Cooperation-based bi-level rescheduling method for multi-objective distributed hybrid flow shop with unrelated parallel machines under multi-type disturbances

Distributed hybrid flow shop production styles are gaining popularity, and many of them feature unrelated parallel machines (UPMs). However, most existing studies fail to consider or solely focus on a single disturbance type. Moreover, they seldom consider processing stability metrics. To closely resemble real production scenarios, this paper first studies the distributed hybrid flow shop scheduling problem with UPMs under two common disturbances (random arrival of new jobs and machine breakdowns), considering processing efficiency and stability metrics. To solve this problem, a cooperation-based bi-level rescheduling method is designed based on the nature of the problem. Among them, a factory load balancing mechanism is proposed to improve factory assignment schemes rapidly. A greedy job insertion partial rescheduling (first level) and improved genetic algorithm (IGA) (second level) are designed to obtain high-quality inner-factory scheduling schemes. In IGA, a dynamic decoding and guided elite individual optimization strategy are presented to balance conflicting objectives. An adaptive tabu search is designed to improve local search efficiency. The experiments involve 120 randomly generated instances. The results illustrate that the proposed methods are very effective. Compared to other efficient full rescheduling algorithms, IGA performs best in spread, generational distance, and inverted generational distance.