A Metaheuristic Method for the Task Assignment Problem in Continuous-Casting Production
Table 6
Computational experiments of solving large-scale instances by CPLEX, PSO, and PSOB algorithms.
Number Of Machines
Number Of Tasks
CPLEX
PSO
PSOB
Best Solution
CPU(s)
Best Solution
CPU(s)
Best Solution
CPU(s)
2
20
575
3600
598.0
105.0
584.0
104.0
22
643727
3600
867.7
116.3
820.3
469.0
25
607637
3600
1345.0
202.3
894.3
263.3
30
16612880
3600
2218.0
347.3
1990.0
337.3
Average
-
3600
1257.2
192.7
1072.2
293.4
3
20
382432
3600
465.0
3042.0
410.7
2719.0
22
629671
3600
708.7
2965.0
697.8
2778.9
25
10491201
3600
1128.0
3234.3
1101.3
3008.9
30
-
3600
2010.7
3517.0
1050.3
3379.1
-
3600
1078.1
3189.6
815.0
2971.5
4
20
390400
3600
463.0
3354.0
424.0
3155.2
22
635660
3600
689.7
3248.7
689.7
3145.4
25
10341439
3600
1175.0
3440.0
1095.0
3456.0
30
-
3600
3972.0
3549.2
2477.4
3600.0
Average
-
3600
1574.9
3398.0
1171.5
3339.2
Note. The en dash means we did not find any solution within the time limits. Two values of best solutions solved by CPLEX are the lower bound and upper bound of the corresponding instance.