Enhancing remanufacturing efficiency: a genetic teaching-learning-based optimisation algorithm for human-robot shared-workstation disassembly line balancing problem

Human-robot collaborative technology leverages the complementary capabilities of both agents, offering diversified operational scenarios for the remanufacturing industry. In this study, the human-robot shared-workstation disassembly line balancing problem (HRSW-DLBP) is addressed, where humans and robots operate concurrently. The HRSW-DLBP facilitates the rapid release of precedence constraints on components while processing both hazardous and complex components. Recognising the prevalence of sequence-dependent setup times (SDSTs) in practical applications, this study extends the HRSW-DLBP with SDST to more accurately model real-world scenarios. The HRSW-DLBP-SDST presents a more complex challenge than its predecessors, which do not consider such setup times. Therefore, devising an effective method for solving this problem is crucial. Given the NP-hard nature of the HRSW-DLBP-SDST, this study introduces a genetic teaching-learning-based optimisation (GTLBO) algorithm tailored for large-scale problem solving, incorporating a double-layer encoding and decoding strategy informed by the characteristics of the problem and enhancing local search operator to better align with the GTLBO structure. The performance of the proposed GTLBO algorithm was benchmarked against established optimisation algorithms across four cases, demonstrating its superiority. Finally, the HRSW-DLBP-SDST was applied to a liquid crystal display TV disassembly scenario, yielding multiple optimal allocation schemes. These case studies confirm the efficacy of the proposed method in resolving the HRSW-DLBP-SDST.