Abstract: To solve the problems such as low efficiency of design knowledge acquisition and lack of effective expression and organization for product conceptual design,and to improve the ability of contradiction resolution in product design process,a product conceptual design method driven by patent knowledge graph and TRIZ was proposed.According to the design knowledge requirements of product conceptual design,combined with text compression and data enhancement methods,an entity relationship joint extraction model based on improved UIE (TC-UIE-senMixup) was established to accurately achieve extraction of entities and relationships in the absence of labeled samples.Based on a Function-Behaviour-Structure-Patent (FBSP) mapping relationship,a patent knowledge graph of the product was constructed.Then,the contradiction resolution method of TRIZ theory was used to eliminate the contradiction in conceptual design of products.Finally,the proposed method was applied to the shearer conceptual design process,and the optimization design of shearer cutting drum structure was taken as an example to verify the feasibility and effectiveness of the method.

Key words: product conceptual design, patent knowledge mining, design knowledge graph, design automation

摘要： 针对产品在概念设计中的设计知识获取效率低、缺乏有效表达组织等问题,为提高产品设计过程中的矛盾消解能力,提出了一种专利知识图谱与TRIZ融合驱动的产品概念设计方法。首先,根据产品概念设计所需的设计知识需求,结合文本压缩与数据增强方法,建立了一种基于改进通用信息抽取(UIE)的实体关系联合抽取模型(TC-UIE-senMixup),实现缺乏带标注样本时实体和关系的准确抽取;进而基于功能-行为-结构-专利(FBSP)映射关系构建了产品的专利知识图谱,然后,利用TRIZ理论的矛盾解决方法,实现了产品概念设计中的矛盾消除。最后将该方法应用于采煤机概念设计过程,以采煤机截割滚筒结构的优化设计为例,验证了该方法的可行性和有效性,提升了概念设计方案生成效率。 

关键词: 产品概念设计, 专利知识挖掘, 设计知识图谱, 设计自动化