Aiming　at　the　problem　that　a　single　equivalent　circuit　model　is　difficult　to　accurately　describe　the　full-time　lithium-ion　battery　and　the　low　accuracy　of　battery　state　of　charge　(SOC)　estimation,　a　joint　estimation　algorithm　of　lithium-ion　battery　state　of　charge　with　a　multi-mode　model　is　proposed.　Electrochemical　impedance　spectroscopy　is　used　to　analyze　the　impedance　distribution　of　lithium-ion　batteries　under　different　SOC,　and　an　equivalent　circuit　model　is　constructed　to　describe　the　lithium-ion　batteries　during　the　entire　charging　and　discharging　process.　A　multi-mode　model　based　on　the　variable-order　RC　model　is　obtained.　The　accuracy　and　practicability　of　the　Bayesian　order　criterion　are　adopted　to　determine　the　specific　order,　the　recursive　least　square　method　with　forgetting　factor　is　chosen　to　identify　the　model　parameters　online,　and　the　extended　Kalman　filter　algorithm　(EKF)　is　adopted　to　obtain　real-time　SOC　of　Li-ion　battery.　Under　constant　current　conditions　and　dynamic　stress　test　conditions,　a　multi-group　experimental　comparison　with　the　traditional　EKF　algorithm　based　on　the　first-order　RC　model　and　the　second-order　RC　model　is　carried　out.　The　results　show　that　the　joint　algorithm　with　the　multi-mode　model　works　well　under　different　conditions.　The　accuracy　of　the　estimated　SOC　is　improved　by　more　than　30%,　and　the　accurate　value　can　be　tracked　within　two　iteration　cycles.　©　2021,　Editorial　Office　of　Journal　of　Xi＇an　Jiaotong　University.　All　right　reserved.