Accurate　prediction　of　renewable　energy　generation　acts　as　a　critical　role　which　not　only　provides　short-term　power　generation　in　the　future,　but　also　facilitates　scheduling　and　pre-configuration　of　energy　storage　sys-tems.　More　importantly,　the　power　generation　prediction　is　of　great　significance　to　the　demand　response　man-agement　(DRM)　of　renewable　energy　to　participate　in　the　electricity　spot　market.　Therefore,　DRM　helps　improve　the　stability　and　reliability　of　renewable　energy　systems.　This　paper　presents　a　novel　prediction-smoothing　based　methodology　to　reduce　and　eliminate　the　influence　caused　by　the　uncertainty　of　renewable　energy　output.　Firstly,　the　Whale　Optimization　Algorithm　(WOA)　is　combined　with　Long　Short-Term　Memory　(LSTM)　to　predict　short-term　wind　power　output.　Then,　the　Hampel-Butterworth-SG　filtering　strategy　with　outlier　regression　and　specific　risk　band　elimination　is　introduced.　After　that,　according　to　the　short-term　output　forecast　results,　the　scheduling　and　pre-configuration　scheme　of　peak-filling　type　energy　storage　is　developed.　Finally,　based　on　the　predicted　renewable　energy　output　and　electricity　price,　the　dynamic　changes　of　demand　response　(DR)　are　calculated　based　on　Logistics　function,　optimistic　response　and　pessimistic　response　factors.　Through　extensive　case　studies,　it　is　demonstrated　that　the　assessment　deviations　in　different　scenarios　is　less　than　5%,　which　are　0.64%　for　Scenario　1　and　3.64%　for　Scenario　2,　and　no　additional　penalty　is　required　at　this　time.　In　addition,　the　proposed　Demand　Response　Load　Adjustment　Rate　(DRLAR)　help　compare　the　differences　between　predicted　and　actual　DR,　which　are　DRLAR　=　0.03%　in　Scenario　1　and　DRLAR　=　0.01%　in　Scenario　2.　Users　are　able　to　adjust　the　DR　dynamically　according　to　the　electricity　price　to　realize　the　optimal　scheduling　of　their　renewable　re-sources.　The　proposed　methodology　creates　a　special　connection　between　DRM　and　renewable　energy　prediction,　which　provides　a　reliable　reference　for　future　work.