Resource Virtualization In Cloud Manufacturing And Their Applications

Cloud manufacturing, a new collaborative manufacturing paradigm, is a service-oriented and networked manufacturing mode. Production modes are driven to transform from product-centric to service-centric rapidly. Resource virtualization is crucial for the realization of cloud manufacturing, which enhances the robustness of cloud manufacturing systems. In this disserta-tion, the key problems for resource virtualization are investigated, which refer to heterogeneous manufacturing resources and capabilities of distributed manufacturing enterprises as scenarios. Complex features of manufacturing resources and capabilities are analyzed. A resource vir-tualization framework is designed to support the service contents and service modes in cloud manufacturing. Theories and methods are investigated to achieve global sharing and circulation of manufacturing resources and capabilities, with the following main contents.(1)A multi-granularity and multi-view framework is constructed for resource virtualization in cloud manufacturing. According to the service contents and service modes of cloud manufac-turing, the existing forms of manufacturing resources, the relationships among existing forms, the complex features of each existing form, and the complexities of resource usage are analyzed. The key techniques and the operating mechanism of the proposed framework are introduced. The unique nature of the proposed framework is demonstrated by comparing with the existing frameworks.(2)Normalized templates are designed and a resource semantic enrichment method is pro-posed for describing manufacturing resources and capabilities. Multi-granularity and multi-view are defined and a formalized model is established to express the complex features of manufac-turing resources and capabilities. An algorithm is proposed for total-ordered partition sequence for a resource granularity set, which performs the transformation among multi-granularity and the interaction among multi-view. A manufacturing resource conceptual model focusing on manufacturing capabilities is designed, which ensures that manufacturing resources are loosely coupled with their manufacturing capabilities. The key factors related to semantics of man-ufacturing resources and capabilities are analyzed from several aspects, such as the meaning and application modes of manufacturing resources, the restriction conditions and the context of manufacturing capabilities, cloud manufacturing knowledge. A manufacturing resource and capability semantic model is established and the important meta-data are organized based on the multilevel approach. A resource semantic enrichment method is proposed based on the domain ontology.(3)A multi-granularity virtualized resource generation algorithm and a multi-objective vir-tualized resource generation algorithm are proposed. Diverse requirements of cloud users and the difference between multi-granularity and multi-view gaps among complex manufacturing tasks are considered. The impact of three key factors (workflow, activity, resource) on the de-composition of complex manufacturing tasks is analyzed. Based on the similarity of manufactur- ing capabilities, the contribution degree of a manufacturing resource to a manufacturing activity, and resource collaboration, the attribute-oriented, activity-oriented, process-oriented resource generation algorithms are proposed, respectively. A multi-objective virtualized resource gener-ation model is formalized taking into account the compactness of a virtualized resource and the connectivity of virtualized resources. An evolutionary virtualized resource generation algorithm is proposed by integrating the efficient fuzzy C-Means method with the effective gravitational search algorithm. The algorithms could provide multiple choices for resource applications with different scales and features in cloud platforms.To validate the effectiveness of the proposed framework and methods, a prototype system is developed for resource virtualization of an aerospace manufacturing enterprise. Taking the manufacturing of the rotor blade of a helicopter tail for example, normalized description and semantic enrichment for manufacturing resources and capabilities are implemented. The pro-posed virtualized resource generation algorithms are validated by simulation experiments. The results shows the proposed methods provide comprehensive information for cloud platform and ensure semantic consistency. The global circulation and sharing are achieved by expanding the usage scope of manufacturing resources.