XYB-DS-202301 基于多模态知识图谱的产品构型上下文推理设计方法研究

发布时间：2023-05-14

摘要：AI 赋能智能制造转型升级，智能装备的产品构型设计过程也逐渐趋于复杂化与精细化。传统的 CAD 系统缺乏针对智能制造背景下产品构型知识的上下文检索与匹配。因此，亟需获取全新的构型知识库， 建立构型之间的上下文关系。本项目首先阐明基于图像、语音和文本的产品构型知识多模态获取、表达、转译机制。然后，探究产品构型的上下文关系，建立产品构型知识图谱。进而，通过多模态知识图谱构建上下文推理设计机制。最后开发知识推理原型系统，以智能制造控制机柜的构型设计予以应用验证。本项目的研究成果将提高产品设计效率，有助于进一步拓展计算机辅助设计和智能设计等领域的理论与应用创新。

AI enables the transformation and upgrading of intelligent manufacturing, and the product structural design process of intelligent equipment is gradually becoming more complicated and refined. Traditional CAD systems lack contextual retrieval and matching for product structural knowledge of intelligent manufacturing. Therefore, it is urgent to acquire a new structural knowledge base and establish the contextual relationship between structures. This project firstly illustrates the multimodal acquisition, expression and translation mechanism of product structural knowledge based on image, speech and text. Then, the contextual relationship of product structures is explored and the product structural knowledge graph is established. Furthermore, context reasoning mechanism is constructed by multimodal knowledge graph. Finally, the prototype system for knowledge reasoning will be developed, and the structural design of taking the structural design of intelligent manufacturing control cabinet as an example. The research results of this project will improve the efficiency of product design and help further expand the theoretical and application innovation in the fields of computer aided design and intelligent design.

关键词：知识图谱；智能设计；上下文推理

Knowledge graph; Intelligent design; Contextual reasoning