Hybrid artificial bee colony algorithm with branch and bound for two–sided assembly line balancing

The two-sided assembly-line balancing (2SALB) is widely used in various production systems especially in high-volume large-size products. However, not many types of research are focused on the study of the 2SALB problem. Recently, the artificial bee colony (ABC) algorithm was used in the solution process where it was considered as a very useful, effective and well-known algorithm. Nevertheless, the ABC is also known to be a slow converging method in achieving an optimal solution. This research is intended to improve the ABC performance in solving the 2SALB problem with the objectives to hybrid ABC algorithm with branch and bound concept and to evaluate the performance of this algorithm in minimizing idle time and number of the workstation . The two-sided assembly line data was tested in modified spreadsheet model which is equipped with random priority rules in order to get multi-solution of ABC approach for balancing two-sided assembly line. The feasible number of workstations was determined with the minimum idle time of every mated station and maintains the minimum one in each cycle. This was done by calculating the partial fitness of the mated station. The branch and bound concept was applied by using mated workstations to overcome the slow convergence of the ABC algorithm and solve the problem optimality. The visual basic application software was used to generate different solutions with the various partial fitness of the proposed approach. The modified ABC algorithm was tested with the 2SALB problems involving P9, P12, P16, P24, P44 and P65 tasks. The results were tested by comparing hybrid ABC with pure ABC, where it was found that hybrid ABC had less number of workstation with minimum partial fitness value. In addition, the comparison was done with other research on ABC with full constraints and the proposed approach shows improvement by reduction of two workstations in sixteen task problem and reduction of one workstation at nine, twenty four and sixty five task problem sizes. As a conclusion, the proposed hybrid of ABC with Branch and Bound concept has increased the effectiveness of 2SALB solutions through the evaluation of many alternative mated stations results before going to assign to next mated stations to obtain minimum workstations with given cycle time solution