A Genetic Algorithm with Weight-Based Encoding for One-Dimensional Bin Packing Problem

Abstract

This paper proposes a specialized genetic algorithm (GA) based on an expended relational representation named weight-based encoding for solving one-dimensional bin packing problem (BPP-1). The encoding provides a totally constraint-handling scheme to address general and specific constraints, while naturally eliminates redundancy and infeasibility of previous representations for BPP-1. The current study performs experiments for solving some problem instances from a benchmark data set by our specific coded genetic algorithm with one-point, two-point and grouping crossovers. Experimental results show that the proposed methodology works well for solving BPP-1 and performs well on experimented benchmark instances. In addition, the results also show that two-point and grouping crossovers work better than one-point crossover in our experiments.