Abstract: With the emergence of the service-based manufacturing model,the industrial Internet platform can configure the required production factor services to complete production tasks.In the industrial production process of industrial Internet environment,a task may require the coordination of multiple or multiple production factor services to complete.As the variety and number of manufacturing service resources increase,service composition becomes more difficult.At the same time,the effectiveness of service collaboration and service quality are the key factors for the success of the entire service composition process.In view of this,based on the historical data of service invocation in the industrial Internet platform,the effectiveness of candidate service collaboration was evaluated by combining Frequent Pattern growth(FP-Growth)and SimRank algorithm.Based on the collaboration effectiveness index and combined with the multi-objective evolutionary algorithm,a filtering method named Multi-Objective Evolutionary Algorithm with Collaboration Effectiveness Indicator Filter(MOEA+CEIF)was proposed to recommend a service combination scheme with high collaboration effectiveness and service quality optimization for users.Finally,the production and assembly process of ships was taken as an example to verify the effectiveness of the proposed method through a simulation data set.

Key words: industrial Internet, production factors, service composition, collaboration effectiveness, evolutionary algorithms, service recommendation

摘要： 随着服务型制造模式的出现,工业互联网平台可配置所需的生产要素服务来完成生产任务。在工业互联网环境下的工业生产过程中,一个任务可能需要多个或多种生产要素服务的协同来完成。随着制造服务资源的种类和数量增多,服务组合变得更加困难。同时,服务协同的有效性和服务质量是整个服务组合流程执行成功的关键因素。鉴于此,本文基于工业互联网平台中服务调用历史数据,结合FP-Growth和SimRank算法来评估候选服务协作的有效性。随后,基于该协作有效性指标并结合多目标进化算法,提出一种过滤方法MOEA+CEIF,为用户推荐具有高协作有效性且服务质量优化的服务组合方案。最后,以船舶的生产组装流程为例,通过仿真数据集验证同了所提方法的有效性。

关键词: 工业互联网, 生产要素, 服务组合, 协作有效性, 进化算法, 服务推荐