Abstract
The increasing complexity and customization of mechanical products have put forward higher requirements on the quality and process iteration speed of machining parts. The effective organization and utilization of the vast knowledge found in past production processes has become a critical concern in refining improvements to the subsequent machining process. To overcome this challenge, this paper proposes a framework for constructing a machining domain knowledge graph based on as-fabricated information model combining the inherent characteristics and development needs of machining process. Firstly, the state of machining part between procedures is described with as-fabricated information model concerning geometric and non-geometric information, after which the corresponding subgraph plotting method is presented with the definition of nodes, relationships and properties. Secondly, the overall route in building machining domain knowledge graph is analyzed, in which the semantic extraction mechanism of unstructured texts is proposed. Moreover, a case study is presented, including the instantiation of as-fabricated information model and the construction of knowledge graph with further approach of quality prediction based on a set of manufacturing process documentations of the example machining parts. The case demonstrates that the machining domain knowledge graph construction framework proposed in this paper is capable of recording the multi-source heterogeneous data generated during the machining process and providing data support for subsequent analysis and optimization.
Original language | English |
---|---|
Article number | 102638 |
Number of pages | 16 |
Journal | Advanced Engineering Informatics |
Volume | 62 |
DOIs | |
Publication status | Published - Oct 2024 |
Keywords
- As-fabricated information model
- Domain knowledge graph
- Machining
ASJC Scopus subject areas
- Information Systems
- Artificial Intelligence