Particle　swarm　optimization　algorithm　was　applied　to　reduce　power　loss　and　to　prevent　the　decline　of　the　power　supply　quality　caused　by　the　imbalance　of　reactive　power,　but　reactive　power　optimization　is　a　mixed　non-linear　programming　problem　with　lots　of　variables　and　uncertain　parameters,　PSO　algorithm　also　has　some　limitations　such　as　premature　convergence,　which　causes　the　bad　accuracy　of　convergence.　And　then　the　coevolution　of　Particle　Swarm　Optimization　(PSO)　with　nonlinear　inertia　weight　factor　(w)　and　Simulated　Annealing　algorithm　(SA)　is　established　to　improve　the　original　algorithm　which　is　named　as　SA-NLWPSO.　Compared　with　the　algorithms　such　as　PSO,　SA-PSO　and　SA-WPSO,　SA-NLWPSO　is　better　for　global　convergence　and　higher　accuracy　of　reactive　power　optimization　by　using　IEEE-10　bus　system　as　a　model　for　the　simulation.　Copyright　©　2012　by　The　Institute　of　Electrical　and　Electronics　Engineers,　Inc.