Abstract: To solve the problems of long product design cycle,limited objectivity of product function solution and lack of effective conceptual design aids in current product conceptual design,a product innovation conceptual design method based on patent knowledge graph was proposed.A patent knowledge construction technology oriented to conceptual design was developed,where design knowledge was defined as Subject-Action-Object (SAO) structure.Entities and their relationships were extracted automatically by Bidirectional Long Short Term Memory (BiLSTM),and then the graph was completed by using relational reasoning model based on graph theory and logical regression.Using the concept generation technology combined with product function expression model and network analysis method,the product concept generation and innovation evaluation were realized.The method was applied to generate the Automated Guided Vehicle (AGV) design scheme automatically,which proved the effectiveness of the proposed method.

Key words: design automation, conceptual design, patent mining, knowledge graph, innovation evaluation

摘要： 为解决当前产品概念设计中,产品设计周期长、产品功能求解客观性受限、缺乏有效的概念设计辅助工具等问题,提出一种基于专利知识图谱的产品创新概念设计方法。首先,利用面向概念设计的专利知识图谱构建技术,将专利中蕴含的设计知识定义为主体—动作—对象(SAO)结构,利用双向长短期记忆网络(BiLSTM)自动提取知识实体及其关系,进而构建知识图谱,并采用基于图论和逻辑回归的关系推理模型完成了图谱补全;然后,利用产品概念设计方案生成技术,结合产品功能表达模型与网络分析法,实现了产品概念设计方案生成,并能够进行创新性评价。最后,将该方法应用于自动导引车(AGV)设计方案的自动生成,证明了所提方法的有效性。

关键词: 设计自动化, 概念设计, 专利挖掘, 知识图谱, 创新性评价