The　variability　and　uncertainty　of　wind　energy　is　becoming　a　huge　threat　to　the　security　and　stability　of　the　grid　with　high　wind　penetration.　This　paper　focused　on　the　stochastic　dynamic　model　for　aggregated　generation　of　large-scale　wind　farm　in　cyber　physical　energy　system,　and　presented　a　dynamic　system　description　aggregated　generation　of　wind　turbines　based　on　stochastic　Burgers　equation　from　near-surface　wind　field＇s　dynamics　and　wind-generation　function　with　uncertainties.　Then　we　developed　a　particle　filter　based　solutions　to　estimate　the　τ-step-ahead　generation　of　this　nonlinear　dynamic　system.　Finally,　a　30-turbine　study　case　shows　that　this　method　is　efficient　and　effective.　©　2016　Chin.　Soc.　for　Elec.　Eng.