Cloud manufacturing (CM) establishes a collaborative manufacturing services chain among dispersed producers, which enables the efficient satisfaction of personalized manufacturing requirements. To further strengthen this effect, the manufacturing service composition and optimal selection (SCOS) in CM, as a NP-hard combinatorial problem, is a crucial issue. Quality of service (QoS) attributes of manufacturing services, as the basic criterion of functions and capabilities, are decisive criterions of SCOS. However, most traditional QoS attributes of CM ignore the dynamic equilibrium of manufacturing services and only rely on initial static characterizations such as reliability and availability. In a high uncertainty and dynamicity environment, a major concern is the equilibrium of manufacturing services for recovering their functions after dysfunctional damage. Therefore, this paper proposes a hybrid resilience-aware global optimization (HRGO) approach to address the SCOS problem in CM. This approach helps manufacturing demanders to acquire efficient, resilient, and satisfying manufacturing services. First, the problem description and resilience measurement method on resilience-aware SCOS is modeled. Then, a services filter strategy, based on the fuzzy similarity degree, is introduced to filter redundant and unqualified candidate services. Finally, a modified non-dominated sorting genetic algorithm (MNSGA-III) is proposed, based on diversity judgment and dualtrack parallelism, to address combination optimization step processing in SCOS. A series of experiments were conducted, the results show the proposed method is more preferable in optimal services searching and more efficient in scalability.

中文翻译：

---

一种HRGO方法，用于增强云制造中的弹性增强服务组合和最佳选择

云制造（CM）在分散的生产者之间建立了协作制造服务链，从而可以有效满足个性化制造需求。为了进一步增强这种效果，CM中的制造服务组合和最佳选择（SCOS）作为一个NP难组合问题，是一个至关重要的问题。制造服务的服务质量（QoS）属性作为功能和能力的基本标准，是SCOS的决定性标准。但是，CM的大多数传统QoS属性都忽略了制造服务的动态平衡，而仅依赖于诸如可靠性和可用性之类的初始静态特征。在高度不确定性和动态性的环境中，一个主要关注的问题是制造服务的失衡，以便在功能受损后恢复其功能。因此，本文提出了一种混合的弹性感知全局优化（HRGO）方法来解决CM中的SCOS问题。这种方法有助于制造需求者获得有效，有弹性且令人满意的制造服务。首先，建立了对具有弹性的SCOS的问题描述和弹性测量方法的模型。然后，提出了一种基于模糊相似度的服务过滤策略，对冗余和不合格的候选服务进行过滤。最后，基于分集判断和双轨并行性，提出了一种改进的非支配排序遗传算法（MNSGA-III），以解决SCOS中的组合优化步骤处理问题。进行了一系列实验，