Much　research　has　been　made　to　study　the　evaluation　function　of　connect6　based　on　path.　However,　most　of　the　study　just　use　a　fix　set　of　parameters　which　is　inflexible　and　can＇t　meet　various　game　situations.　In　order　to　handle　this　problem,　we　propose　a　double　parameter　evaluation　function.　Different　from　the　former　study,　we　combine　the　global　and　local　scanning　method　and　use　two　sets　of　parameters　to　meet　different　situations.　Experimental　results　demonstrate　that　our　method　raise　the　win　rate　of　the　game　application　with　comparing　efficiency.