Abstract
The objective of this work is to compare the Augmented Neural Network (AugNN) metaheuristic with Minimum Bin Slack (MBS) heuristic to solve Combinatorial Optimization Problems, specifically, in this case, the bin packing problem, a class of Cutting and Packing Problems (CPP). CPPs are easily found among various industry sectors and its proper treatment can impact directly in savings of raw material and/or physical space of enterprises. In order to optimize the parameters of AugNN, a Full Factorial Design of Experiment (DOE) was applied. The tests, developed in many benchmark problems found in the literature, showed that MBS heuristic was generally superior, both in terms of quality of solution (approximately 70% better) and computational time (about 90% shorter).
Keywords:
Neural network; Cutting and packing; Design of experiments; Bin Packing; Minimum Bin Slack