Studies On Agent Based Supply Chain Management

The object of supply chain management (SCM) is to integrate the business processes in the supply chain, for the purpose of enforcing an enterprise's over all competition ability. The business processes integration requires the integration of SCM softwares. Current operation method of software or component encounters many problems such as low level abstract, passive process call, bad cooperation and coordination ability, and the semantic mismatch in the inter-operation, which affects the compaction and coherence of software or component coordination, and can not effectively support the quick construction of SCM information fundament. The introduction of Agent technology can help to settle those problems. But current Agent technology implies the coordination and coherence maintain in the description of Agent mental states, making it difficult to transparently describe the integrated architecture, not sufficient to support the integration of management information system needed in the business processes integration.This thesis studies Agent-based SCM, with the emphases of resolving the collaboration of heterogeneous Agents in supply chain, decreasing the complexity of Agent design, constructing information infrastructures with openness, reconstructability and extensibility. We use the following methods to realize our object:1. An OIL-based ontology is build to support enterprise modeling and Agent communication terms definition. Ontology and knowledge base technology are used to build a knowledge base to describe a MAS, which is used to control the collaboration between Agents, making it easy for heterogenous Agents to co-operate.2. This paper proposes an Agent-based SCM integration framework, which has two components: description models and Agent engine. The former uses six models to describe a MAS. Because the model description is in a highly abstract knowledge level, easy to be modified, making the SCM information system open and extensible. Based on the description models, we design an Agent engine to support the cooperation problem-solving, so as to decrease the complex of Agent design, promote the business processes integration.3. A difficulty that the integration framework is confronted with is the semantic mismatch when two heterogeneous Agents using different ontology to communicate with each other. To solve this problem, we design an OIL-based approximate ontology translation arithmetic. The key to a translation arithmetic is to allow a complex class to be defined as a logical combination of simple classes. So the translation between two ontologies is simplified to the translation between simple classes. Then an arithmetic is designed to compare the degree of the similarity of arbitrary two simplified classes. At last, the rules defining how a source ontology class instance translated into a target oneare constructed. By now, the OIL-based ontologies can be approximately translated to each other.At the end of the paper, we use an Agent based SCM system to simulate the integration of business processes of the constituent corporation of Special Mechanical Bearing. The integration is used as a prototype to introduce SCM modeling environment, and the process and steps to use this environment to model SCM. The result of the model simulation indicates that our framework is feasible.