A Reinforced Tabu Search Approach for 2D Strip Packing

This paper deals with a particular “packing” problem, namely the two dimensional strip packing problem, where a finite set of objects have to be located in a strip of fixed width and infinite height. The variant studied here considers regular items (they are rectangular to be more precise) that must be packed without overlap, rotations being not allowed. The objective is to minimize the height of the resulting packing. For this problem, we present a local search algorithm based on the well-known tabu search metaheuristic. We reinforce two important components of our tabu search strategy to try to include problem knowledge. The fitness function incorporates a measure related to the empty spaces and diversification relies on a set of historically “frozen” objects. The resulting reinforced tabu search approach is evaluated on a set of well-known hard benchmark instances and compared with some stateof-the-art algorithms.