抄録
To solve m-objective k-knapsack problems (mk-KPs) by using evolutionary algorithms, we propose a repair method that transforms infeasible solutions into feasible ones. In evolutionary multi-objective optimization, each solution in the population has a role in approximating a part of the Pareto front. However, since the conventional weighted scalar repair method (WSR) does not consider the position of each solution in the objective space, the solution diversity to approximate a wide range of the Pareto front is deteriorated. To improve the search performance of evolutionary algorithms for solving mk-KPs by enhancing the diversity of solutions, we propose a repair method considering the positions and repair directions of infeasible solutions in the objective space. Experimental results show that the proposed method improves the diversity of solutions and achieves higher search performance than the conventional WSR in mk-KPs.
