reinforcement learning for DFJSP

Citation Author(s):
jiepin
ding
East China Normal University
Submitted by:
Jiepin Ding
Last updated:
Wed, 05/04/2022 - 08:04
DOI:
10.21227/z199-mj36
License:
31 Views
Categories:
0
0 ratings - Please login to submit your rating.

Abstract 

We conducted the experiments on seven different size instances for training and testing.

We used the notation |J| X |M| to denote that a DFJSP instance consists of |J| jobs and |M| machines.

Specifically, 10X10, 20X20, 30X30 are three static instances and four dynamic instances of various scales considering continuous arrival of jobs are set as 20X10, 30X15, 50X20, 100X20.

Instructions: 

Number of machines           [10, 30]

Number of initial jobs          [10, 30]

Total number of jobs            [0, 100]

Number of operations per job       [10, 20]

Processing time            [30, 50]

Available machine rate for each operation       [0.3, 0.7]