FUNCTIONAL SIMILARITY AT VARYING LEVELS OF ABSTRACTION

摘要

Function modeling is often used in the conceptual design phase as an approach to capture a form-independent purpose of a product. Current research efforts have focused on the formalization of functional models, development of function-based design repositories, and concept generation based on a quantitative functional similarity metric. In this paper, three levels of abstraction of function models are obtained by including supporting functions, excluding supporting functions, and applying abstraction rules to function models of 128 products in a design repository. The similarity of these products is computed using the Functional Basis controlled vocabulary and a matrix-based similarity metric. A matrix-based clustering algorithm is then applied to the similarity results to identify groups of similar products. A subset of these products is then studied to further compare the three levels of abstraction and to validate the results. Similarity between consumer products depends on the level of abstraction of the models, with higher levels of abstraction producing better results.