An evolutionary knowledge transfer framework for solving workgroup reconfiguration problem in aircraft final assembly

Final assembly is a crucial aspect of aircraft manufacturing. The introduction of reconfigurable workgroups to aircraft final assembly lines (AFALs) significantly increases assembly flexibility and efficiency. However, the number of reconfigurations and their corresponding assignments in the constraint-complex AFAL must be considered jointly during the assignment of reconfigurable workgroups. This presents a great difficulty when optimizing the reconfiguration strategy. In this study, an optimization framework based on evolutionary knowledge transfer (EKT) is proposed to efficiently search the metameric-encoded search space. EKT extracts and transfers valuable knowledge in different reconfiguration sub-problems according to the temporal relations between encodings. Meanwhile, the scope of the EKT is constrained based on the historical individual changes of different sub-problems and different regions of the objective space. The experimental results show that the proposed framework outperforms state-of-the-art algorithms on real-world production data. The behavior and mechanism of the EKT approach during the optimization process are also investigated in detail.