A Hybrid Evolutionary Algorithm for Stochastic Robot Disassembly Line Balancing Problem

Advanced robotic systems are increasingly vital in remanufacturing, bringing about revolutionary changes in end-of-life (EOL) product recovery, disassembly, and remanufacturing. However, a key characteristic of EOL products is the uncertainty in the condition of their components, resulting in stochastic disassembly times. Hence, this research addresses the stochastic robot disassembly line balancing Problem (SRDLBP), which arises from the inherent stochasticity in the operation times during practical disassembly line operations. To address this challenge, the study formulates the mathematical model of SRDLBP, with the number of workstations, balance index, and energy consumption as optimization objectives, while also considering the constraint of equivalent workstation operation time meeting the cycle time requirement. Furthermore, specific methods for generating and decoding initial solutions are proposed to handle the uncertainty in disassembly times. Additionally, an improved differential evolution operation is introduced within the framework of a non-dominated sorting genetic algorithm (NSGA-II), developing a hybrid evolutionary algorithm (HEA) designed to solve SRDLBP. Finally, the proposed HEA is applied to a printer case study involving 55 components. The results demonstrate that the HEA exhibits superior global optimization capabilities to NSGA-II under low and high stochasticity scenarios.