Optimization Methods For Service Clustering, Composition And Scheduling In Cloud Manufacturing

In recent years, many new network technologies(such as Service oriented Computing, Cloud Computing and Internet of Things) have emerged, providing new applications for agile and intelligent cooperation among various organizations based on virtual enterprises. Mirroring this emergence, Cloud manufacturing(CMfg) was proposed to open a new way for manufacturing industry in China achieving the shift from production-oriented to service-oriented. CMfg is a new service-oriented networked manufacturing model able to enhance the resources utilization within its network, to construct intelligent service development for enterprise information, and realize the dynamic sharing and intelligent distribution of manufacturing resources and manufacturing capabilities. CMfg service provides dynamic on-demand access to manufacturing resources or manufacturing capabilities for virtual manufacturing chains generation, as to support the entire service manufacturing life cycle cross entities. CMfg service is the result a of new resources organization and integration channel including service clusters for manufacturing service processing optimization. However, CMg is a new research field still in its early stage, and theoretical researches and successful applications remain fuzzy. Consequently this thesis takes advantages of the novelty of CMfg and tackles its specifications to introduce innovative solutions along the manufacturing service processing.The thesis is articulated around the manufacturing service processing model and a cluster-based approach. The CMfg framework includes the following key research points i.e. manufacturing resource clustering, CMfg service composition and cloud service scheduling. Hence this thesis can be appreciated as a state-of-the-art on CMfg, and a sustainable basis for the development of core features. The objective of this work is to enhance manufacturing service processing with innovative approaches and optimization algorithms alongside realistic constraints and variables. All the presented methods are discussed through their model to translate the strategy and the direction of the present research, and analyzed toward existing methods’ benefits and / or drawbacks. Then, the precision and performances of the present methods are discussed through the simulation and experiments sections. Detailed research works include:(1) An initial Mfg resource clustering solution based on a density-based clustering supporting CMfg service decomposition model. Since Mfg resources can be countless, a clustering strategy is a fundamental basis for the development of a sustainable service processing. The objective is to distinguish Mfg resources sharing similar functional parameters and group them into service clusters based on a density evaluation. For this purpose, an innovative density-based clustering strategy is introduced based on a modified DBSCAN(Density-Based Spatial Clustering of Applications with Noise) alongside a space definition model.(2) By extension of the initial service clusters structure, additional research work takes advantages of the Mfg resource selection experience to present a very innovative approach for service cluster restructuring. The solution proposed is based on probability density function to a see a couple of Mfg resources selected together to restructure service clusters according to their probability to see all their virtual resources selected. The solution developed is ABC(Artificial Bee Colony) optimized to reach the most efficient solution and to be applicable to any already existing cluster structure.(3) Cloud Mfg service composition represents the core features for manufacturing cloud generation(i.e. virtual Mfg chains generation). Although service composition is largely presents in the literature, several aspects ar e omitted when it comes to manufacturing spheres. Transportation is one of these fundamentals aspects to consider in the composition process, alongside the complexity of Mfg interoperability. Additionally we can also notice that Qo S-based solutions do not include Qo S unmatched scenario. Consequently the solution proposed in this thesis is built on the foundation of Qo S, transportation and Mfg interoperability evaluation alongside a strategy for Qo S unmatched substitution. This innovative framework relies on an improved ABC optimization for cloud Mfg service composition(i.e. ABC_Cs CCMfg).(4) Task scheduling is a complex process in collaborative virtual Mfg chains. Keeping in mind that manufacturers often rely on long term commitment to their own scheduling and organizational model, an orchestration framework must be addressed for cloud Mfg scheduling. It has to insure the integrity of service manufacturers while satisfying global time consumption optimization. For this purpose, this thesis tackles cloud Mfg service scheduling orchestration based on availability and timeslots analysis. To address an optimal solution exploration, the CMfg scheduling orchestration is ABC optimized.(5) At last, the thesis presents a use case based on the shift from ASEM’s manufacturing model for the HT700 to a cloud Mfg service model based on a cluster-based CMfg scenario. The clustering and composition framework process are integrated to generate a virtual manufacturing chain for the HT700. In such way we can perceive the benefits and the future of CMfg for small-medium enterprises.The research work in this thesis has been supported by the China Ministry of Science and Technology(MOST) 863 High-Tech R&D Project, ‘The Key Technology of Cloud Manufacturing Service Platforms(No. SQ2010AA0400507009)’ and the China Natural Science Foundation(NSFC) projects i.e. “Research on Service Domain-Oriented Artificial Bee Colony Algorithm paradigm and Optimization Theory”(No. 61472106); “Value-oriented, Software service, Methodology: Theory, Methods and Applications”(No. 61033005).