Product innovation design based on patent knowledge graph provides an effective means of overcoming the limitations of designers' experience and supporting intelligent design. However, incomplete design paths and limited design path retrieval often lead to insufficient reuse of latent innovative knowledge during conceptual design, thereby constraining the efficiency and quality of concept generation. To address these challenges, an enhanced product innovation design approach supported by a patent knowledge graph is proposed. First, an attribute-enhanced requirement-function-effect-structure (A-RFES) ontology model is constructed to develop a comprehensive patent knowledge graph. Second, a design path completion approach based on graph data and comprehensive semantic similarity is proposed to uncover potential function-effect-structure mappings within the graph. Third, an innovation-oriented design solution approach combining depth-first and breadth-first search algorithms with the theory of inventive problem solving (TRIZ) is established to facilitate efficient design knowledge reuse and concept generation. Finally, a peristaltic pipeline inspection robot design is presented as a case study to validate the proposed approach. The validation results show that the proposed approach improves the design path completion rate and expert-evaluated design innovativeness compared with the baseline, indicating its effectiveness in improving design path completeness and supporting the generation of innovative concept schemes.
Contemporary product user demand exhibits dynamic and diversified characteristics, making the effective identification of product innovation opportunities from dynamically evolving demand data a critical challenge. To address this issue, a method for product opportunity identification and innovation strategy generation based on the dynamic evolution analysis of demand is proposed by integrating the Large Language Model (LLM) and the Temporal Graph Convolutional Network (TGCN). First, multi-granularity demand extraction is realised through the Multi-grained Requirement Feature Extraction Model (MRFEM), and a dynamic Requirement Evolution Graph (REG) is constructed. Second, the spatial semantic correlation and temporal dependency between demands are fused, and the evolution indicators such as demand evolution intensity and future attention are established. Finally, the opportunity identification and scoring method is proposed, and the demand value correction coefficient is introduced to quantify the innovation value, while the LLM is used to generate interpretable innovation strategies. A smartphone product is used as an example to verify the effectiveness and practicality of this method.
Bio-inspired design (BID) is an important approach for generating innovative product design solutions, yet its practical application remains challenging due to the lack of systematic mapping strategies and over-reliance on designers' experience. This paper proposes a systematic BID process model to support intelligent decision-making in concept generation. The methodology begins with biological prototype screening based on functional keyword search and Theory of Inventive Problem Solving (TRIZ) scientific effects, followed by quantitative similarity evaluation using fuzzy mathematics. Subsequently, a biological system function model is constructed using the Function Analysis System Technique (FAST) to analyze functional relationships among biological coupling elements. Key coupling elements are then identified by calculating their functional contribution degrees. Based on these elements, biological substance-field (BS-BF) models are established. By mapping the property characterisation functions of bio-substance (BS) and bio-field (BF), the product substance (S) and product field (F) are solved, enabling product function unit generation. Material, energy, and signal flow transmission paths are then analyzed to establish connections between function units and construct a bionic product functional structure. A case study is presented to illustrate the conceptual design generation process using the proposed method, and a user experiment is conducted to validate its effectiveness.
