Analysis And Design Of Mobile Agent Based Open DSS In The Grid Platform

With the continuous development of the economic globalization and Web Technology, it is impossible or inconvinient for many large scale decision activities to proceed in the centralized mode. The objects of decision support are limited in neither one decision maker nor a decision group in the same organization, but in a variety of interdependent decision organizations, which makes decision problem solving distributed. The characters of decision support objects and problem solving have changed in a great extent. Therefore, it needs DSS to enhance its intelligence and the ability of parallel problem solving .This thesis puts forward a new model of mobile agent based open DSS in the grid environment to meet the growing needs of decision support. The model adopts the Service-Oriented Architecture, utilizes the abundant decision resources in the grid environment and combines the intelligence and flexibility of mobile agent to construct open DSS in the grid environment. The MABODSS uses the popular XML based Web Service technology and Artificial Intelligence to achieve the distribution,intelligence,integration and loose coupling of the DSS.Traditional DSS usually utilizes the settled mode of three local bases, while the design thought of the MABODSS is constructing DSS in the Service-Oriented Architecture so as to realize the sharing and cooperation of decision resources in the largest extent. The grid environment can be looked as a huge dynamic model base,knowledge base and database of the MABODSS. The intelligence and self-adaption ability can also be used distribute a complicated decision problem solving to the nodes in the grid environment so as to do the parallel asynchronism problem solving and construct the really open distribution DSS.In order to enhance the intelligence and productivity of the MABODSS, the thesis adopts the Case-based Reasoning to do the task disassembling and grid decision resources matching. It explains in details the work flow and reaction mechanism of the CBR based Agent reasoning mechanism in the MABODSS scheme layer. This method can enhance the self-learning ability and intelligence of agents dynamicly and gradually by evaluating and analyzing the existing cases thoroughly. A simple MABODSS design and realization based on the transportation optimization problems of third party logistic company.