A Tolerance Type Screening Method Based on Similarity Evaluation

In the context of automatic generation methods for tolerance specifications, the current approach entails designers subjectively selecting geometric tolerances suitable for annotation from the tolerance types generated through automatic reasoning. This process is susceptible to the occurrence of mismatches between tolerance types and product functionality. To address this issue, this paper proposes a similarity calculation and evaluation method for screening tolerance types, which retrieves tolerance types that meet design requirements from resource cases. The proposed method commences with the establishment of a case data structure, predicated on the tolerance design problem. This structure incorporates features for similarity comparison, encompassing tolerance element characteristics that determine tolerance types, part component element characteristics, topological relationship features of component elements, and usage performance. Concurrently, past tolerance specification problems and their design solutions are defined as resource cases, while new tolerance specification problems are defined as target cases. Subsequently, analogous elements for similarity assessment are identified in both resource and target cases. Utilizing these elements, a similarity calculation method is devised, based on feature sets, to compute the similarity between the target case and previous cases. Finally, through case implementation and comparative analysis, the proposed similarity evaluation method is verified to have significant advantages in case discrimination ability.