Biological Knowledge-enabled BERT for Innovation in Biomimetic Design: A Case Study

分类： 机械工程 >> 机械设计 提交时间： 2021-11-17

Sun,Feng Xu,He

摘要： The biomimetic design provides an adequate solution to achieve an excellent design. However, the prototype space for biomimetic design is relatively large, and it becomes more and more challenging to find the required biological prototypes quickly and efficiently. In order to improve the design efficiency and enrich the means of biomimetic innovation, this paper proposes a biological knowledge-enabled bidirectional encoder representation from transformers (BERT) model to assist biomimetic design, namely BioDesign. We extract the biological strategies, functions and extract dimensional information from the Asknature as the data source. The linguistic expression model-BERT was used to recommend biomimetic strategies or functions combined with the biological strategies data. Finally, we take the biomimetic erosion wear resistant design of the valve core as an example and use the proposed BioDesign model to recommend biomimetic inspired functions. According to the recommended content of the BioDesign model, we obtained the erosion wear resistance strategies and designed the biomimetic structure. The erosion wear experiment proved the feasibility and effectiveness of the proposed method.

同行评议状态:待评议

 点击量 5193  下载量 930  评论