Retrieval of solid models based on assembly similarity

ABSTRACT

The research objective is to (re)use geometric part information to mine databases for process design information. Specifically, this research investigates an approach to use assembly solid model similarity to mine databases for assembly work instructions from the perspective of an automotive original equipment manufacturer. Results from this research will allow generation of assembly work instructions based on input of a solid model. This research presents an approach to determine similarity between query assembly solid model and database assembly solid models. In this method, similarity scores for the overall assembly solid model are used in conjunction with the similarity score of individual components. These similarity scores are obtained by computing histogram-based similarity scores, surface area differences and tessellation area distribution differences. A multi-index sort of the computed values for each (query-database) pair of assembly solid models results in a list of similar assembly solid models based on a query assembly solid model.

In addition, human designers were asked to identify similar geometric models in a controlled study. The results from the human study and the similarity algorithm are compared. It is found that using assembly model similarity, in conjunction with component model similarity, yields better correlation to survey results, as compared to the correlation between assembly model similarity alone and survey. Testing also shows that the proposed method has a statistically significant better correlation to survey results than traditional histogram based similarity approach. The approach to determine similarity of assembly solid models will be used to mine databases of solid models and retrieve their related assembly work instructions.

GRAPHICAL ABSTRACT

KEYWORDS:

• Solid model similarity
• knowledge reuse
• assembly work instructions
• shape histogram
• tessellation density

ORCID

Rahul Renu http://orcid.org/0000-0002-7069-9545

Gregory Mocko http://orcid.org/0000-0001-8845-4581