Abstract: To solve the problem of service composition optimization in the cloud manufacturing environment,by taking the machining features as task granularity,a bi-layer optimization approach was proposed.The manufacturing resource supply graph and demand-feature graph were constructed to describe and store the information of suppliers and demanders respectively.Four indicators that were processing qualification rate,satisfaction evaluation values,on-time delivery rate and supply preference values were set sp as optimization indexes,and the first-level optimization was established through screening suppliers relied on the coefficient of variation algorithm for weighting the above indexes.Taking total process cost,total process time and total carbon emission as the optimization objectives,and the priority relationship among the features were added as constraints,an optimization model was developed.An improved genetic algorithm NSGA-Ⅲ that incorporated sub-candidate set variation operators was used to solve the model,and the bi-layer optimization was established.With process flow sequence analysis to example results,the feasibility of the proposed approach and the effectiveness of the algorithm were verified.

Key words: cloud manufacturing, service composition, small-batch customization, NSGA-Ⅲ, knowledge graph

摘要： 为解决云制造环境下小批量定制产品的服务组合优选问题,提出一种以加工特征作为任务粒度的资源服务组合双层优选方法。该方法首先构建了需求—特征图谱、制造资源供应图谱对供需两方信息进行描述与存储;然后设立加工合格率、满意度评价值、交付按时率、供应偏好值4个指标,并依托变异系数法进行赋权来初筛供应商,实现一层优选;最后建立了以工艺总成本、工艺总时间及总碳排放作为优化目标,特征间的优先级关系作为约束的优化模型,并使用融入次候选集变异算子的改进遗传算法NSGA-Ⅲ对其进行求解,完成二层优选。通过对实例结果的工艺流动顺序分析验证了该方法的可行性,并证明了改进后的算法求解效果更佳。

关键词: 云制造, 服务组合, 小批量定制, 第三代非支配排序遗传算法, 知识图谱