基于动态偏好学习与不确定性建模的AIGC产品设计闭环优化

王沈策; 吴京; 牛虹苏

doi:10.19554/j.cnki.1001-3563.2026.14.015

PDF(5838 KB)

包装工程（设计栏目） ›› 2026, Vol. 47 ›› Issue (14) : 162-180. DOI: 10.19554/j.cnki.1001-3563.2026.14.015

工业设计

基于动态偏好学习与不确定性建模的AIGC产品设计闭环优化

王沈策, 吴京^*, 牛虹苏

作者信息 +

Closed-loop Optimization of AIGC-driven Product Design Based on Dynamic Preference Learning and Uncertainty Modeling

WANG Shence, WU Jing^*, NIU Hongsu

Author information +

文章历史 +

摘要

目的针对生成式人工智能（AIGC）驱动的产品方案生成中存在的评价不稳定、用户偏好难以动态适配以及生成结果难以持续优化等问题,提出一种面向AIGC产品设计的偏好收敛闭环优化框架。方法构建基于动态偏好学习与不确定性建模的闭环优化框架(Model for Uncertainty-aware Preference- driven AIGC Closed-loop Optimization,MUPA),形成“生成-评价-偏好学习-再生成”的端到端流程。在评价环节,引入q阶正交对模糊集（q-rung Orthopair Fuzzy Sets,q-ROF）刻画多源数据中的不确定性,并结合模糊加权零不一致法（Fuzzy-Weighted Zero-Inconsistency Method,FWZIC）、贝塔分布（Beta）与随机多准则可接受性分析（Stochastic Multicriteria Acceptability Analysis,SMAA）实现稳健赋权和概率化排序;在优化环节,引入贝叶斯动态偏好学习与差异维度提示词注入机制,实现用户反馈驱动下的迭代再生成。结果以手持式开沟机外观设计为例对比验证,所提方法在有限迭代内实现偏好的快速收敛,稳定识别优胜解,扩展解空间并提升排序一致性。结论 MUPA将不确定性建模与动态偏好学习融合,为AIGC辅助产品设计的持续优化提供了可解释、可推广的方法框架。

Abstract

To address the issues of unstable evaluation, difficulty in dynamically adapting user preferences, and limited capability for continuous optimization in AIGC-driven product concept generation, the work aims to propose a preference-convergent closed-loop optimization framework for AIGC-driven product design. The Model for Uncertainty-aware Preference-driven AIGC Closed-loop Optimization (MUPA) was constructed, forming an end-to-end process of "generation-evaluation-preference learning-regeneration". In the evaluation stage, q-rung Orthopair Fuzzy Sets (q-ROF) were employed to characterize uncertainty in multi-source data, while Fuzzy-Weighted Zero-Inconsistency (FWZIC), Beta distribution, and Stochastic Multicriteria Acceptability Analysis (SMAA) were integrated to achieve robust weighting and probabilistic ranking. In the optimization stage, Bayesian dynamic preference learning and a difference-dimension prompt injection mechanism were introduced to enable iterative regeneration driven by user feedback. A case study on the appearance design of a handheld trenching machine demonstrated that the proposed method achieved rapid preference convergence within limited iterations, reliably identified superior solutions, expanded the design space, and improved ranking consistency. By integrating uncertainty modeling with dynamic preference learning, MUPA provides an interpretable and generalizable framework for continuous optimization in AIGC-assisted product design.

导出引用1

王沈策, 吴京, 牛虹苏. 基于动态偏好学习与不确定性建模的AIGC产品设计闭环优化[J]. 包装工程. 2026, 47(14): 162-180 https://doi.org/10.19554/j.cnki.1001-3563.2026.14.015

WANG Shence, WU Jing, NIU Hongsu. Closed-loop Optimization of AIGC-driven Product Design Based on Dynamic Preference Learning and Uncertainty Modeling[J]. Packaging Engineering. 2026, 47(14): 162-180 https://doi.org/10.19554/j.cnki.1001-3563.2026.14.015

中图分类号： TB472

参考文献

[1] XIE B S, LI J X.Research on the Workability of AIGC Generated Pictures[J]. Progress in Social Sciences, 2024, 6(2): 424-441.
[2] HUANG K L, LIU Y C, DONG M Q, et al.Integrating AIGC into Product Design Ideation Teaching: An Empirical Study on Self-Efficacy and Learning Outcomes[J]. Learning and Instruction, 2024, 92: 101929.
[3] ZHU Z J, YU T, WANG Y J, et al.Revolutionizing Imaging Design Content with AIGC: User-Centered Challenges, Opportunities, and Workflow Evolution[J]. IEEE Access, 2025, 13: 87600-87620.
[4] SHEN X T, ZHANG R, ZHAO X Y, et al.PMG: Personalized Multimodal Generation with Large Language Models[C]// Proceedings of the ACM Web Conference 2024. Singapore: ACM, 2024: 3833-3843.
[5] ZHANG P, WANG H X, LI X D, et al.Research on Digital Characterization and Identification Process Model of Functional Genes for Intelligent Innovative Design[J]. Advanced Engineering Informatics, 2023, 56: 101983.
[6] LEE C H, CHEN C H, LI F, et al.Customized and Knowledge-Centric Service Design Model Integrating Case-Based Reasoning and TRIZ[J]. Expert Systems with Applications, 2020, 143: 113062.
[7] LU M L, SUN Y J, MA Z J.Multi-Objective Design Optimization of Multiple Energy Systems in Net/nearly Zero Energy Buildings under Uncertainty Correlations[J]. Applied Energy, 2024, 370: 123620.
[8] XIAO L M, HUANG G Q, ZHANG G B.Improved Assessment Model for Candidate Design Schemes with an Interval Rough Integrated Cloud Model under Uncertain Group Environment[J]. Engineering Applications of Artificial Intelligence, 2021, 104: 104352.
[9] ZHOU T T, CHEN Z H, MING X G.Multi-Criteria Evaluation of Smart Product-Service Design Concept under Hesitant Fuzzy Linguistic Environment: A Novel Cloud Envelopment Analysis Approach[J]. Engineering Applications of Artificial Intelligence, 2022, 115: 105228.
[10] TIWARI V, JAIN P K, TANDON P.An Integrated Shannon Entropy and TOPSIS for Product Design Concept Evaluation Based on Bijective Soft Set[J]. Journal of Intelligent Manufacturing, 2019, 30(4): 1645-1658.
[11] LAINWRIGHT N, PEMBERTON M. Assessing the Response Strategies of Large Language Models Under Uncertainty: A Comparative Study Using Prompt Engineering[EB/OL]. [2024-08-01](2025-12-11). https://www. researchgate.net/publication/382804673_Assessing_the_ Response_Strategies_of_Large_Language_Models_Under_ Uncertainty_A_Comparative_Study_Using_Prompt_Engineering.
[12] LAN L T H, HIEN D T T, THONG N T, et al. An ANP-TOPSIS Model for Tourist Destination Choice Problems under Temporal Neutrosophic Environment[J]. Applied Soft Computing, 2023, 136: 110146.
[13] SARWAR M, BASHIR F.Design Concept Evaluation Based on Cloud Rough Model and Modified AHP-VIKOR: An Application to Lithography Tool Manufacturing Process[J]. Advanced Engineering Informatics, 2024, 60: 102369.
[14] LIU P D, ZHU B Y, WANG P.A Weighting Model Based on Best-Worst Method and Its Application for Environmental Performance Evaluation[J]. Applied Soft Computing, 2021, 103: 107168.
[15] ALSALEM M A, ALAMOODI A H, ALBAHRI O S, et al.Evaluation of Trustworthy Artificial Intelligent Healthcare Applications Using Multi-Criteria Decision-Making Approach[J]. Expert Systems with Applications, 2024, 246: 123066.
[16] MOHAMMED R T, ZAIDAN A A, YAAKOB R, et al.Determining Importance of Many-Objective Optimisation Competitive Algorithms Evaluation Criteria Based on a Novel Fuzzy-Weighted Zero-Inconsistency Method[J]. International Journal of Information Technology & Decision Making, 2022, 21(1): 195-241.
[17] QAHTAN S, SHARIF K Y, ZAIDAN A A, et al.Novel Multi Security and Privacy Benchmarking Framework for Blockchain-Based IoT Healthcare Industry 4.0 Systems[J]. IEEE Transactions on Industrial Informatics, 2022, 18(9): 6415-6423.
[18] DEMIREL H O, GOLDSTEIN M H, LI X, et al.Human- Centered Generative Design Framework: An Early Design Framework to Support Concept Creation and Evaluation[J]. International Journal of Human-Computer Interaction, 2024, 40: 933-944.
[19] TSANG Y P, WU C H, LIN K Y, et al.Unlocking the Power of Big Data Analytics in New Product Development: An Intelligent Product Design Framework in the Furniture Industry[J]. Journal of Manufacturing Systems, 2022, 62: 777-791.
[20] CONG J C, CHEN C H, MENG X, et al.Conceptual Design of a User-Centric Smart Product-Service System Using Self-Organizing Map[J]. Advanced Engineering Informatics, 2023, 55: 101857.
[21] WANG Z X, LI X Y, ZHENG P, et al.Smart Product- Service System Configuration: A Novel Hypergraph Model-Based Approach[C]//2020 IEEE 16th International Conference on Automation Science and Engineering (CASE). Hong Kong: IEEE, 2020: 323-328.
[22] XIE X, CAI X W, ZHOU J P, et al.A Semantic-Based Method for Visualizing Large Image Collections[J]. IEEE Transactions on Visualization and Computer Graphics, 2019, 25(7): 2362-2377.
[23] ROCHLIN D.Hope and Grit: How Human-Centered Product Design Enhanced Student Mental Health[J]. California Management Review, 2024, 66(2): 108-120.
[24] 尹虎, 殷莹熙彤, 陈殿生, 等. 基于AIGC协同工业设计流程的气道廓清辅具设计研究与实践[J]. 包装工程, 2024, 45(16): 51-65.
YIN H, YIN Y, CHEN D S, et al.Research and Practice of Airway Clearance Assistive Device Design Based on AIGC Collaborative Industrial Design Process[J]. Packaging Engineering, 2024, 45(16): 51-65.
[25] WANG Y, ZHONG W, LI L, et al. Aligning Large Language Models with Human: A Survey[EB/OL]. [2023- 07-24] (2025-12-11). https://arxiv.org/abs/2307.12966.
[26] LIN L, ZHANG S H, FU S, et al.FD-LLM: Large Language Model for Fault Diagnosis of Complex Equipment[J]. Advanced Engineering Informatics, 2025, 65: 103208.
[27] ZHANG K, LIN K Y, WANG J F, et al.UNISON Framework for User Requirement Elicitation and Classification of Smart Product-Service System[J]. Advanced Engineering Informatics, 2023, 57: 101996.
[28] LI Q X, YANG Y, LI C J, et al.Energy Vehicle User Demand Mining Method Based on Fusion of Online Reviews and Complaint Information[J]. Energy Reports, 2023, 9: 3120-3130.
[29] 卢裕弘, 封颖超杰, 朱琳, 等. PromptVis: 面向文本生成图片的提示词的交互式可视分析方法[J]. 计算机辅助设计与图形学学报, 2025, 37(4): 688-696.
LU Y H, FENG Y, ZHU L, et al.PromptVis: Prompt-Based Interactive Visual Analysis Method for Text-to-Image Creation[J]. Journal of Computer-Aided Design & Computer Graphics, 2025, 37(4): 688-696.
[30] 路鹏, 吴凡, 唐建. 基于人工智能生成内容的产品造型设计与评价方法[J]. 图学学报, 2024, 45(6): 1277-1288.
LU P, WU F, TANG J.Product Design and Evaluation Methods Based on AI-Generated Content[J]. Journal of Graphics, 2024, 45(6): 1277-1288.
[31] ABDEEN W, UNTERKALMSTEINER M, WNUK K. Auxiliary Artifacts in Requirements Traceability: A Systematic Mapping Study[EB/OL].2025: arXiv: 2504.19658. https://arxiv.org/abs/2504.19658
[32] SANCHEZ-LOZANO J M, MARTINEZ M F, LARRAZ M A. A Comparison Between Fuzzy Topsis and Vikor for the Selection of Aircraft for Airspace Defense[J]. International Journal of Information Technology & Decision Making, 2025, 25(2): 121-133.
[33] SHIDPOUR H, DA CUNHA C, BERNARD A.Group Multi-Criteria Design Concept Evaluation Using Combined Rough Set Theory and Fuzzy Set Theory[J]. Expert Systems with Applications, 2016, 64: 633-644.
[34] QI J, HU J, PENG Y H.Integrated Rough VIKOR for Customer-Involved Design Concept Evaluation Combining with Customers' Preferences and Designers' Perceptions[J]. Advanced Engineering Informatics, 2020, 46: 101138.
[35] AYDOĞAN S, GÜNAY E E, AKAY D, et al. Concept Design Evaluation by Using Z-Axiomatic Design[J]. Computers in Industry, 2020, 122: 103278.
[36] KUKREJA V.Hybrid Fuzzy AHP-TOPSIS Approach to Prioritizing Solutions for Inverse Reinforcement Learning[J]. Complex & Intelligent Systems, 2023, 9(1): 493-513.
[37] SARWAR M, BASHIR F.Design Concept Evaluation Based on Cloud Rough Model and Modified AHP-VIKOR: An Application to Lithography Tool Manufacturing Process[J]. Advanced Engineering Informatics, 2024, 60: 102369.
[38] SAHA A, MISHRA A R, RANI P, et al.A Q-Rung Orthopair Fuzzy FUCOM Double Normalization-Based Multi-Aggregation Method for Healthcare Waste Treatment Method Selection[J]. Sustainability, 2022, 14(7): 4171.
[39] DEVECI M, GOKASAR I, BRITO-PARADA P R. A Comprehensive Model for Socially Responsible Rehabilitation of Mining Sites Using Q-Rung Orthopair Fuzzy Sets and Combinative Distance-Based Assessment[J]. Expert Systems with Applications, 2022, 200: 117155.
[40] LIU V, QIAO H, CHILTON L.Opal: Multimodal Image Generation for News Illustration[C]//Proceedings of the 35th Annual ACM Symposium on User Interface Software and Technology. Bend: ACM, 2022: 1-17.
[41] ZENG H P, WANG X B, WANG Y, et al.GestureLens: Visual Analysis of Gestures in Presentation Videos[J]. IEEE Transactions on Visualization and Computer Graphics, 2023, 29(8): 3685-3697.
[42] 牛晓霆, 刘祥源. 基于AIGC技术下圈椅数字化修复应用研究[J]. 家具与室内装饰, 2025, 32(3): 9-15.
NIU X T, LIU X Y.Research on the Digital Restoration and Application of Circle Chairs Based on AIGC Technology[J]. Furniture & Interior Design, 2025, 32(3): 9-15.
[43] YANG J, FAN J P, HUBBALL D, et al.Semantic Image Browser: Bridging Information Visualization with Automated Intelligent Image Analysis[C]//2006 IEEE Symposium on Visual Analytics Science and Technology. Baltimore: IEEE, 2006: 191-198.
[44] LI Z J, ZHANG Y, ZHOU S Z, et al.RealtimeGen: An Intervenable AI Image Generation System for Commercial Digital Art Asset Creators[J]. International Journal of Human-Computer Interaction, 2025, 41(11): 6613-6636.
[45] 缪翌, 张卫锋, 徐领. 基于CLIP的视频时刻检索预训练模型[J]. 计算机应用研究, 2024, 41(12): 3866-3872.
MIAO Y, ZHANG W F, XU L.Pre-Training Model for Video Moment Retrieval Based on CLIP[J]. Application Research of Computers, 2024, 41(12): 3866-3872.
[46] WANG Z, LI J S, PAN H R, et al.Research on Multimodal Generative Design of Product Appearance Based on Emotional and Functional Constraints[J]. Advanced Engineering Informatics, 2025, 65: 103106.
[47] HORVAT N, ŠKLEBAR J, ŠTORGA M, et al.Create or Revise? A Comparative Study on CAD Rework after Team-Based Engineering Design Review in Virtual Reality and Desktop Interface[J]. Advanced Engineering Informatics, 2025, 65: 103177.
[48] MARUKATAT S.Tutorial on PCA and Approximate PCA and Approximate Kernel PCA[J]. Artificial Intelligence Review, 2023, 56(6): 5445-5477.
[49] KURITA T. Principal Component Analysis (PCA)[M]// Computer Vision. Cham: Springer International Publishing, 2021: 1013-1016.
[50] YAGER R R.Generalized Orthopair Fuzzy Sets[J]. IEEE Transactions on Fuzzy Systems, 2017, 25(5): 1222-1230.
[51] PELISSARI R, OLIVEIRA M C, BEN AMOR S, et al.SMAA Methods and Their Applications: A Literature Review and Future Research Directions[J]. Annals of Operations Research, 2020, 293(2): 433-493.
[52] LI D Y, ZHANG H X, DING H, et al.User- Preference-Learning-Based Proactive Edge Caching for D2D- Assisted Wireless Networks[J]. IEEE Internet of Things Journal, 2023, 10(13): 11922-11937.
[53] ZHANG T, CHEN C, WANG D, et al.A VAE-Based User Preference Learning and Transfer Framework for Cross- Domain Recommendation[J]. IEEE Transactions on Knowledge and Data Engineering, 2023, 35(10): 10383-10396.
[54] WOIKE J K, HERTWIG R, GIGERENZER G.Heterogeneity of Rules in Bayesian Reasoning: A Toolbox Analysis[J]. Cognitive Psychology, 2023, 143: 101564.