12.07.2015 Views

A Simulated Annealing Based Multi-objective Optimization Algorithm ...

A Simulated Annealing Based Multi-objective Optimization Algorithm ...

A Simulated Annealing Based Multi-objective Optimization Algorithm ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

100.70.60.50.40.30.20.100.81.41.210.80.60.40.201.50.60.40.20.10 01(a)0.20.30.40.50.60.710.5000.21(b)0.40.60.8121.41.21.5110.50.80.60.40.20201.51.510.50.50 02(a)11.522.510.500.40.202(b)0.60.811.21.41.41.210.80.60.40.201.51.41.210.80.60.40.201.510.5000.23(a)0.40.60.811.21.410.500.40.203(b)0.60.811.21.4Fig. 8.The final non-dominated front obtained by (a) AMOSA (b) MOSA for the test problems (1) DTLZ1 (2) DTLZ2 (3) DTLZ3TABLE IIINEW MEASURE displacement ON THE TEST FUNCTIONS FOR BINARY ENCODING<strong>Algorithm</strong> SCH2 Deb4 ZDT 1 ZDT 2 ZDT 6AMOSA 0.0230 0.0047 0.0057 0.0058 0.0029PAES 0.6660 0.0153 0.0082 0.0176 0.0048NSGA-II 0.0240 0.0050 0.0157 0.0096 0.0046TABLE IVTIME TAKEN BY DIFFERENT PROGRAMS (IN SEC) FOR BINARY ENCODING<strong>Algorithm</strong> SCH1 SCH2 Deb1 Deb4 ZDT 1 ZDT 2 ZDT 6AMOSA 15 14.5 20 20 58 56 12PAES 6 5 5 15 17 18 16NSGA-II 11 11 14 14 77 60 21length, distribution index for the crossover operation=10, distributionindex for the mutation operation=100.In MOSA [17] authors have used unconstrained archivesize. Note that the archive size of AMOSA and the populationsize of NSGA-II are both 100. For the purpose ofcomparison with MOSA that has an unlimited archive [17],the clustering procedure (adopted for AMOSA), is used toreduce the number of solutions of [32] to 100. Comparisonis performed in terms of Purity, Convergence and MinimalSpacing. Table V shows the Purity, Convergence, MinimalSpacing measurements for DTLZ1-DTLZ6 problems obtainedafter application of AMOSA, MOSA and NSGA-II. It can

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!