Intelligent data-driven system for mold manufacturing using reinforcement learning and knowledge graph personalized...

Traditional manufacturing models are heavily dependent on standardized processes, which makes it challenging to accommodate customized production needs. To address this limitation, this study presents an optimized mold digitalization system grounded in knowledge engineering. The proposed system integrates knowledge graphs with intelligent algorithms to support the development of a smart quality control framework tailored to personalized manufacturing. This study places particular emphasis on the knowledge engineering module within the digitalization system. It focuses on converting domain-specific expertise into computable models that can be applied in real-world manufacturing scenarios. Additionally, reinforcement learning and graph neural networks are used to efficiently extract and utilize manufacturing knowledge. The experimental results reveal two key findings: (1) Within the enhanced learning knowledge graph framework, the algorithm-optimized using a graph convolutional network-achieves consistently higher qualification rates across test samples. When actual qualification rates exceed 88.1%, the model's regression fit also surpasses 88.1%, indicating strong alignment between predicted and actual performance. (2) Compared with other algorithmic models, the proposed approach achieves a predictive accuracy of over 94.7%. Overall, the proposed system significantly improves the level of customization in mold manufacturing while enhancing production efficiency and maintaining quality standards. The outcomes offer a new direction for the digital transformation of the manufacturing sector. Moreover, the approach holds practical value for enabling intelligent, flexible production processes, helping manufacturers better meet the growing demand for personalized products.