Research On Semantic Web Based Resource Integration And Scheduling For Manufacturing Grid

Manufacturing Grid is the combination of some most prominent outcomes from information technology and cooperative manufacturing. "Semantic Manufacturing Grid (SMG)" is proposed in this dissertation by introducing Semantic Web into MG. Its theory, architecture and realization are studied in detail, by focusing this research on manufacturing resource integration, discovery and scheduling o The main contributions of this dissertation can be described as follows.(1) MG is analyzed and compared with Networked Manufacturing and computing grid. An infrastructure of semantic grid service is proposed by applying Semantic Web to WSRE A framework of MG ontology is designed, which is a key infrastructure to enable semantics in MG. It is made up of three sub-ontologies including grid service ontology, manufacturing domain-specific ontology and SLA ontology. Then the five-layered architecture of SMG is proposed based on the two infrastructures. OWL DL, which is based on Description Logic (DL), is employed as the ontology language of SMG. Its formal semantics and reasoning mechanism are analyzed in detail.(2) Approaches of grid service ontology design by extending OWL-S, and manufacturing domain-specific ontology design based on feature extraction technology, are studied. Scalable models of these two ontologies are also established. Manufacturing resources are organized and managed as isolated, cooperative manufacturing cells. Each cell is multi-view modeled towards its capabilities of job types, and each view is encapsulated as a semantic grid service by employing MG ontology. A Hierarchically distributed framework of service registry is introduced into SMG, in order to organize and manage the semantic information of services. By this means, manufacturing resources are integrated in the manner of service information integration.(3) SMG is for Virtual Organization (VO), so it is crucial for SMG to discovery services efficiently. This issue is analyzed and formally described. The data structure and physical model of SMG service registry (named as UDDIe4SMG) are designed based on UDDI, in which a special indexing component is used to improve the efficiency of semantic service matchmaking. A text similarity based approach of matchmaking is proposed on UDDIe4SMG, which is divided into four parts of match, including manufacturing type, input, output and resource properties. Possible "degree of match" of each part and the total "degree of match" of two service descriptions are defined. Then, algorithms of matchmaking are implemented and given in detail.(4) The issue of multiple QoS indices based scheduling optimization of SMG service is analyzed. A Directed Acyclic Graph (DAG) based model of service composition is proposed, which transforms the issue of service scheduling optimization into the corresponding issue of service composition optimization. The SLA ontology is constructed based on semantic WS-Agreement specification, by extending the QoS indices of WS-Agreement towards SMG. A new Analytic Hierarchy Process (AHP) based on Triangular Fuzzy Number - Fuzzy Consistent Matrix is proposed, which is used to calculate the weight of each QoS index to help decision-making. Then the model of service composition optimization is established, and a malformed-solution proof algorithm, which is based on Simulated Annealing (SA) algorithm and named as CPSA, is proposed to solve this model.(5) The theories and approaches proposed in this dissertation are condensed into a prototype system, which is implemented in the background of MEMS manufacturing. Its architecture and working pattem embodies almost all the key characteristics of SMG. The theories and approaches proposed in this dissertation are verified by the prototype system with some instances from MEMS domain.