Constructing a knowledge graph-driven intelligent data-enabled design system for mold using deep semantic...

To address the inefficiency and high error rates in traditional methods for handling complex design processes in modern mold design, this study proposes a Knowledge Graph-driven Intelligent Data-enabled Design System for molds. Initially, deep semantic understanding techniques are employed to intelligently parse a large volume of mold design documents and data. Using Bidirectional Encoder Representations from Transformers (BERT) and Random Forest (RF) algorithms, key information and knowledge points are accurately extracted from the design documents, laying a solid foundation for constructing the knowledge graph. The study collects a significant number of representative mold design documents, followed by detailed data preprocessing and cleaning. Subsequently, the BERT model is utilized for semantic analysis to precisely extract various entities (such as components, materials, and process parameters) and their complex relationships during the design process. Research findings show that the system significantly reduces the error rate in mold design processes, decreasing from 0.15 to 0.0975. Regarding design efficiency, the average completion time per design task reduces from 20 h to 12 h. Compared to traditional design methods, the system shortens the average design cycle from 30 days to 22.5 days, achieving a reduction of 0.25. Validation through examples further demonstrates that the system exhibits notable advantages in intelligence and automation during mold design processes, effectively enhancing design quality and efficiency. Additionally, it reduces related labor costs by 0.2. In summary, the proposed Knowledge Graph-based mold design system not only demonstrates significant innovation and application prospects theoretically but also shows substantial effectiveness and value in practical applications. Future research directions include further optimizing system performance, expanding application domains, and exploring integration with other intelligent manufacturing technologies to elevate the overall level of smart manufacturing.