关键词: 再制造作业车间, 预调度, 动态调度, 鲁棒优化, 扰动事件, 生物地理学优化算法

Abstract: Climate change has become a significant challenge for humanity. Therefore, developing a low-carbon economy is extremely urgent and critical. Remanufacturing is a sustainable manufacturing process that emphasizes the recycling of waste materials and promotes low-carbon economic development, which has received widespread attention from both enterprises and academia. However, the scalable and standardized development of the remanufacturing industry relies on advancements in remanufacturing job shop scheduling technology. Thus, research on the remanufacturing job shop scheduling problem holds significant theoretical and practical value in advancing the growth of the remanufacturing industry.
Currently, there exists a research gap in the current studies that solve the problem of multiple uncertainties and disturbances in remanufacturing shops. To address the aforementioned gap, this study proposes a new robust optimization-based remanufacturing job shop dynamic scheduling (RO-RJSDS) model, which divides the remanufacturing scheduling process into two phases: a pre-scheduling phase and a dynamic scheduling phase. In the pre-scheduling phase, a discrete scenarios set is utilized to characterize the multiple uncertainties in the remanufacturing job shop. The robust optimization approach is employed to construct a mathematical model with the objective of minimizing the makespans and the difference in makespans under various scenarios, in order to obtain stable solutions that perform effectively in all possible scenarios. Once the remanufacturing process begins, the disturbances will trigger the remanufacturing system to enter the dynamic scheduling phase. In the dynamic scheduling phase, the final scheduling solution is determined based on the pre-scheduled solution and disturbances, while ensuring scheduling efficiency and minimizing performance differences with the pre-scheduled solution. Therefore, a hybrid rescheduling strategy is designed to handle disturbances that may decrease the efficiency of the remanufacturing system. In this study, an effectiveness indicator is used to evaluate the efficiency of the dynamic scheduling solution, which is measured by the makespan. Meanwhile, a robustness indicator is utilized to assess the performance gap, which is measured by the difference in performance between the pre-scheduling solution and the dynamic scheduling solution.
The proposed RO-RJSDS model is a typical NP-hard problem that cannot be solved using exact methods, such as the Lagrangian relaxation method and branch-and-bound methods. The biogeography-based optimization algorithm is widely used to solve NP-hard problems due to its good performance. However, the basic biogeography-based optimization algorithm suffers from insufficient population diversity and premature convergence. Therefore, an extended biogeography-based optimization (EBBO) algorithm with a new two-dimensional unequal-length representation scheme is proposed to solve the RO-RJSDS model. The proposed algorithm employs a sinusoidal migration model and designs new migration and mutation operators to facilitate efficient population migration. In addition, a local search strategy is proposed to enhance the convergence and distribution of the EBBO algorithm.
In this study, each simulation experiment data is randomly generated within the corresponding range. Simulation experiments are conducted to verify the algorithm's superior performance and validate that the proposed hybrid rescheduling strategy can efficiently respond to disturbances. Firstly, the EBBO algorithm is compared with three other baseline algorithms on eight different-scaled instances to comprehensively evaluate its performance. Secondly, simulation experiments are conducted to compare the performance of the proposed hybrid rescheduling strategy with that of a complete rescheduling strategy in the dynamic scheduling phase. The results indicate that the solution obtained by executing the hybrid rescheduling strategy dominates all the solutions obtained by executing the complete rescheduling strategy. Finally, simulation experiments are conducted to compare the proposed robust optimization-based remanufacturing job shop scheduling model with the traditional deterministic remanufacturing job shop scheduling model.
These findings offer valuable insights for future research. Firstly, future studies should consider various disturbances such as worker factors, quality failures, and urgent order insertions in the model. Secondly, researchers should adopt search strategies with lower computational complexity in the algorithm and explore the integration of machine learning methods to enhance the efficiency and performance of the algorithm.

Key words: remanufacturing job shop, pre-scheduling phase, dynamic scheduling phase, robust optimization, disturbances, biogeography-based optimization algorithm