On　the　basis　of　its　working　principle　and　topology,　a　mathematical　model　of　three-phase　AC-DC-AC-type　solid　state　transformer　is　built.　The　control　strategies　of　the　AC-DC　rectifier,　DC-DC　converter　and　DC-AC　inverter　are　studied　respectively.　Due　to　the　nonlinearity　and　time-varying　uncertainty　in　a　SST　system,　it　is　difficult　to　get　a　precise　model,　resulting　in　the　unsatisfactory　control　results　gained　via　adopting　the　traditional　PID　regulator.　To　enhance　the　robustness　and　adaptability　of　the　SST　system,　a　neural　network　PID　controller　is　designed　which　modifies　the　proportional　coefficient,　integral　coefficient　and　differential　coefficient　in　the　real-time　online　adjustment　process.　A　simulation　model　is　constructed　based　on　Simulink　and　the　results　show　that　the　SST　model　can　achieve　unit　input　power　factor,　constant　output　voltage　and　stability　against　load　variation　and　grid　disturbance.