AI driven dual constraint cooptimization of affective semantics and engineering parameters for biomimetic product design

人工智能驱动的仿生产品设计中情感语义和工程参数的双重约束协同优化

阅读:1

Abstract

In response to the limitation of the separation between emotion and rationality in traditional bionic design of product forms, an artificial intelligence-based dual-constraint generative design framework (AI-DCF) is proposed, which deeply integrates computer vision and visual dynamics modeling technology to achieve systematic coupling of biological semantics and engineering parameters. Firstly, an emotional semantic graph is constructed through the TextRank algorithm in natural language processing (NLP), combined with OpenCV contour detection to generate biological form tensor representations, forming a cross-modal semantic-morphological knowledge graph, breaking through the barrier of traditional design’s reliance on subjective experience. Secondly, a dual-channel optimization engine based on deep reinforcement learning (DRL) is developed, where the Alpha channel fusion technology realizes shape matching based on visual attention, and the visual dynamics model completes the inference of structural parameters guided by mechanics, and a real-time gradient feedback mechanism is embedded to form a data-algorithm-decision closed loop. Finally, taking Mingyu forklift as the experimental carrier, a comparative experiment is conducted to verify the significant advantages of the dual-constraint model over traditional methods in terms of biological feature recognizability (+ 37.5%), comprehensive performance score (+ 25%), and design iteration efficiency (+ 31.25%). These findings demonstrate case-level feasibility of aligning emotional semantics with engineering parameterization within the study’s constraints; it provides a preliminary feasibility verification for the mapping relationship between the emotional semantic descriptors based on artificial intelligence and engineering parameterization. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1038/s41598-026-42297-2.

特别声明

1、本页面内容包含部分的内容是基于公开信息的合理引用;引用内容仅为补充信息,不代表本站立场。

2、若认为本页面引用内容涉及侵权,请及时与本站联系,我们将第一时间处理。

3、其他媒体/个人如需使用本页面原创内容,需注明“来源:[生知库]”并获得授权;使用引用内容的,需自行联系原作者获得许可。

4、投稿及合作请联系:info@biocloudy.com。