Resolution of one-dimensional bin packing problems using augmented neural networks and minimum bin slack

Abstract

The objective of this work is to compare the augmented neural network (AugNN) metaheuristic to minimum bin slack (MBS) heuristic to solve combinatorial optimisation problems, specifically, in this case, the one-dimensional bin packing problem (1D-BPP), a class of cutting and packing problems (CPP). CPP are easily found among various industry sectors and its proper treatment can improve the use of stocks in cutting problems or optimise physical space in packing problems. In order to optimise AugNN parameters, a design of experiment (DOE) was applied in order to guide a statistical analysis of different configurations of AugNN. The tests, developed in many benchmark problems found in the literature, showed that MBS heuristic was, in general superior, both in terms of the solution quality, which is about 70% better, and computational time, which is about 90% less.