A discrete teaching-learning-based optimisation algorithm for realistic flowshop rescheduling problems

In this study, we proposed a discrete teaching-learning-based optimisation (DTLBO) for solving the flowshop rescheduling problem. Five types of disruption events, namely machine breakdown, new job arrival, cancellation of jobs, job processing variation and job release variation, are considered simultaneously. The proposed algorithm aims to minimise two objectives, i.e., the maximal completion time and the instability performance. Four discretisation operators are developed for the teaching phase and learning phase to enable the TLBO algorithm to solve rescheduling problems. In addition, a modified iterated greedy (IG)-based local search is embedded to enhance the searching ability of the proposed algorithm. Furthermore, four types of DTLBO algorithms are developed to make detailed comparisons with different parameters. Experimental comparisons on 90 realistic flowshop rescheduling instances with other efficient algorithms indicate that the proposed algorithm is competitive in terms of its searching quality, robustness, and efficiency.