Journal of Advanced Mechanical Design, Systems, and Manufacturing (Sep 2024)
Ontology-based methodology for the intelligent detection of product manufacturing information semantic representation errors in the model-based definition
Abstract
In model-based definition (MBD), product manufacturing information (PMI) semantic representation errors (SREs) are known to hinder the model readability, understandability, and processibility. Existing model quality test tools cannot detect PMI SREs efficiently and reliably; considering ontology's powerful abilities for semantic representation and logical consistency check, an ontology-based systematic methodology is proposed for their automatic and intelligent detection. Through the analysis of typical characteristics of PMI SREs’ content and style, four aspects of PMI SREs were identified, namely, the associated objects, structural integrity, parameters, and symbols. A three-dimensional PMI ontology (3DPMIOnto) framework was established, which included a PMI extraction module, a PMI ontology instance generation module, a rule-based inference module, and a message feedback module. In the Web Ontology Language, the object and data property of these ontology classes, including dimension annotation, geometric tolerance annotation, datum annotation, roughness annotation, and associated object classes, have been thoughtfully designed, which allows an automatic generation of ontology instances. The Semantic Web Rule Language was used to present four types of PMI SRE detection rules as per their classification. The application program interface of the commercial computer-aided design (CAD) software was used to extract PMI from the MBD model and imported to 3DPMIOnto to execute an intelligent inference process, where PMI SREs were filtered and fed back to the CAD software for redesign. A prototype system was developed and two cases study were performed. The results demonstrated the efficiency, feasibility, and openness of the proposed methodology.
Keywords
