Microgrid　is　an　effective　means　to　integrate　distributed　generations　and　manage　their　power　flows.　Since　the　high　penetration　rate　of　DCs　in　microgrid　and　its　capability　to　operate　in　grid-connected　mode　or　islanded　mode,　the　fault　patterns　of　microgrid　are　not　that　obvious　as　those　shown　in　the　traditional　low　voltage　distributed　networks.　Thus　an　extensive　knowledge　of　fault　transient　characteristics　is　beneficial　to　design　effective　protection　schemes　for　microgrid.　Since　conventional　methods　have　difficulties　in　analyzing　transient　signals,　discrete　wavelet　transform　is　employed　in　this　paper　to　identify　fault　patterns　from　fault　current/voltage　signals.　Aiming　to　provide　a　solution　to　the　problems　arise　from　applying　wavelet　in　microgrid　with　short　line　segments,　this　paper　proposed　an　improved　wavelet　transform　method　that　using　Daubechies　complex　wavelet　with　cubic　spline　interpolation.　A　modified　benchmark　microgrid　network　is　modeled　in　Matlab/Simulink　to　verify　the　proposed　method.　Four　types　of　fault　cases　in　bus-bar　and　feeder　are　investigated　respectively,　and　the　simulation　results　indicate　the　unique　fault　patterns　in　the　form　of　modulus　maxima　and　wavelet　energy.