김정연, 노명일, 공민철, 장형운, "설계 P&ID와 3D 배관 데이터 간 매칭 데이터 자동 생성", 2026년도 대한조선학회 춘계학술발표회, 제주, p. 97, 2026.05.27-05.29
| Abstract | Legacy piping design data on ships is stored in multiple formats, including P&IDs and 3D CAD models. Since no explicit associations are defined between these design data sources, it is difficult to integrate them into a unified dataset. Nevertheless, such design data is essential for generating training data for ship design, as it embodies past design practices. This study proposes a methodology for automatically extracting associations between identical elements across P&ID and 3D CAD piping design data to generate training data without human intervention. For the P&ID data, the DXF format, which contains only geometric information, was used. For the 3D CAD data, a database that accumulates design information was used. Symbols in the P&IDs were classified into image-based detection symbols (valves) and rule-based detection symbols (nozzles, reducers, etc.), and short segments from valves to equipment nozzles were identified as the target search regions. For the 3D CAD piping data, piping information for the entire vessel was converted into a graph. The nodes corresponding to specific valves were retrieved by name, and the target pipe segments between them were identified. Through this approach, it was confirmed that matching data for valves and equipment nozzles can be automatically generated solely from design information. Furthermore, this study demonstrates the potential to extend the methodology toequipment- and equipment-nozzle-level matching datasets for use as training data. |
|---|---|
| Publication Date | 2026-05-28 |