Benchmark scheduling policies for MO-DFJSP

Citation Author(s):
Yong Zhou
Submitted by:
Yong Zhou
Last updated:
Tue, 05/17/2022 - 22:17
Data Format:
Research Article Link:
0 ratings - Please login to submit your rating.


To investigate the generalization performance of the evolved scheduling policies(SPs), which are generated by the hyper-heuristic coevolution, the evolutionary SPs extracted from the aggerate Pareto front were applied to 64 testing scenarios to compare with the combinations of 320 existing man-made SPs which include 32 job sequencing rules and 10 machine assignment rules. This dataset provides the simulation performance of the evolved SPs and the 320 existing man-made SPs on the multi-objective dynamic flexible job shop scheduling problem. We applied a design of experiments (DOE) approach to design the testing scenarios, six experimental factors that each factor with two levels (including 64 combinations) are used to construct the test set.


Using software jasima,which can be found at, the combinations of 320 existing man-made scheduling policies are applied to 64 testing training scenarios (more details shown below), and 100 simulation replications are performed for each scenario.

Simulation Parameters
Weights of jobs 4:2:1(20%:60%:20%)
Simulation Length 2500
Warmup Length 500
Machine number 10
Operation number distribution Missing, Full
Process time distribution U[1,99], N(120,20)
Allowance factor 1, 3
Utilization 80%, 95%
Optional device number U[1,2], U[1,4]