Permutation Flow Shop Dataset

Contains the datasets used in our study "Permutation Flow Shop Scheduling with Multiple Lines and Demand Plans Using Reinforcement Learning". Two datasets are provided. The main dataset (folder data) contains 1050 problem instances for the multi-line permutation flow shop problem. The generation follows the method of Taillard (1993) and generates random processing times in the interval [1,99]. To create the demand plan we draw randomly from a multinomial distribution with equal probability for each job type. The additional dataset (folder data_disturbed) contains 150 problem instances for the multi-line permutation flow shop problem with short term-disturbances. Dataset structure Each PFSP dataset is structured in 15 subfolders. Each folder contains problem instances for a combination of line layout and processing time variation. Notation: Tai_PFSP_AL_B / Tai_D_PFSP_AL_B A: Number of Lines (1-3) B: Number of processing time variation (1-5) Each folder contains 70 problem instances. A problem file is a combination of one problem characteristic (number of jobs, machines and stations) and a demand plan variation. The processing times are fixed for one problem characteristic. Notation: tCD_E_F_G_H.mix C: Number of Lines (1-3) D: Number of problem characteristic (1-7) E: Number of jobs (20,100,500) F: Number of machines (5,10,20) G: Number of sorts (5,10,20) H: Number of demand plan variation (1-10) Each file represents a different problem in text format. Line Notation: L1: Demand plan L2: Layout Type L3: Number of machines L4: Number of machines per line L5: Number of total machines with synchronization machine L6: Number of sorts L7-end: Processing times matrix for the combination of machine (row) and job type (column) 1. Taillard, E. (1993). Benchmarks for basic scheduling problems. European Journal of Operational Research, 64 (2), 278-285.