A　two-step　predictive　current　control　method　based　on　a　predictive　evaluation　function　and　a　sliding　mode　observer　is　proposed　to　improve　the　current　bandwidth　and　dynamic　performance　of　servo　systems　with　permanent　magnet　AC　servo　system.　Firstly,　a　predictive　evaluation　function　is　designed　to　obtain　the　optimal　voltage　vector　and　to　attenuate　the　current　fluctuation　in　the　existing　two-step　predictive　current　control　caused　by　system　noise.　Then,　a　sliding　mode　observer　is　designed　based　on　a　stator　voltage　equation　with　parameter　perturbation　and　a　new　switch　function.　When　the　gain　matrix　of　the　observer　is　selected　properly,　the　observed　equivalent　current　error　converges　to　the　current　prediction　error　caused　by　mismatch　of　the　electrical　parameters,　and　the　corresponding　voltage　is　compensated　and　corrected　according　to　the　equivalent　current　error.　The　q-phase　current　step　response　experiment　with　the　amplitude　of　3　A,　the　current　loop　bandwidth　test　and　the　controller　robustness　test　are　carried　out　on　a　motor　control　performance　testing　platform.　Results　show　that　the　proposed　control　method　effectively　improves　the　dynamic　response　performance　and　the　bandwidth　of　the　current　loop　of　permanent　magnet　synchronous　motor　(PMSM),　and　has　sufficient　robustness.　©　2020,　Editorial　Office　of　Journal　of　Xi＇an　Jiaotong　University.　All　right　reserved.