Abstract
Duplicate designs consume a significant amount of company resources during product development. Search for similar parts for a given query part, which facilitates design reuse, is crucial to avoiding this problem. Previous studies have only compared parts on a complete scale, not on a partial scale. This paper proposes a novel scheme which incorporates the concept of LOD (Levels of Detail) into 3D part comparison in order to assess partial similarity. Different LOD variants are generated from negative feature decomposition of a solid model. A human comparison behavior model (HCBM), mainly consisting of a back-propagation artificial neural network (ANN), is established by training with the result of a similarity ranking experiment. It combines the dissimilarity value at each LOD based on a modified D2 distribution. Test examples show that the proposed scheme is effective in 3D part search with LODs.