Task　scheduling　is　the　core　research　content　of　cloud　computing.　It　studies　how　to　allocate　tasks　between　computing　nodes,　so　that　tasks　are　evenly　allocated　or　the　execution　cost　of　each　task　is　minimized　or　the　overall　performance　of　the　system　is　optimized.　Different　from　the　previous　task　slices　that　perform　independent　tasks　sequentially　in　a　model　with　processing　time　as　the　goal,　we　construct　a　mathematical　model　with　the　goal　of　optimizing　response　time,　in　which　task　slices　are　executed　in　parallel.　Then,　a　solution　method　based　on　the　improved　adjustment　entropy　function　is　given　and　a　new　task　scheduling　algorithm　is　designed.　Finally,　we　implement　our　proposed　task　scheduling　algorithm　and　compare　it　with　the　game　theory　algorithm　and　the　balanced　scheduling　algorithm.　From　the　experimental　results,　by　comparing　the　response　time　of　the　task　with　the　size　of　the　control　system,　the　system　load　and　the　communication　bandwidth,　it　is　found　that　the　task　scheduling　algorithm　we　proposed　is　superior　to　the　game　theory　algorithm　and　the　balanced　scheduling　algorithm　in　response　time,　and　the　algorithm　we　proposed　is　fair.