Branch, bound and remember algorithm for two-sided assembly line balancing problem

Abstract

This research presents a new branch, bound and remember (BBR) algorithm to minimize the number of mated-stations in two-sided assembly lines. The proposed methodology modifies the Hoffman heuristic to achieve high-quality upper bounds, and employs two new dominance rules, referred to as memory-based maximal load rule and memory-based extended Jackson rule, to prune the sub-problems. The BBR algorithm also employs several other improvements to enhance the performance, including renumbering the tasks and new lower bounds. Computational results demonstrate that BBR achieves the optimal solutions for all the tested instances within 1.0 s on average, including two optimal solutions for the first time. Comparative study shows that BBR outperforms the current best exact method (branch and bound algorithm) and the current best heuristic algorithm (iterated greedy search algorithm). As a consequence, the proposed BBR can be regarded as the state-of-the-art method for TALBP.