Abstract
This paper addresses the harness dissimilarity problem — a generalization of the path dissimilarity problem — in which the goal is to select ℎ dissimilar yet low-cost wire harness routing layouts, providing alternatives for engineering decision-making. Our approach involves two stages: first, generating a diverse pool of candidate routings; second, constructing subsets of size ℎ that are well-distributed along an approximated Pareto front. We analyze three methods for generating the candidate pool: the iterative penalty method, the pathway reward method, and NSGA-II. To approximate a Pareto front for the bi-objective problem of maximizing diversity while minimizing the total cost of selected candidates, we propose a weighted-sum constructive heuristic combined with local search, which outperforms a GRASP metaheuristic from related work. The method is also evaluated against an exact solver for small instances.