In　this　paper,　a　dynamic　recombined　multi-population　particle　swarm　optimization　algorithm　based　on　chaotic-mutation　(CDMSPSO)　is　proposed　to　realize　self-tuning　of　the　weighting　factors　when　model　predictive　control　algorithm　(MPC)　is　dealing　with　multi-objective　and　multi-constraint　conditions.　By　analyzing　the　design　principle　of　cost　function　in　the　model　predictive　torque　control　(MPTC),　taking　the　root　mean　square　of　the　current　error　in　the　two-phase　rotating　coordinate　system　as　a　reference,　the　objective　function　of　particles　in　particle　swarm　optimization　is　designed　with　reducing　the　torque　ripple　and　reducing　the　current　total　harmonic　distortion　(THD)as　the　main　control　objectives.　The　whole　population　was　divided　into　several　small　sub-particle　swarms　by　using　CDMSPSO,　and　the　particles　were　randomly　recombined　with　a　certain　recombination　period,　then　a　random　sub-particle　swarm　is　selected　and　chaotic　sequence　is　generated　iteratively　on　the　basis　of　any　particle,　and　the　selected　sub-particle　swarm　is　replaced　by　the　new　chaotic　sequence　to　realize　chaotic　mutation　of　particles.　Simulation　and　experimental　results　show　that　this　method　can　solve　the　problem　of　weighting　factors　setting　well　and　achieve　excellent　steady-state　performance.　©　2021,　Electrical　Technology　Press　Co.　Ltd.　All　right　reserved.