A Cooperative Coevolutionary Genetic Programming Approach to Generate Robust Priority Rules for the Integrated Problem of Assembly Line Balancing and Part Feeding by AGVs

Assembly lines play a major role in large-scale and mass-scale productions. The assembly line balancing problem (ALBP) arises when designing or redesigning an assembly line, which aims to distribute assembling tasks optimally to workstations with regard to a certain objective. However, without proper design of the part feeding system, it is almost impossible to achieve the original objective. Automated guided vehicles (AGVs) have been widely employed by material handling systems to ensure a reliable and flexible production flow. In order to solve ALBP and AGV scheduling problems, priority rule-based methods (PRBM) have been used in static and dynamic environments to rank the assembling tasks for assigning to a workstation in the ALBP heuristic and to sequence delivery tasks for AGV dispatching. Since ALBP and AGV scheduling problems are dependent on each other, this paper studies these two problems in an integrated manner. First, a mathematical model is developed with the goal of maximizing the throughput of the assembly line. In order to incorporate practical industrial settings, the model considers preventive maintenance (PM) of assembly workstations, AGV breakdown and product volume variation scenarios. Afterward, with the aim of generating a robust priority rule combination for solving both problems simultaneously, cooperative coevolutionary genetic programming (CCGP) with collaborative archive is designed. Finally, through extensive computational experiments, the robustness of the CCGP-generated rules is revealed by testing the rules on ALBP benchmark problem instances such as those of Gunther and Killbridge with different line layouts, which also involve PM, AGV breakdown and product volume variation scenarios. The effectiveness of those rules is also compared with the existing priority rules from the literature.