Research Of Case Composition Method Based Representation And Retrieval Of Case Of Discrete-event Simulation Model

In area of model management of discrete-event simulation, there are problems ofduplication of modeling and difficulty in model reuse. The application of case-basedreasoning in model management of discrete-event simulation, provides a powerfulsolution for researchers to get rid of duplicate modeling and create high-quality discretesystem simulation models quickly.In case-base method, case coverage is often likely to be lower in a large problemspace than in a small problem space. This may cause the case-base to return a mediocrematch that will require considerable adaptation. Similar Problem occurred in modelmanagement area of discrete-event simulation.To solve the above problems, this paper study two key issues about representationand retrieval, in the process of combination of case-based reasoning theory, casecomposition method and model management of discrete-event simulation, for the basis forresearch of discrete simulation modeling approach using case-based reasoning theory andcase composition method.In the research of case representation for discrete simulation model, this paperanalyzes the structure of Arena discrete simulation model, proposed the concept “case ofdiscrete simulation”, and the representation method for which with three aspects,including available scenarios, composition constraints and description of simulation model,which additionally represented by NOG(Nested Object Graph) based on the Arena modelcharacteristics of discrete characteristics, module class hierarchy and the characteristics ofnested sub-model were proposed. On this basis, this paper build XML-based (ExtensibleMarkup Language) discrete simulation case storage method, and proposed classification ofcases for standard case of discrete-event simulation based on structure of hierarchyprocess.In the research of case retrieval for discrete-event simulation, overall retrieval methodfor case is proposed, based on non-precious three strategies which include availablescenarios, features of models and composition restraints for different modeling needs, andprecious retrieve for standardized and non-standardized case of discrete-event simulationbased on similarity retrieval algorithm, which calculated by genetic heuristic algorithm,under consideration of that similarity measure for graph is NPC problem. After the realization of similarity algorithm under Arena VBA programming environment, fivecategories of tests with Smarts for effective of simulation retrieval algorithm wereestablished. Then analysis, evaluation and improvement of the similarity algorithm weregiven. Tests result show that this method is fit for retrieval of case fragment. At last, thispaper discussed something issues about case composition of discrete-event simulation.