Multi-objective-based differential evolution for balancing production cost, diversity and aggregated performance attributes in product family design
Ismail M. Ali,Hasan H. Turan,Ripon K. Chakrabortty,Sondoss Elsawah
DOI: https://doi.org/10.1007/s10696-022-09480-9
IF: 2.2092
2022-12-27
Flexible Services and Manufacturing Journal
Abstract:In product family design (PFD), deciding on a platform design strategy can be viewed as a multidisciplinary optimization problem that involves several factors, such as design variables, manufacturing costs, customizability, supplier reliability, and customer satisfaction. In this study, a multi-objective based differential evolution (MO-based DE) algorithm has been proposed for tackling the module-based PFD problem. The MO-based DE aims to find the best balance between many objectives, such as total production cost, diversity index, and a combination of other objectives (performance attributes). These objectives include commonality, modularity, and suppliers' reliability and all are aggregated to provide a goodness score. To effectively improve the DE's efficiency while solving such a complex optimization problem, the proposed DE integrates new elements such as (i) a novel solution representation, (ii) an improved heuristic technique for platform development, (iii) a weighted aggregation to combine different objectives, and (iv) a proposed platform-based crossover. To validate its performance, the proposed MO-based DE has been compared with (1) the standard DE to assess the effect of the incorporated new elements on DE's performance, and (2) well-known fast non-dominant sorting genetic algorithms NSGA-II and (3) NSGA-III for solving a real case study of a family of kettles. The experimental results confirmed the efficacy of the proposed MO-based DE as follows: in terms of average cost value, MO-based DE outperformed standard DE and NSGA-II by 26.40% and 11.69%, respectively. While in terms of goodness score, it achieved 20.69% and 8.05% better scores compared to standard DE and NSGA-II, respectively. Moreover, the proposed MO-based DE attained a very competitive performance against NSGA-III as it reached a better average cost and goodness score of 1.74% and 0.82%, respectively.
engineering, manufacturing, industrial,operations research & management science