To　implement　structural　damage　detection　in　the　case　of　a　low　signal-to-noise　ratio　(SNR),　this　paper　first　presents　an　adaptive　stochastic　resonance　(SR)　optimal　algorithm　that　is　able　to　adaptively　determine　the　parameters　in　SR　system　and　to　obtain　the　optimal　output　responses,　and　then　a　two-stage　damage　detection　approach　by　integrating　the　nonlinear　SR　denoising　and　improved　multi-particle　swarms　cooperative　optimization　(IMPSCO)　is　proposed.　In　the　first　stage,　structural　dynamic　responses　and　background　noise　are　processed　through　the　nonlinear　SR　system　for　the　purpose　of　enhancing　the　response　signal.　Afterwards,　the　RobustICA　is　used　to　extract　the　feature　independent　components　with　time-domain　signal　spike　property　and　the　maximum　of　normalized　source　distribution　vector　(NSDV)　is　computed　to　initially　locate　the　damage.　In　the　second　stage,　the　dynamic　responses　are　divided　into　several　segments　at　the　identified　damage　instant,　which　are　used　to　identify　the　structural　parameters　by　the　IMPSCO　algorithm　so　that　the　damage　location　and　extent　can　be　estimated　accurately.　A　numerical　experiment　of　three-story　frame　is　validated　the　proposed　novel　method,　and　the　results　show　that　the　proposed　method　can　successfully　implement　the　instant,　location　and　extent　of　damage　in　the　case　of　a　low　SNR.　©　2015,　International　Society　for　Structural　Health　Monitoring　of　Intelligent　Infrastructure,　ISHMII.　All　rights　reserved.