A　novel　multi-function　rescue　attachment　with　tonging,　shearing　and　grasping　capabilities　is　proposed　to　improve　the　rescue　operation　efficiency　and　save　time　when　switching　between　different　attachments　during　rescue　operations.　The　tonging　and　shearing　form,　as　well　as　the　grasping　form,　of　the　rescue　attachment,　is　analyzed.　Furthermore,　the　tonging　diameter,　shearing　diameter　and　grasping　force　are　simultaneously　chosen　as　objective　functions,　and　respective　mathematical　models　are　established　for　estimating　the　attachment＇s　performance.　The　nondominated　sorting　genetic　algorithm　(NSGA-II)　is　used　for　multi-objective　optimization　of　the　attachment.　To　improve　the　population　diversity　and　increase　the　search　capability　and　uniformity　of　the　distribution　of　Pareto　fronts　of　NSGA-II,　the　elitist　strategy,　crossover　operator　and　mutation　operator　are　improved.　The　Pareto　front　is　visualized　in　a　specific　polygon,　and　the　best　solution　is　then　selected.　More　importantly,　the　simulation　results　indicate　that　the　astringency,　population　diversity　and　search　ability　of　the　proposed　algorithm　can　be　improved.　The　uniformity　of　the　Pareto　front　distribution　can　also　be　improved,　which　may　lead　to　an　increased　number　of　available　design　schemes.　The　values　of　the　three　objectives　of　the　best　solution　obtained　via　the　improved　algorithm　are　superior　to　those　for　other　algorithms.