The　state　of　charge　(SOC)　estimation　is　a　core　function　in　the　battery　management　system.　In　order　to　estimate　the　battery　SOC　more　effectively,　the　objective　of　this　paper　is　to　add　a　bias　compensation　to　the　original　least　squares　recursive　algorithm　with　forgetting　factor,　aiming　at　solving　the　problem　that　the　battery　parameter　identification　are　disturbed　by　measuring　noise.　In　this　paper,　an　improved　algorithm　is　used　to　identify　the　parameters　of　the　battery　model　online,　and　then　estimate　the　battery　SOC　combined　with　the　unscented　Kalman　filter　algorithm.　With　battery　charging　and　discharging　experiment　result,　the　parameter　identification　methods　with　and　without　bias　compensation　are　compared　and　analyzed.　The　results　show　that　the　improved　algorithm　has　better　accuracy　and　robustness.　©　2022　IEEE.