This　paper　presents　an　intelligent　structural　damage　identification　model,　where　three　kinds　of　intelligent　information　processing　techniques,　i.e.　fractal　theory,　probabilistic　neural　network　(PNN)　and　data　fusion,　are　integrated　to　implement　damage　identification　from　multi-sensor　data.　This　intelligent　model　proposed　consists　of　4　modules,　which　are　data　preprocessing,　box-counting　dimension　extraction,　PNN　decision,　and　fusion　decision　output　modules.　The　efficiency　of　the　intelligent　model　proposed　is　validated　by　detecting　both　single-　and　multi-damage　patterns　of　a　two-span　concrete-filled　steel　tubular　arch　bridge　in　service.　The　results　show　that　the　intelligent　model　proposed　can　not　only　extract　nonlinear　features　through　the　fractal　theory　from　the　vibration　response　data,　but　also　provide　more　reasonable　and　reliable　damage　assessment　results,　and　have　excellent　tolerance　and　robustness　capabilities.　Copyright　©　2010　Binary　Information　Press.