Faced　with　massive　data　and　the　complexity　of　query　requirements,　how　to　improve　the　query　speed　of　database　has　become　a　research　hotspot.　This　paper　analyses　the　artificial　intelligence　algorithms,　genetic　ant　colony　algorithm　(GA-ACA),　which　is　used　for　database　query　optimization.　The　GA-ACA　is　prone　to　decrease　the　diversity　in　multi-connection　search　of　database,　which　results　in　inefficiency　and　local　extremum.　To　solve　this　problem,　our　paper　proposes　an　improvement　algorithm　on　multi-connection　query.　Based　on　the　premise　of　population　diversity,　the　algorithm　analyses　the　population　entropy　and　variance.　And　it　chooses　the　equal　probability　crossover　or　the　unequal　probability　crossover　according　to　the　evolutionary　state,　which　effectively　avoids　the　phenomenon　of　local　optimum　due　to　the　iteration　of　similar　individuals.　This　paper　improves　the　crossover　operation　and　redefines　the　generation　mode　of　new　population.　Experiments　show　that　the　improved　algorithm　avoids　the　local　optimal　solution　to　some　extent,　meanwhile　shortens　the　searching　time.