• CN: 11-2187/TH
  • ISSN: 0577-6686

Journal of Mechanical Engineering ›› 2025, Vol. 61 ›› Issue (3): 142-153.doi: 10.3901/JME.2025.03.142

Previous Articles    

Knowledge Graph Construction and Intelligent Reasoning Methods for Human-computer Hybrid Interaction in Product Design

GUO Xin1,2, HUANG Zechuan1,2, WANG Jie1,2, ZHANG Kai1,2, ZHAO Wu1,2   

  1. 1. School of Mechanical Engineering, Sichuan University, Chengdu 610065;
    2. The Innovation Method and Creative Design Key Laboratory of Sichuan Province, Chengdu 610065
  • Received:2024-02-18 Revised:2024-08-10 Published:2025-03-12

Abstract: Product design is a creative activity oriented to the iterative and progressive evolution of user requirements, which centers on the application of multi-domain knowledge. To address the problems of low knowledge visualization in product design, weak knowledge reasoning under human-computer interaction and to improve the degree of product design intelligence, a product interactive design knowledge application model based on knowledge graph (KG) is proposed. Firstly, based on the role and interaction evolution mechanism between product design activities, the correlations between product design problems and solution knowledge are analyzed, and an expandable multi-layer knowledge graph for product design (m-KGPD) is constructed and used to structurally organize cross-domain, multi-disciplinary solution knowledge and to establish a knowledge requirement-solution knowledge information retrieval channel. Secondly, the data annotation platform doccano is used to carry out knowledge text annotation and model training set construction, and based on the BERT-BiLSTM-CRF model to carry out solution knowledge entity relationship extraction, to alleviate the repetitive manual operations in the large-scale textual knowledge extraction. The knowledge graph visualization platform GraphXR is used to complete the graph construction. Finally, based on the interactive genetic algorithm (IGA), an iterative inference method is proposed to satisfy the evolution rule of interactive product design, and match the optimal solution knowledge set for the product design knowledge requirements through the hybrid human-computer interaction. The feasibility and effectiveness of the method are verified by taking the interaction design process of the complex multi-rock formation hole-forming equipment as an example.

Key words: product design, knowledge graph, interactive genetic algorithm, knowledge reasoning

CLC Number: