A heuristic algorithm enhanced with probability-based incremental learning and local search for dynamic facility layout problems

摘要

The dynamic facility layout problem (DFLP) involves finding an arrangement of facilities that minimises the sum of material handling cost and rearrangement cost over multiple periods. In this paper, the DFLP is modelled as a multiple quadratic assignment problem (QAP), one for each period. Probability-based incremental learning algorithm with a pair-wise exchange local search (PBILA-PWX) is proposed for solving the QAP for each period. The proposed heuristic and 16 algorithms available in the literature are applied for solving a set of 48 benchmark instances of the DFLP. For most of the problem instances, the proposed heuristic provides better results in comparison with an existing robust algorithm. The deviations of the solutions for the proposed heuristic are found to be within 5% of the best known solutions. A case study conducted for determining the machine shop layout of firm manufacturing printing machines is also presented.