1 数据名称: DMU
2 来源：Purdue Electronics Manufacturing Research Group page
3 网址：http://pst.istc.cnr.it/~angelo/OUdata/
4 文章：
1) Demirkol, E., Mehta, S.V., Uzsoy, R. (1998) Benchmarks for Shop Scheduling Problems. European Journal of Operational Research 109(1): 137-141
2) Ovacik, I.M., Uzsoy, R., Decomposition Methods for Complex Factory Scheduling Problems, Kluwer Academic Publishing (1997)
3) Ovacik I, Uzsoy R. (1994) Exploiting Shop Floor Status Information to Schedule Complex Job Shops. Journal of Manufacturing Systems 13(2): 73-84
5 数据形式：
Each benchmark contains 160 files (one file for each instance) with names st_psp_N.sch (N is a number from 1 to160). These benchmarks were originally downloaded from the URL:http://cobweb.ecn.purdue.edu/~uzsoy/ResearchGroup，which is not available anymore.
6 数据形式
1) First line: first entry is the number of activities nA (excluded the source activity with id = 0), the second entry is the number of jobs nJ, and the last one is the number of machine nM.
2) Each of the following nA lines contains 6 entries for each activiti i:
i. id i of the activity in the job’s sequence
ii. id of the job containing activity i
iii. id of the predecessor activity i in the job’s sequence
iv. duration of activity i
v. deadline time of the job containing i
vi. id of resource used by activity i
3) Each of all the remaining lines defines a setup time constraints between two activities, and contains 3 entries:
i. id of the start activity s
ii. id of the end activity e
iii. setup time st from start s to end e
4) Ready times of activities are represented as setup time from the source activity (id = 0) to the other activities i