In　order　to　solve　the　problem　of　insufficient　smoothing　ability　and　excessive　smoothing,　this　paper　proposes　a　method　combining　exponential　smoothing　short-term　wind　power　forecasting　with　wavelet　packet　decomposition.　According　to　the　optimal　decomposition　level,　the　output　power　in　line　with　the　grid　connection　standard　can　be　obtained.　To　improve　the　service　life　and　reliability　of　HESS,　this　paper　further　introduce　the　concept　of　standby　storage.　Then,　this　work　propose　a　method　combining　CEEMDAN　with　NEE　and　NSE　to　achieve　the　optimal　power　allocation　instructions　of　the　HESS.　According　to　the　residual　uniqueness　and　adaptive　frequency　division,　the　IMF　and　residual　components　of　different　orders　can　be　achieved.　Then,　the　cut-off　point　with　the　largest　proportion　of　energy　will　be　obtained　by　NEE　and　the　cut-off　point　with　the　lowest　self-similarity　by　NSE,　which　leads　to　the　reasonable　power　allocation.　In　the　simulation　and　experiment　analysis,　it　further　verifies　that　the　introduction　of　standby　storage　element　can　effectively　reduce　the　charge　and　discharge　times　of　the　HESS,　decrease　the　fluctuation　index　of　SOC,　which　improve　the　service　life　and　identification　accuracy　of　HESS.　The　result　shows　a　certain　feasibility　and　actual　significance　in　the　practical　applications.