An empirically efficient two-phase heuristic algorithm for two-dimensional bin packing problem

Abstract

This paper reconsidered the Two-Dimensional Bin Packing Problem from the perspective of practical application based on actual needs of small furniture companies in China. Although there have been many researches on this problem for many years, most of the existed researches put their foremost attention on developing an algorithm with low complexity or approximately best performance which means that it can nearly work out the best solution most of the time. This is the common situation when we do research on such problems, but unfortunately it will leave out abundant practical needs. For example, the computational resources of those small companies are seldom, which makes it impossible to use algorithms that almost gets the best solution but takes much time. The significant goal is to find out a solution satisfactory enough within a short time, so some complex algorithms which aim to get a nearly best result aren't suitable here. Meanwhile, the cutting times are also a rather critical matter for them. Cutting too many times will cause high wastage to machines and is therefore unaffordable for small companies, but this was often ignored in previous researches especially in classic meta-heuristic algorithms since they frequently lead to complicated packing patterns with massive cutting times. This paper reviewed previous related researches and summarized their shortages and some necessary characteristics of a practically efficient algorithm. Based on classic algorithms, it gained a two-stage heuristic algorithm which performed better than current used ones in those small companies, as well as meets their special needs stated in this paper which is a rather important issue. And this algorithm itself may be used to meet needs of those factories or be used as a starter of the further optimization search.

Research on Bin Packing Problem Based on Supply Chain Economic Efficiency

“…The experimental results show that the effect of using scorer scoring function and α-R rule at the same time is better, and the improved rule is more suitable for actual data. In addition to some heuristic algorithms mentioned above, Qu Yuan and Wang Xuelian [21] use tabu algorithm to solve large-scale packing problem, which uses natural number coding and introduces penalty function to represent space utilization constraints. The convergence process of the algorithm is stable and can ensure the superiority of the solution obtained in large-scale situations.…”