Considering　the　comprehensive　inversion　of　different　physical　parameters　of　soil　layers,　the　hybrid　global　optimization　inversion　algorithm　of　Horizontal-to-Vertical　Spectral　Ratio　(HVSR)　was　improved.　Furthermore,　through　a　sensitivity　analysis　of　inversion　parameters　of　soil　layers　such　as　S　and　P　wave　velocities,　thickness,　Poisson＇s　ratio,　density　and　S　and　P　wave　quality　factors,　a　hybrid　global　optimization　inversion　algorithm　was　developed　for　simultaneous　inversion　for　thicknesses　and　shear　wave　velocities　of　soil　layers.　Taking　two　vertical　borehole　strong　motion　arrays,　the　Garner　Valley　Downhole　Array　(GVDA)　in　U.　S.　and　the　KiK-net　Vertical　Array　IWTH27　in　Japan　as　examples,　the　velocity　structures　of　the　arrays　were　inverted　from　the　theoretical　and　observed　HVSRs,　respectively,　and　compared　with　the　model　of　soil　layers　revealed　by　drilling　holes　in　the　vertical　borehole　strong　motion　array　sites,　thus　the　rationality　and　applicability　of　the　developed　inversion　algorithm　were　verified.　The　results　in　this　study　show　that　the　HVSR　global　optimization　inversion　algorithm　based　on　seismic　acceleration　records　is　an　effective　way　to　obtain　velocity　structure　of　soil　beds　in　the　site.