Research On Hierarchical Representation For Similarity Assessment And Reuse Of 3D CAD Models

With the upgrade of manufacturing industry to digital and intelligent manufacturing,3D models have been recognized as the center of CAD,CAM and CAE. Reuse of existing models could reduce product design cost, accelerate re-innovation and improve competitiveness of company. Currently, the increase of 3D models in repository brings pressing challenges to PDM. While text based approach and semantic feature based approach are not applicable to name duplication or variation of feature definition between heterogenous CAD platforms, content based similarity assessment method is becoming a key technology of intelligent design, product data management and design reuse.The content of 3D models are of differeces in variational applications. Given that there are aboundant geometry and topology features in neutral STEP format models, the share and reuse of which are of pressing challenges between multiple platforms and applications. How to retrieve, to evaluate similarity and to reuse models efficiently, including the complex solid model, free-form/compound surface, as well as assembly with complex kinematic pair and function, are hot topics deserving promotion research in the field of intelligent digital design in recent years.To support design reuse, to improve efficiency of matching and precision of similar-ity evaluation, this work studied the fundamentals of 3D model representation in multiple resolutions, and proposed reuse oriented hierarchical structures, similarity assessment al-gorithms and strategies for application. For solid model, a geometric reasoning approach is proposed to abstract face adjacent topology, to build hierarchy with feature relationship. For free-form surface with continuous curvature variation, the representation is built by hi-erarchical partition, where curvature distribution is described by quantitative and qualitative properties from coarse to fine. For compound surface with notable curvature change, hier-archy is built based on morphological relationship. For kinematic pairs, lower pair is solved by merging degree of freedom restricted by faces of assembly feature, higher pair is solved by configuration space method, to abstract function of assembly in hierarchical tree struc- ture, which derived a reuse strategy that functional equivalent is precondition of geometric similarity. Based on the hierarchy proposed, geometry and topology are combined, to sup-port hierarchical matching from coarse to fine, and to integrate feature property as weight to calculate total similarity. The results of model retrieval validated that multiple resolutional hierarchical approach improved precision of similarity assessment. While the matching re-sults correspond to global or partial feature, the proposed approach is applicable to global/ partial reuse.Based on the proposed similarity assessment method, Python language, PythonOCC geometry kernel and PythonXY scientific calculation package are used to develop visualiza-tion tool and prototype system for model retrieval. C++language, NXopen API and Sikuli API are used to realize similarity based design reuse prototype and image matching based automation testing. The application of the system in complex injection mold design of au-tomotive interiors, improved reuse efficiency of design recources yet reduced time of mold design, which validated the feasiblity of proposed methods.