Dragonfly　algorithm　is　a　novel　meta-heuristic　algorithm,　which　solves　the　problem　by　simulating　the　biological　behavior　of　dragonfly＇s　predation　and　migration.　There　are　also　some　common　problems　of　meta-heuristic　algorithms,　such　as　slow　convergence　speed　and　entrapment　in　local　optima.　To　improve　the　performance　of　dragonfly　algorithm,　this　study　combines　singer　chaos　and　sine-cosine　mechanism　to　reinforce　the　original　dragonfly　algorithm.　Singer　chaos　is　used　to　optimize　the　population　initialization　of　the　algorithm,　and　the　sine-cosine　mechanism　is　applied　to　modify　the　iterative　updating　formula　of　the　position.　The　performance　of　the　proposed　approach　is　examined　by　eight　numerical　benchmark　functions.　From　the　optimal　solution　and　standard　deviation　of　the　experimental　data,　it　can　be　seen　that　the　convergence　accuracy　and　stability　of　the　proposed　dragonfly　algorithm　are　significantly　strengthened.　Besides,　SC-DA　is　applied　to　optimize　the　design　of　cantilever　beam　problem　with　good　performance.　©　2019　IEEE.