A new genetic algorithm based on anti-Darwinism for multi-objective part-tool grouping problem

Abstract

Grouping parts and tools is an essential problem that arises in the set-up of a flexible manufacturing system (FMS). In the part-tool grouping problem (PGP), the process of assembling parts is assigned to several machines so as to optimize plural performance criteria. In this paper, the PGP is formulated as a multiobjective optimization problem. Then, for sampling various nondominated solutions from along the entire Pareto-optimal front of the PTP, a new genetic algorithm (GA) based on the evolutionary theory advocated by Kinji Imanishi is proposed. While conventional GAs mimic the process of natural selection, the proposed GA realizes the situation of habitat segregation, i.e., a principle of coexistence. The Imanishism-based GA can find various Pareto-optimal solutions effectively, because it keeps the diversity of population in both of the objective and the problem spaces without harming the power of local search operations. The advantage of the Imanishism-based GA is confirmed quantitatively through computational experiments conducted on a practical problem instance of the PGP.