Patent knowledge-driven intelligent conceptual design method: innovation path based on style transfer and optimisation algorithm

The effective integration of cross-domain knowledge remains a significant challenge in the conceptual design stage. While various methods exist to leverage interdisciplinary knowledge, they often face limitations in automatically handling the heterogeneity of knowledge representation (e.g. text, images) and in supporting the automated generation of conceptual solutions. In order to solve the above problems, this study proposes a patent knowledge-driven intelligent conceptual design method. Firstly, starting from the user ‘s needs, the function analysis is carried out to determine the function of the system to be designed; secondly, according to the function of the system to be designed, relevant patents are retrieved, and natural language processing technology is used for intelligent extraction to obtain the source system and its functional information and structural information. Thirdly, the functional information and structural information obtained by the style migration algorithm are migrated as different features, and the migration results are saved for subsequent design. Finally, the computer-aided design scheme is intelligently generated, and the generation scheme is screened according to the Louvain algorithm and the integer linear programming algorithm to obtain the optimal matching path, and the proposed method is verified by the intelligent parking system.