Accurate　state　of　charge　(SOC)　and　state　of　health　(SOH)　estimation　is　very　important　to　ensure　safe　and　efficient　operation　of　electric　vehicle　battery　system.　In　this　study,　an　improved　co-estimation　method　of　SOC　and　SOH　based　on　a　fractional　model　is　proposed.　A　fractional　second-order　model　is　established.　The　identification　of　model　parameters　(including　the　order　of　fractional　elements)　is　realized　by　adaptive　genetic　algorithm　and　the　SOC　is　estimated　using　multi-innovations　unscented　Kalman　filter　(MIUKF).　At　the　same　time,　the　unscented　Kalman　filter　(UKF)　is　used　to　predict　SOH　to　update　the　actual　capacity　of　the　SOC　estimator.　The　effectiveness　of　the　proposed　co-estimation　method　is　validated　by　experiment　data　under　different　test　cycles　and　battery　aging　degrees.　The　results　show　that　the　root　mean　square　error　of　SOC　at　25　degrees　C　is　less　than　0.38%　under　different　test　cycles,　and　the　root　mean　square　error　of　SOH　is　less　than　0.002%.　Compared　with　UKF,　fractional-order　unscented　Kalman　filter　and　fractional-order　MIUKF,　the　SOC　estimation　error　of　the　proposed　method　is　the　lowest.　Under　different　aging　degree,　the　root　mean　square　error　of　SOC　and　SOH　at　25　degrees　C　is　lower　than　1.21%　and　0.007%,　respectively.　It　indicates　that　the　proposed　method　has　good　adaptability　and　high　accuracy.