In　vibration-based　structural　system　identification　(SI)　of　a　civil　structure,　the　commonly　used　SI　method,　such　as,　the　one　based　on　Newmark-β　algorithm　usually　needs　the　known　excitation.　In　fact,　it　is　difficult　to　measure　the　actual　excitation　and　full　output　responses　of　a　structure　system.　Here,　a　modified　Newmark-β　algorithm　was　proposed　through　introducing　polynomial　fitting　into　Newmark-β　algorithm　to　effectively　solve　the　drift　phenomenon　brought　by　the　error　of　SI.　Furthermore,　combining　the　modified　Newmark-β　algorithm　with　the　improved　multi-particle　swarm　co-evolution　optimization　(IMPSCO),　an　identification　approach　of　SI　and　excitation　adopting　responses　of　parts　of　a　structure　system　was　presented.　Finally,　numerical　simulation　for　a　seven-story　steel　frame　and　its　model　tests　were　conducted　to　verify　the　effectiveness　and　feasibility　of　the　proposed　method.　The　effects　of　noise　and　completeness　of　structural　response　were　also　discussed.　The　results　showed　that　the　proposed　method　can　be　used　to　effectively　identify　a　structural　system　and　its　excitation　only　adopting　responses　of　parts　of　the　structure　system;　it　has　stronger　noise　tolerance　and　robustness.　©　2017,　Editorial　Office　of　Journal　of　Vibration　and　Shock.　All　right　reserved.