the　proposed　strategy　effectively　reduces　the　flux　and　torque　ripples,　and　requires　a　lower　SF　and　execution　time.