ABSTRACT Aiming at addressing several conflicting criteria of quality of service (QoS) that should be trade-off optimized during service composition and optimal selection (SCOS) in cloud manufacturing (CMfg), the improved non-dominated sorting genetic algorithm III (NSGA-III) is proposed and employed to address the SCOS issue. This is the first time that a preference-based multi-objective algorithm has been used to address the SCOS problem. In this paper, a novel K-layer preference reference point set approach is proposed for generating a reference point set with this algorithm to guide the search towards the interesting parts of the Pareto optimal region based on customer preferences, which improves the efficiency of the algorithm and the convergence of the obtained solution. A new fitness assignment strategy and environment selection scheme is developed accordingly to balance the relationship of diversity and convergence of preserved individuals in each generation. Additionally, the memetic algorithm is integrated into the evolutionary mechanism of the algorithm to address the insufficiency of local search. To validate the performance of the proposed algorithm, several test cases are conducted. The results demonstrate that the proposed algorithm is more competitive than other considered algorithms.

中文翻译：

---

一种基于偏好的云制造服务组合优化多目标算法

摘要 针对在云制造 (CMfg) 中的服务组合和最优选择 (SCOS) 过程中应权衡优化的服务质量 (QoS) 几个冲突标准，改进的非支配排序遗传算法 III (NSGA-III ) 被提议并用于解决 SCOS 问题。这是首次使用基于偏好的多目标算法来解决 SCOS 问题。本文提出了一种新的K层偏好参考点集方法，利用该算法生成参考点集，以引导搜索基于客户偏好的帕累托最优区域的有趣部分，提高了算法的效率。以及所得解的收敛性。相应地开发了一种新的适应度分配策略和环境选择方案，以平衡每一代保存个体的多样性和收敛性的关系。此外，模因算法被集成到算法的进化机制中，以解决局部搜索的不足。为了验证所提出算法的性能，进行了几个测试用例。结果表明，所提出的算法比其他考虑的算法更具竞争力。进行了几个测试用例。结果表明，所提出的算法比其他考虑的算法更具竞争力。进行了几个测试用例。结果表明，所提出的算法比其他考虑的算法更具竞争力。