Research On Multi-Agent Based Logistic Scheduling System

With the rapid development of Internet and modern information technologies, the traditional logistic is transiting to the modern one. There are many substaintial technical problems need our in-depth research.Modern logistic scheduling system is a complicate information processing system interacting and cooperating with other information systems.Designing of the architecture of logistic scheduling system with appropriate structure and openability is the essence of success in constructing a modern logistic scheduling system. Artificial Intelligence and Multi-Agent technologies greatly improve the abilities of intelligence and automation in logistic scheduling application, so the complicacy of human scheduling process is decreased, the strategy of scheduling is optimized and the utilization of the logistic resources is improved.In this thesis, research work are mainly focusing on three topics, including the architecture of modern logistic scheduling system, application mode of multi-Agent technology in logistic scheduling system, and strategies and algorithms of automatic logistic scheduling. The following results gained:An architecture of multi-agent based logistic scheduling system is proposed. Through the hierarchy structure, the fundamental interface, intelligent agent, fundamental business logic and the application logic are separated. So the open ability and the flexibility in business logic processing are enhanced, and the extension cost in application logic is decreased.A agent model of logistic scheduling system is proposed. Based on the BDI model, the concept and physical model of the agent in logistic scheduling system is setup, according to the operating mode and relations between business entities in the system.A communication mechanism between agents in a logistic system is designed. Through the extending of KQML semantic, the grammar and semantic of inter-agent communication in the system are defined. And the competition and cooperation mechanism can be reached by contract net.The effect of heuristic scheduling algorithm and C-W algorithm on scheduling strategy is researched, and a hybrid scheduling strategy is proposed. The centralized scheduling strategy is used in mass similar task distribution. The distributed self competing strategy is used in scattered small task distribution. So the balance of scheduling cost and target optimization reached.