of MDOD and MDRP have not found the solutions at given error level,there- fore,it is considered as solving fault. Given in Fig.5 are the preformance curves for MDOD,MDCP MDRP and 15 12 MOHP MDCP MDGP 12 MORP 8 HOGP ·2 0 0.5 2 6 810 15 0 MORP Fraction of average time 0.0050.020.04 Fract ion of average time Fig.2 Algorithm preformance at e,=0.05 and Fig.3 Algorithm performance fraction of average time in 0.05 2te,=0.05 30 MDHP 25 20 10 10 15 Fraction of average time Fig.4 Algorithm performance at e,=0.001 with TEST-I MDHP with TEST-I testing at error level 0.001,and the general conclusions are about the same with above, Given in Fig,6 is the plot of number of problems solved versus the relative error of objective function.It will be seen from this diagram,the MDHP is an algorithm which hasn't been influenced by the computational accuracy,it is illustrated to be a better algorithm for the ability to find global optimum. The computative results as shown in Tablc I that be used combination of two algorithms,indicate that the package MOD on the whole have high robust- 345， ， ‘ 户 ， 夕 巴 态爪 启沙 ， ‘ 州口尸 一 尸尸 月 一，了夕 厂 相日 幻 一 ‘ 一 产 一 曰 、厂料即 一 口。夕的任﹄山。 。已已山肠的山。 提 扩 ‘ 吕 气 「 一 户 「， · ￡ 了二 又 公 · 夕 峪。﹄。 逻 一 痔荞释 ‘ ， 日 · 公 “ 一 一 。 ， ， 五 ， ， 五 ， 五 ， ， ， 川 。 ，