A Hybrid Remora Optimization Algorithm with Variable Neighborhood Search for the Flexible Job Shop Scheduling Problem

Reducing the production makespan of modern small-batch, multi-variety jobs can be abstracted into a flexible job shop scheduling problem (FJSP). The organization can greatly improve production efficiency by scheduling according to the optimal solution of FJSP. However, due to the high computational complexity of FJSP, it is often difficult to achieve higher performance in actual production. To optimize the makespan of the FJSP, a hybrid remora optimization algorithm with variable neighborhood search (HROA-VNS) is proposed in this work. A hybrid initialization method based on machine load balancing is utilized to improve the quality of the initial population. To solve discrete FJSP, a conversion method for individual positions and scheduling solutions is proposed, and a host switching mechanism (HSM) is designed to enhance the development and exploration capabilities of the algorithm. Furthermore, for enhancing the accuracy and effectiveness of the optimal solution, a variable neighborhood search (VNS) strategy which contains multiple structures is designed for local search. According to the experimental data, compared with the basic algorithm, the HROA-VNS reduced the makespan by an average of 33.02% and 14.40% on the two benchmark test sets respectively, and it shows better performance than related algorithms.