目的 针对注射成形工艺调试依赖经验、试错成本高、周期长的问题,提出知识库增强的生成式大模型智能优化方法,旨在探索人工智能在制造工艺优化中的应用。方法 构建融合领域知识与实时反馈的优化框架,首先自动搜集注射成形工艺相关中英文文献共819篇;其次通过文档切分、嵌入模型处理,将文档信息转化为大模型可理解的向量知识库;最后结合本地部署的生成式大模型,根据制品实际与目标质量的偏差,动态生成参数调整建议,并提供推理依据。结果 实验结果表明,所提出的知识库增强生成式大模型方法在PC和PMMA等多种材料的注射成形优化中均能在3轮内精准收敛至目标质量。与无知识库增强的大模型相比,后者因采用参数空间中心点初始化且缺乏对材料特性和成形机理的理解,导致初始保压压力设置过低、首轮偏差较大,并出现参数反向调整等不合理操作;与传统人工试错方法相比,后者依赖经验判断,优化路径主观性强,在相同任务中需7轮以上迭代才能达标,而本文方法通过知识库增强的自动推理显著缩短了调试周期。结论 本文提出的方法将工艺机理与智能推理相融合,显著提升了注射成形工艺优化的效率与泛化能力,为大模型在工业制造中的可靠应用提供了有效路径。
Abstract
The work aims to propose a knowledge-enhanced generative large model-based intelligent optimization method, to explore the application of artificial intelligence in the optimization of manufacturing processes and address the challenges in injection molding process tuning, such as heavy reliance on empirical knowledge, high trial-and-error costs, and long optimization cycles. An optimization framework that integrated domain knowledge and real-time feedback was constructed. Firstly, 819 Chinese and English technical documents on injection molding were automatically collected; Then, these documents were chunked and embedded into a vectorized knowledge base accessible to the large model; Finally, combining locally deployed generative large model, parameter adjustment recommendations were dynamically generated based on the deviation between actual and target part weight, accompanied by interpretable reasoning. Experimental results showed that the proposed method achieved precise convergence to the target weight within three iterations for multiple materials, including PC and PMMA. In contrast, a knowledge-agnostic large model, initialized at the center of the parameter space and lacking understanding of material properties and molding physics set an excessively low initial packing pressure, leading to low initial holding pressure, large first-round deviations and even counterproductive adjustments. Compared with traditional manual trial-and-error, which relies on subjective experience and requires more than seven iterations to meet the target, our approach significantly shortens the tuning cycle through knowledge-guided automated reasoning. By effectively fusing process mechanism knowledge with intelligent inference, the method substantially enhances the efficiency and generalization capability of injection molding optimization, offering a reliable pathway for deploying large models in industrial manufacturing.
关键词
检索增强生成 /
生成式大模型 /
工艺参数优化 /
注射成形 /
高效收敛
Key words
retrieval-augmented generation /
generative large model /
process parameter optimization /
injection molding /
effective convergence
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] TSOU H H, HUANG C C, ZHAO T W, et al.Design and Validation of Sensor Installation for Online Injection Molding Sidewall Deformation Monitoring[J]. Measurement, 2022, 205: 112200.
[2] 刘怀举, 卢泽华, 朱才朝. 塑料齿轮传动高承载技术发展与应用[J]. 中国机械工程, 2025, 36(1): 2-17.
LIU H J, LU Z H, ZHU C C.State-of-the-Art and Trend of High Loading Capacity Plastic Gear Drives[J]. China Mechanical Engineering, 2025, 36(1): 2-17.
[3] 郭昊, 陈浩, 杨卫民. 汽车轻量化与高分子材料加工的创新发展机遇[J]. 石油化工, 2025, 54(10): 1486-1490.
GUO H, CHEN H, YANG W M.Innovative Development Opportunities of Automotive Lightweighting and Polymer Material Processing[J]. Petrochemical Technology, 2025, 54(10): 1486-1490.
[4] KHOSRAVANI M R, NASIRI S.Injection Molding Manufacturing Process: Review of Case-Based Reasoning Applications[J]. Journal of Intelligent Manufacturing, 2020, 31(4): 847-864.
[5] 黄佳文, 孙瑞, 阮宇飞. 基于PCA与K-Means的注射成形制品质量在线检测[J]. 电子技术与软件工程, 2021, 10(21): 117-120.
HUANG J W, SUN R, RUAN Y F.On-Line Quality Inspection of Injection Molded Products Based on PCA and K-Means[J]. Electronic Technology & Software Engineering, 2021, 10(21): 117-120.
[6] 吴俊超. 基于Critic权重法和阀浇口开启时序的车门内饰板注塑优化分析[J]. 工程塑料应用, 2025, 53(10): 141-149.
WU J C.Optimization Analysis of Injection Molding of Automobile Door Interior Panel Based on Critic Weight Method and Valve Gate Opening Time Sequence[J]. Engineering Plastics Application, 2025, 53(10): 141-149.
[7] GASPAR-CUNHA A, MELO J, MARQUES T, et al.A Review on Injection Molding: Conformal Cooling Channels, Modelling, Surrogate Models and Multi-Objective Optimization[J]. Polymers, 2025, 17(7): 919.
[8] 邓建新, 刘光明, 王令, 等. 制造工艺参数的智能优化设计方法进展[J]. 制造技术与机床, 2023(5): 74-80.
DENG J X, LIU G M, WANG L, et al.Research Progress of Intelligent Optimization Design of Manufacturing Process Parameters[J]. Manufacturing Technology & Machine Tool, 2023(5): 74-80.
[9] HOPMANN C, KÖBEL T. Influence of the Injection Velocity Profile on the Properties of Injection Moulded Parts[J]. International Polymer Processing, 2024, 39(3): 378-391.
[10] CHEN Y Y, HU S C, LI A S, et al.Parameters Optimization of Electrical Discharge Machining Process Using Swarm Intelligence: A Review[J]. Metals, 2023, 13(5): 839.
[11] ZHU J J, QIU Z W, HUANG Y Z, et al.Overview of Injection Molding Process Optimization Technology[J]. Journal of Physics: Conference Series, 2021, 1798(1): 012042.
[12] LAZIM H M, JUSOH M S, AHAMAD M S F, et al. Optimization of Warpage Defects Using the Taguchi Method: A Failure Analysis in Plastic Injection Molding[J]. International Journal of Innovative Research and Scientific Studies, 2025, 8(2): 867-877.
[13] MOAYYEDIAN M, DINC A, MAMEDOV A.Optimization of Injection-Molding Process for Thin-Walled Polypropylene Part Using Artificial Neural Network and Taguchi Techniques[J]. Polymers, 2021, 13(23): 4158.
[14] YANG W K, LU S H, LIU W H.Optimization Method to Select Temperature Based on Chemorheological and Exothermal Reaction of RTM[J]. Journal of Applied Polymer Science, 2019, 136(46): 48245.
[15] WANG X Y, LI H X, GU J F, et al.Pressure Analysis of Dynamic Injection Molding and Process Parameter Optimization for Reducing Warpage of Injection Molded Products[J]. Polymers, 2017, 9(3): 85.
[16] RAIMI O A, LEE B K. Artificial Neural Network (ANN)-Based Prediction Model of Demolding Force in Injection Molding Process[J]. Advances in Polymer Technology, 2025, 2025: 1528204.
[17] JUNG J, PARK K, CHO B, et al.Optimization of Injection Molding Process Using Multi-Objective Bayesian Optimization and Constrained Generative Inverse Design Networks[J]. Journal of Intelligent Manufacturing, 2023, 34(8): 3623-3636.
[18] HEINISCH J, LOCKNER Y, HOPMANN C.Comparison of Design of Experiment Methods for Modeling Injection Molding Experiments Using Artificial Neural Networks[J]. Journal of Manufacturing Processes, 2021, 61: 357-368.
[19] DONG Z Y, ZHAO P, ZHENG J G, et al.Intelligent Injection Molding: Parameters Self-Learning Optimization Using Iterative Gradient-Approximation Adaptive Method[J]. Journal of Applied Polymer Science, 2021, 138(29): 50687.
[20] 金镖, 潘毅峰, 姚涵非, 等. 注射成形中的工艺参数自适应优化方法[J]. 精密成形工程, 2025, 17(5): 220-228.
JIN B, PAN Y F, YAO H F, et al.A Self-Learning Optimization Method for Process Parameters in Electric Driven Injection Molding[J]. Journal of Netshape Forming Engineering, 2025, 17(5): 220-228.
[21] MALLIKARJUN S C, PERIYASAMY B K, BENSINGH R J.A Review of Materials, Manufacturing, and Process Optimization for Polymeric Ophthalmoscope Lenses[J]. Polymer-Plastics Technology and Materials, 2025, 64(16): 2485-2517.
[22] ZHAO N Y, LIAN J Y, WANG P F, et al.Recent Progress in Minimizing the Warpage and Shrinkage Deformations by the Optimization of Process Parameters in Plastic Injection Molding: A Review[J]. The International Journal of Advanced Manufacturing Technology, 2022, 120(1): 85-101.
[23] 张熙, 杨小汕, 徐常胜. ChatGPT及生成式人工智能现状及未来发展方向[J]. 中国科学基金, 2023, 37(5): 743-750.
ZHANG X, YANG X S, XU C S.Current State and Future Development Directions of ChatGPT and Generative Artificial Intelligence[J]. Bulletin of National Natural Science Foundation of China, 2023, 37(5): 743-750.
[24] 曹平. 工业控制系统中人工智能的应用前景[J]. 石化技术, 2025, 32(10): 376-378.
CAO P.Application Prospect of Artificial Intelligence in Industrial Control System[J]. Petrochemical Industry Technology, 2025, 32(10): 376-378.
[25] PERES R S, JIA X D, LEE J, et al.Industrial Artificial Intelligence in Industry 4.0-Systematic Review, Challenges and Outlook[J]. IEEE Access, 2020, 8: 220121-220139.
基金
国家自然科学基金重点项目（52535010）; 宁波市“科创甬江2035”重点研发计划（2025Z010）; 浙江省创新创业领军人才团队（2024R01002）
PDF(676 KB)
渝公网安备 50010702501719号 渝ICP备15012534号-4
