An efficient hybrid optimization method for Fuzzy Flexible Job-Shop Scheduling Problem: Steady-state performance and analysis

Due to the uncertainty pervasive in manufacturing and production systems, a crisp processing time is no longer appropriate. Thus, the flexible job-shop scheduling issue comes with fuzzy processing times. Fuzzy numbers are used to depict the uncertainty associated with processing times. This paper proposes a Hybrid Fuzzy Flexible job-shop Scheduling Approach (HFFSA) for addressing the Fuzzy Flexible Job-Shop Scheduling Problem (FFJSSP). An encoding-and-decoding scheme is employed for initializing the population. HFFSA incorporates four operators to enhance the quality of solutions: the Different Positions Shuffling (DPS) operator, the Randomly Selected Positions Shuffling (RSPS) operator, the Block Shuffling (BS) operator and the Inversion Mutation (IM) operator. The DPS operator shuffles the different positions between a randomly generated solution and the current one. On the other hand, the RSPS shuffles random positions to maintain population diversity. The BS and IM work on a subset of positions from the current solution to obtain near-optimal solutions. Furthermore, we integrate the pair-wise local search strategy with HFFSA to improve the best solution. HFFSA is compared to twenty-five algorithms using Lei and remanufacturing benchmarks. Several statistical measures are employed, like the CPU time, best, average and worst fuzzy makespan values, boxplots and the Wilcoxon signed-rank test. The statistical analyses affirm the superiority of the proposed HFFSA.

Automated summary

Due to the uncertainty in manufacturing and production systems, a flexible job-shop scheduling approach using fuzzy processing times is proposed. The approach incorporates four operators to improve solution quality and integrates a local search strategy. The proposed method is compared to twenty-five algorithms and shown to be superior based on statistical analyses.