An Intelligent Bio-Inspired Design Solution Generation Method Integrating Dual-Granularity Knowledge Bases

ZHANG Xiaoming; LIU Chengxiang; ZHONG Kai; XIE Kun; QIAN Hulei

doi:10.12263/DZXB.20250456

您当前的位置：

首页 >

文章列表页 >

An Intelligent Bio-Inspired Design Solution Generation Method Integrating Dual-Granularity Knowledge Bases

更新时间：2026-05-08

- An Intelligent Bio-Inspired Design Solution Generation Method Integrating Dual-Granularity Knowledge Bases
  增强出版
- ACTA ELECTRONICA SINICA Pages: 1-17(2026)
- 作者机构：
  
  1.安徽大学物质科学与信息技术学院，安徽合肥 230601
  2.青海理工学院，青海西宁 810016
- 作者简介：
- 基金信息：
  
  Foundation Items:　Commissioned Technology Development Project of the Chinese Academy of Sciences (No.动合(科)2025-069)
- DOI：10.12263/DZXB.20250456
  CLC： TP182;TP391
- Received：31 May 2025，
  
  Accepted：18 March 2026，
  
  Online First：07 May 2026，
- 稿件说明：
移动端阅览
ZHANG Xiaoming, LIU Chengxiang, ZHONG Kai, et al. An Intelligent Bio-Inspired Design Solution Generation Method Integrating Dual-Granularity Knowledge Bases[J/OL]. ACTA ELECTRONICA SINICA, 2026, 1-17.
DOI：

ZHANG Xiaoming, LIU Chengxiang, ZHONG Kai, et al. An Intelligent Bio-Inspired Design Solution Generation Method Integrating Dual-Granularity Knowledge Bases[J/OL]. ACTA ELECTRONICA SINICA, 2026, 1-17. DOI： 10.12263/DZXB.20250456.

摘要

仿生学通过借鉴自然界生物的优异特性，为工程创新提供了重要的方法支撑。然而，如何从丰富的生物信息中提取准确的知识并与设计需求有效对接，仍具有挑战。尽管大语言模型（Large Language Models，LLM）在自动化仿生设计过程中展现了潜力，但其依然存在局限性，包括：认知缺乏导致对仿生原理理解偏差；参数化知识约束引发时效性与准确性不足；外部知识库利用不足使得生成方案缺乏领域支撑。为此，本文提出一种融合双粒度知识库与LLM的智能化仿生设计方案生成方法（Intelligent Bio-Inspired Design，IBID），其核心是整合双粒度仿生知识库与LLM协同生成与评审机制，以实现从问题出发生成完整且有效的仿生设计方案。首先，针对知识表示与检索，本文收集生物优异特性及仿生实例，通过将每种生物策略拆分为连贯的片段，并依据主题相似性对其进行聚类，形成粗细两种粒度的知识表示，构建了双粒度仿生知识库，并设计了双粒度检索机制以提升知识检索的效率与准确性。其次，针对方案生成与优化，本文设计了分层协同的LLM工作流程：初始由基于LLM的仿生设计方案写作专家根据需求生成结构化大纲，进而设计了专家会商评审机制，由多个LLM形成普通专家组，基于知识证据独立扩展草案，再通过多轮德尔菲法进行意见交换、评估与迭代修订，以此集成多元视角、提升一致性与可靠性；最后，由基于LLM的高级专家进行终审，执行逻辑校验、细节补充与整体提升，避免观点漂移，提升方案的完整性与有效性。为验证IBID的有效性，本文在自主构建的仿生数据集上开展了实验分析，结果表明：IBID在Precision@1 （P@1）、Precision@3 （P@3）及Mean Average Precision （MAP）等检索指标上均优于最佳基线模型；在生成质量评估中，IBID在专业质量及内容相关性等大多数指标上优于GraphRAG、RAPTOR及其他基线模型。本方法有效缓解了LLM在专业领域的幻觉与知识局限问题，为高质量、高可靠的智能仿生设计提供了实用框架。

Abstract

Bio-inspired design leverages the superior characteristics of biological organisms in nature

providing essential methodological support for engineering innovation. However

extracting accurate knowledge from abundant biological information and effectively aligning it with design requirements remains challenging. Although large language models (LLMs) have demonstrated potential in automating the bio-inspired design process

they still exhibit notable limitations

including: cognitive deficiencies leading to biased understanding of bio-inspired principles; parametric knowledge constraints resulting in insufficient timeliness and accuracy; and underutilization of external knowledge bases

which leaves generated solutions lacking domain-specific support. To address these issues

this paper proposes an intelligent bio-inspired design (IBID) method that integrates a dual-granularity knowledge base with LLMs for automated design solution generation. The core of IBID lies in the synergistic integration of a dual-granularity bio-inspired knowledge base with an LLM-based collaborative generation and review mechanism

enabling the generation of complete and effective bio-inspired design solutions from problem statements. First

regarding knowledge representation and retrieval

we collect biological superior characteristics and bio-inspired instances

decompose each biological strategy into coherent segments

and cluster them based on thematic similarity to form coarse- and fine-grained knowledge representations

thereby constructing a dual-granularity bio-inspired knowledge base. A corresponding dual-granularity retrieval mechanism is designed to improve the efficiency and accuracy of knowledge retrieval. Second

regarding solution generation and optimization

we design a hierarchically collaborative LLM workflow: an LLM-based bio-inspired design writing expert initially generates a structured outline based on the requirements; subsequently

an expert consultation and review mechanism is introduced

wherein multiple LLMs form a panel of general experts that independently expand drafts based on knowledge evidence

followed by multi-round Delphi-method-based opinion exchange

evaluation

and iterative revision to integrate diverse perspectives and enhance consistency and reliability; finally

an LLM-based senior expert conducts a final review

performing logical verification

detail supplementation

and overall refinement to prevent viewpoint drift and improve the completeness and effectiveness of the solution. To validate the effectiveness of IBID

experimental analyses are conducted on a self-constructed bio-inspired dataset. The results demonstrate that IBID outperforms the best baseline models across retrieval metrics including Precision@1 (P@1)

Precision@3 (P@3)

and Mean Average Precision (MAP). In the generation quality evaluation

IBID surpasses GraphRAG

RAPTOR

and other baseline models on most metrics

including professional quality and content relevance. The proposed method effectively mitigates hallucination and knowledge limitation issues of LLMs in specialized domains

providing a practical framework for high-quality and reliable intelligent bio-inspired design.

关键词

Keywords

references

Budholiya S , Bhat A , Raj S A , et al . State of the art review about bio-inspired design and applications: An aerospace perspective [J ] . Applied Sciences , 2021 , 11 ( 11 ): 5054 .

Domel A G , Saadat M , Weaver J C , et al . Shark skin-inspired designs that improve aerodynamic performance [J ] . Journal of the Royal Society Interface , 2018 , 15 ( 139 ): 20170828 .

Latthe S S , Terashima C , Nakata K , et al . Superhydrophobic surfaces developed by mimicking hierarchical surface morphology of lotus leaf [J ] . Molecules , 2014 , 19 ( 4 ): 4256 - 4283 .

Kapoor R T , Rafatullah M , Qamar M , et al . Review on recent developments in bioinspired-materials for sustainable energy and environmental applications [J ] . Sustainability , 2022 , 14 ( 24 ): 16931 .

Ha N S , Lu Guoxing . A review of recent research on bio-inspired structures and materials for energy absorption applications [J ] . Composites Part B: Engineering , 2020 , 181 : 107496 .

Dressler F , Akan O B . A survey on bio-inspired networking [J ] . Computer Networks , 2010 , 54 ( 6 ): 881 - 900 .

Helms M , Vattam S S , Goel A K . Biologically inspired design: Process and products [J ] . Design Studies , 2009 , 30 ( 5 ): 606 - 622 .

Bar-Cohen Y . Biomimetics: Nature-based innovation [M ] . Boca Raton : CRC Press , 2016 .

Zhang Taiyao . Biologically-inspired design in engineering: Current perspective on biomimicry applications [C ] // Proceedings of 2024 International Conference on Mechanics, Electronics Engineering and Automation . [S.l.] : Atlantis Press , 2024 : 637 - 645 .

Wegst U G K , Bai Hao , Saiz E , et al . Bioinspired structural materials [J ] . Nature Materials , 2015 , 14 ( 1 ): 23 - 36 .

Ng L , Elgar M A , Stuart-Fox D . From bioinspired to bioinformed: Benefits of greater engagement from biologists [J ] . Frontiers in Ecology and Evolution , 2021 , 9 : 790270 .

Shu L H , Ueda K , Chiu I , et al . Biologically inspired design [J ] . CIRP Annals , 2011 , 60 ( 2 ): 673 - 693 .

Cohen Y H , Reich Y , Greenberg S . Biomimetics: Structure-function patterns approach [J ] . Journal of Mechanical Design , 2014 , 136 ( 11 ): 111108 .

Liu N F , Lin K , Hewitt J , et al . Lost in the middle: How language models use long contexts [J ] . Transactions of the Association for Computational Linguistics , 2024 , 12 : 157 - 173 .

Shi F , Chen X , Misra K , et al . Large language models can be easily distracted by irrelevant context [C ] // International Conference on Machine Learning . PMLR , 2023 : 31210 - 31227 .

Li D , Shao R , Xie A , et al . How long can context length of open-source LLMs truly promise [C ] // NeurIPS 2023 Workshop on Instruction Tuning and Instruction Following . 2023 .

Li J , Wang M , Zheng Z , et al . Loogle: Can long-context language models understand long contexts? [C ] // Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) . 2024 : 16304 - 16333 .

Wang Yidong , Guo Qi , Yao Wenjin , et al . AutoSurvey: Large language models can automatically write surveys [C ] // Proceedings of the 38th International Conference on Neural Information Processing Systems . New York : Curran Associates Inc. , 2024 : 3655 .

Shao Yijia , Jiang Yucheng , Kanell T , et al . Assisting in writing Wikipedia-like articles from scratch with large language models [C ] // Proceedings of 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies . [S.l.] : Association for Computational Linguistics , 2024 .

Kandpal N , Deng H , Roberts A , et al . Large language models struggle to learn long-tail knowledge [C ] // International Conference on Machine Learning . PMLR , 2023 : 15696 - 15707 .

Gao Yunfan , Xiong Yun , Gao Xinyu , et al . Retrieval-augmented generation for large language models: A survey [PP/OL ] . arXiv ( 2024-03-27 ). https://arxiv.org/abs/2312.10997 https://arxiv.org/abs/2312.10997 .

Kang H B , Lin D C E , Martelaro N , et al . BioSpark: An end-to-end generative system for biological-analogical inspirations and ideation [C ] // Extended Abstracts of the CHI Conference on Human Factors in Computing Systems . New York : ACM , 2024 : 61 .

Luu R K , Buehler M J . BioinspiredLLM: Conversational large language model for the mechanics of biological and bio-inspired materials [J ] . Advanced Science , 2024 , 11 ( 10 ): 2306724 .

Toukmaji C , Tee A . Retrieval-augmented generation and LLM agents for biomimicry design solutions [J ] . Proceedings of the AAAI Symposium Series , 2024 , 3 ( 1 ): 273 - 278 .

Chen Liuqing , Jiang Zhaojun , Xia Duowei , et al . BIDTrainer: An LLMs-driven education tool for enhancing the understanding and reasoning in bio-inspired design [C ] // Proceedings of 2024 CHI Conference on Human Factors in Computing Systems . New York : ACM , 2024 : 676 .

Ram O , Levine Y , Dalmedigos I , et al . In-context retrieval-augmented language models [J ] . Transactions of the Association for Computational Linguistics , 2023 , 11 : 1316 - 1331 .

Izacard G , Lewis P , Lomeli M , et al . Atlas: Few-shot learning with retrieval augmented language models [J ] . The Journal of Machine Learning Research , 2023 , 24 ( 1 ): 251 .

Gao T , Yen H , Yu J , et al . Enabling large language models to generate text with citations [C ] // Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing . 2023 : 6465 - 6488 .

Huang Lei , Yu Weijiang , Ma Weitao , et al . A survey on hallucination in large language models: Principles, taxonomy, challenges, and open questions [J ] . ACM Transactions on Information Systems , 2025 , 43 ( 2 ): 42 .

Dong Yuxin , Wang Shuo , Zheng Hongye , et al . Advanced RAG models with graph structures: Optimizing complex knowledge reasoning and text generation [C ] // Proceedings of the 2024 5th International Symposium on Computer Engineering and Intelligent Communications . Piscataway : IEEE , 2024 : 626 - 630 .

Ma X , Gong Y , He P , et al . Query rewriting in retrieval-augmented large language models [C ] // Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing . 2023 : 5303 - 5315 .

Gao L , Ma X , Lin J , et al . Precise zero-shot dense retrieval without relevance labels, 2022 [J ] . URL https://arxiv. org/abs/2212.10496 , 2022 .

Shao Zhihong , Gong Yeyun , Shen Yelong , et al . Enhancing retrieval-augmented large language models with iterative retrieval-generation synergy [C ] // Findings of the Association for Computational Linguistics: EMNLP 2023 . [S.l.] : Association for Computational Linguistics , 2023 : 9248 - 9274 .

Yu Wenhao , Iter D , Wang Shuohang , et al . Generate rather than retrieve: Large language models are strong context generators [C/OL ] // Proceedings of the 11th International Conference on Learning Representations , 2023 . https://dblp.org/db/conf/iclr/iclr2023.html#conf/iclr/0002IWXJ000023 https://dblp.org/db/conf/iclr/iclr2023.html#conf/iclr/0002IWXJ000023 .

Sarthi P , Abdullah S , Tuli A , et al . RAPTOR: Recursive abstractive processing for tree-organized retrieval [C/OL ] // Proceedings of the 12th International Conference on Learning Representations , 2024 . https://dblp.org/db/conf/iclr/iclr2024.html#conf/iclr/SarthiATKGM24 https://dblp.org/db/conf/iclr/iclr2024.html#conf/iclr/SarthiATKGM24 .

Edge D , Trinh H , Cheng N , et al . From local to global: A graph rag approach to query-focused summarization [PP/OL ] . arXiv ( 2025-02-19 ). https://arxiv.org/abs/2404.16130 https://arxiv.org/abs/2404.16130 .

Deldin J M , Schuknecht M . The AskNature database: Enabling solutions in biomimetic design [M ] //Goel A K, McAdams D A, Stone R B. Biologically inspired design: Computational methods and tools . London : Springer , 2014 : 17 - 27 .

Kaufman L , Rousseeuw P J . Finding groups in data: An introduction to cluster analysis [M ] . New York : Wiley , 1990 .

郭哲 , 张智博 , 周炜杰 , 等 . 融合引导注意力的中文长文本摘要生成 [J ] . 电子学报 , 2024 , 52 ( 12 ): 3914 - 3930 .

Guo Zhe , Zhang Zhibo , Zhou Weijie , et al . Chinese long text summarization with guided attention [J ] . Acta Electronica Sinica , 2024 , 52 ( 12 ): 3914 - 3930 .

Kim S , Suk J , Longpre S , et al . Prometheus 2: An open source language model specialized in evaluating other language models [C ] // Proceedings of 2024 Conference on Empirical Methods in Natural Language Processing . [S.l.] : Association for Computational Linguistics , 2024 : 4334 - 4353 .

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

A Survey on the Applications of Large Language Models in Electronic Design Automation

Research Progress on Blockchain-Empowered Data Storage Security Service

A Temporal-Pulse Amplitude Clustering-Based Unknown Radar Emitter Signal Sorting Method

LLM-Based Zero-Shot Imputation of Spatiotemporal Data

Multi-Party Private Table Screening for LLM-Driven ODQA: An Enhanced Method with MPC, Publicly Aggregable Audit, and Dynamic Reputation

Related Author

WANG Jing

LIN Liyu

LI Lei

CHEN Linyi

LI Zheng

LAI Yao

LI Yan

YAO Jiafei

Related Institution

School of Computing and Data Science, The University of Hong Kong

School of Integrated Circuits and Microelectronic Innovation, Fudan University

School of Integrated Circuit Science and Engineering, Nanjing University of Posts and Telecommunications

School of Cyberspace Security, Guangzhou University

The 29th Research Institute of China Electronics Technology Group Corporation

⁰