A Mixed-Factor Evolutionary Algorithm for Multi-objective Knapsack Problem
Yanlian Du,Zejing Feng,Yijun Shen
DOI: https://doi.org/10.1007/978-3-031-13870-6_5
2022-01-01
Abstract:Nondominated-sorting plays an important role in multi-objective evolutionary algorithm in recent decades. However, it fails to work well when the target multi-objective problem has a complex Pareto front, brusque nondominated-sorting virtually steers by the conflicting nature of objectives, which leads to irrationality. In this paper, a novel mixed-factor evolutionary algorithm is proposed. A normalization procedure, i.e. mixed-factor, is introduced in the objective space, which links all the objectives for all the solutions of the problem to ease the conflicting nature. In the process of nondominated-sorting, the mixed factors of individual substitute the raw objectives. In order to ensure that the population are thoroughly steered through the normalized objective space, hybrid ageing operator and static hypermutation with first constructive mutation are used to guide the searching agents converge towards the true Pareto front. The algorithm proposed is operated on multi-objective knapsack problem. The effectiveness of MFEA is compared with five state-of-the-art algorithms, i.e., NSGA-II, NSGA-III, MOEA/D, SPEA2 and GrEA, in terms of five performance metrics. Simulation results demonstrate that MFEA achieves better performance.