Research On Key Technologies Of Resource Management And Configuration In Manufacturing Grid

Globalization manufacturing and network manufacturing have become a research focus in modem manufacturing field. Manufacturing Grid is the key technology sustention of network manufacturing and an infrastructure of integration and sharing of manufacturing resources through network. Manufacturing resources are encapsulated as standard and open grid services adopting grid and information technologies and shared in the form of high quality manufacturing services in Manufacturing Grid. To achieve the resource sharing and optimal configuration under Manufacturing Grid environment, in this dissertation, some key technologies of Manufacturing Grid such as architecture, resource modeling, service node construction, resource service capability estimation and resource optimal configuration have been investigated based on related works. The main contributions of this thesis are as followed:(1) A multi-hierarchy architecture of Manufacturing Grid was put forward referring to the Open Grid Services Architecture (OGSA). From bottom to top the architecture is composed of a physical resource layer, a resource connective layer, a virtual resource layer, a service management layer and a manufacturing service application layer. Within the architecture, physical resource entities are mapped as grid services from the physical resource layer to the virtual resource layer. In this way local resources are sharedas whole network resources in Manufacturing Grid, which embodies adequately the kernel of services in Grid.(2) A whole relationship multi-view model was brought forward to depict the architecture characteristics of Manufacturing Grid. In the model, a functional view, a resource view, an information view and a process view were presented to describe respectively the functional fabric, resource composing, information source and system run mode of Manufacturing Grid. The front three views were connected as an organic whole by the process view, which constitutes the whole relationship multi-view model of Manufacturing Grid.(3) The typical procedure and basic operations of resource management for Manufacturing Grid were investigated, which includes the registration, presentation, discovery, estimation, allocation, status monitor of resources and so forth. A Manufacturing Grid resource management model was constructed, in which all resource management functions are implemented through inheriting the management interfaces in OGSA. A resource abstract model under Manufacturing Grid environment was put forward to present resource attribute information and to define access interfaces. Through the model, the service information description, operation interface definition, service implement and deployment are realized depending on Web technologies. As an instance, a sort of standard part lib systems was encapsulated as grid service nodes.(4) The resource allocation process driven by tasks in Manufacturing Grid was discussed. In the process, key approaches including task modeling, task decomposing, resource search, resource service capability esitimate, task status monitor were studied intensively. Among those, resource service capability estimation is the key and difficulty, so a muti-objective integrative evaluation method based on genetic algorithms was presented. A multi-objective optimization mathematical model was established; a fitness function, crossover and copy probability, a select, crossover and variation operator, a keep optimazation operation and an end condition of genetic algorithms were designed. Lastly the validity of the method was proved.(5) The economic problems in Manufacturing Grid were studied and the advantages of introducing economic models to resource allocation of Manufacturing Grid were expatiated. A resource market model for Manufacturing Grid according to market supply and demand relationship theory and a resource allocation framework based on market mechanism were established. A market mechanism based resource allocation principle and process were discussed, which achieves the Pareto optimization of resource allocation at the market supply and demand equilibrium state. A rapid iterative algorithm was introduced to confirm the resource equilibrium price which is the key of resource allocation process, and an instance was given to validate the algorithm.(6) A resoure share oriented service platform of Manufacturing Grid was developed under the research background of this dissertation based on Web technologies and Globus Toolkit, which is a grid open source software middleware toolkit. The platform includes a resource management sub system and a task management sub system. The numerical control machine tool as a machining resource among hardware and a standard part lib system as a design and software resource were encapsulated as grid servce nodes. Also a manufacturing task management process across the platform was simulated through machining a blade of aeroengine, which validates theories and methods presented in this dissertation.