This　paper　proposes　a　data-driven,　multi-objective　optimization　control　method　based　on　policy　iteration　for　flotation　process,　addressing　the　limitations　of　existing　control　methods　that　rely　on　the　system　model　and　cannot　satisfy　multiple　performance　indices　simultaneously.　Firstly,　The　ordinary　linear　quadratic　regulator　algorithm　is　improved　with　process　data,　enabling　the　algorithm　to　obtain　optimal　feedback　gain　without　relying　on　the　internal　dynamics　model　of　the　system.　Then,　This　method　is　further　extended　to　the　situation　where　multiple　objectives　exist,　minimizing　the　deviation　of　the　final　additive　amount　while　meeting　control　requirements　for　both　concentrate　and　tailings　grade.　Finally,　the　convergence　and　effectiveness　of　the　proposed　method　are　verified　by　simulation　experiments.