Abstract
Physical modeling and simulation software is aimed at the support of model-based design of complex engineered systems. Device ontology, which has been employed as the product modeling scheme of the software and allows product modeling with component connection, restricts designers to perform system architecting, a fundamental task at the early stage of product development that determines the topology of system components based on reasoning about physical phenomena. In complementing the drawback of device ontology, this paper proposes another aspect model for model-based design with physical modeling and simulation. The aspect model enables designers to analyze the parameter network of a product model, which is derived from physical phenomena that jointly construct an equation system for numerical simulation representing the model behavior. The model is described with the Modelica language, a physical and system modeling language based on device ontology used in the software for model-based design. The analysis includes classification of simulation variables in the parameter network based on its network structure and the results of numerical simulation processed with a machine learning technique. Such an analysis potentially supports system-level reasoning about the behavior of a product model in system architecting. The proposed method is demonstrated using a simple DC model.
