Two　drawbacks　of　Chaotic　Particle　Swarm　Optimization　(CPSO)　are　analyzed　in　the　paper,　the　first　one　is　the　search　in　the　initial　state　is　helpless　and　the　second　is　it　is　ineffective　in　jumping　out　of　the　local　optimum　at　the　later　stage.　An　improved　strategy　of　CPSO　(HMCPSO)　is　introduced　to　modify　these　problems.　This　method　uses　the　randomness,　ergodicity　and　regularity　of　chaos　to　associates　with　PSO　to　improve　the　abilities　of　jumping　out　of　local　optimum　and　searching　for　the　global　optimum.　Besides,　the　mechanism　of　judging　the　historical　data　of　particles＇　location　is　proposed,　by　using　which　the　excess　operations　of　chaotic　processes　and　the　run　time　are　reduced　significantly.　At　the　end,　HMCPSO　is　applied　in　the　optimization　design　of　pipe　network,　the　result　shows　the　algorithm　finds　a　lower　cost　point　and　the　reliability　is　satisfactory.　©　2013　TCCT,　CAA.