Errors　of　a　battery　model　will　dramatically　enlarge　as　the　internal　parameters　of　a　battery　varying.　To　reduce　the　systematic　errors,　a　parameter　adaptive　battery　model　is　proposed.　Based　on　it,　sliding　mode　algorithm　is　adopted　to　estimate　the　SOC　of　a　battery.　The　experimental　platform　is　constructed　and　the　UDDS　driving　cycles　is　used　to　verify　the　method.　The　results　show　the　error　of　SOC　estimation　is　less　than　2%　and　it　indicates　the　monitoring　algorithm　is　of　great　value　to　power　batteries　which　are　generally　used　in　variable　environment.　(C)　2016　The　Authors.　Published　by　Elsevier　Ltd.