基于半监督学习双模型结构的注塑产品异常检测

doi:10.3969/j.issn.1004-132X.2025.03.021

中国机械工程 ›› 2025, Vol. 36 ›› Issue (03): 576-583.DOI: 10.3969/j.issn.1004-132X.2025.03.021

基于半监督学习双模型结构的注塑产品异常检测

陈昱¹;项薇^1,2*;林文文¹;龚川¹;张怀志¹;虞任豪¹

1.宁波大学机械工程与力学学院，宁波，315211
2.浙江省零件轧制成形技术研究重点实验室，宁波，315211

出版日期:2025-03-25 发布日期:2025-04-23
作者简介:陈昱，男，1999年生，硕士研究生。研究方向为制造系统工程。E-mail：15058023546@163.com。

Abnormal Detection of Injection Molding Products Based on Semi-supervised Learning Dual-model Structure

CHEN Yu¹;XIANG Wei^1,2*;LIN Wenwen¹;GONG Chuan¹;ZHANG Huaizhi¹;YU Renhao¹

1.School of Mechanical Engineering and Mechanics,Ningbo University,Ningbo,Zhejiang,315211
2.Zhejiang Provincial Key Laboratory of Part Rolling Technology,Ningbo,Zhejiang,315211

Online:2025-03-25 Published:2025-04-23

摘要/Abstract

摘要： 质量数据分布的不平衡及分类边界的模糊性限制了传统分类器的性能，阻碍了企业智能生产决策的高效实施。为此，提出了一种基于双模型结构的深度生成模型异常检测方法，根据尺寸数据分布将合格产品等级进行二分类，即优秀及次优，分别用于训练两个深度生成模型，考虑数据分布特点设计加权集成，基于计算的异常分数对产品进行合格性判定。以变分自编码器（VAE）、Wasserstein生成对抗网络(WGAN)为子模型开发了两个双模型结构，测试结果显示，相较于单模型结构，基于双模型的VAE和WGAN在测试集上的分类准确率分别提高了4.5%和6%。

关键词: 产品质量, 异常检测, 变分自编码器, Wasserstein生成对抗网络, 双模型结构

Abstract: The imbalance in quality data distribution and the ambiguity of classification boundaries were identified as the factors limiting the performance of traditional classifiers, which hindered the efficient implementation of intelligent production decision-making in enterprises. Therefore, a dual-model-based deep generative model anomaly detection method was proposed. Qualified products were classified into two levels, excellent and sub-optimal, based on size data distribution, and two deep generative models were trained separately. A weighted ensemble strategy was designed considering characteristics of data distribution, and anomaly scores were calculated to determine product qualification. Two dual-model structures were developed based on VAE and WGAN. Experimental results indicate that, compared to single-model structures, the dual-model-based VAE and WGAN improve classification accuracy on the test sets by 4.5% and 6%, respectively.

Key words: product quality, anomaly detection, variational autoencoder（VAE）, Wasserstein generative adversarial network(WGAN), dual-model structure

中图分类号:

TP181

陈昱1, 项薇1, 2, 林文文1, 龚川1, 张怀志1, 虞任豪1. 基于半监督学习双模型结构的注塑产品异常检测[J]. 中国机械工程, 2025, 36(03): 576-583.

CHEN Yu1, XIANG Wei1, 2, LIN Wenwen1, GONG Chuan1, ZHANG Huaizhi1, YU Renhao1. Abnormal Detection of Injection Molding Products Based on Semi-supervised Learning Dual-model Structure[J]. China Mechanical Engineering, 2025, 36(03): 576-583.

导出引用管理器 EndNote|Ris|BibTeX

链接本文: https://www.cmemo.org.cn/CN/10.3969/j.issn.1004-132X.2025.03.021

https://www.cmemo.org.cn/CN/Y2025/V36/I03/576

参考文献

［1］HSIEH R J, CHOU J, HO C H. Unsupervised Online Anomaly Detection on Multivariate Sensing Time Series Data for Smart Manufacturing［C］∥2019 IEEE 12th Conference on Service-Oriented Computing and Applications (SOCA). Kaohsiung, 2019:90-97.
［2］GUO Yu, ZHANG Jundong, SUN Bin, et al. Adversarial Deep Transfer Learning in Fault Diagnosis:Progress, Challenges, and Future Prospects［J］. Sensors, 2023, 23(16):7263.
［3］CHEN Tingting, LIU Xueping, XIA Bizhong, et al. Unsupervised Anomaly Detection of Industrial Robots Using Sliding-window Convolutional Variational Autoencoder［J］. IEEE Access, 2020, 8:47072-47081.
［4］TANG Weitao, VIAN C M, TANG Ziyang, et al. Anomaly Detection of Core Failures in Die Casting X-ray Inspection Images Using a Convolutional Autoencoder［J］. Machine Vision and Applications, 2021, 32(4):102.
［5］SONG J, LEE Y C, LEE J. Deep Generative Model with Time Series-image Encoding for Manufacturing Fault Detection in Die Casting Process［J］. Journal of Intelligent Manufacturing, 2023, 34(7):3001-3014.
［6］XU Caie, NI Dandan, WANG Bingyan, et al. Two-stage Anomaly Detection for Positive Samples and Small Samples Based on Generative Adversarial Networks［J］. Multimedia Tools and Applications, 2023, 82(13):20197-20214.
［7］LEHR J, SARGSYAN A, PAPE M, et al. Automated Optical Inspection Using Anomaly Detection and Unsupervised Defect Clustering［C］∥2020 25th IEEE International Conference on Emerging Technologies and Factory Automation(ETFA). Vienna, 2020:1235-1238.
［8］CHEN Xuanhao, DENG Liwei, HUANG Feiteng, et al. DAEMON:Unsupervised Anomaly Detection and Interpretation for Multivariate Time Series［C］∥2021 IEEE 37th International Conference on Data Engineering(ICDE). Chania, 2021:2225-2230.
［9］刘永兴, 唐小琦, 钟靖龙, 等. 基于轻量级梯度提升机的非对称风险注塑成形产品尺寸预测模型［J］. 中国机械工程, 2022, 33(8):965-969.
LIU Yongxing, TANG Xiaoqi, ZHONG Jinglong, et al. Asymmetric Risk Injection Molding Product Size Prediction Model Based on LightGBM［J］. China Mechanical Engineering, 2022, 33(8):965-969.
［10］来杰, 王晓丹, 向前, 等. 自编码器及其应用综述［J］. 通信学报, 2021, 42(9):218-230.
LAI Jie, WANG Xiaodan, XIANG Qian, et al. Review on Autoencoder and Its Application［J］. Journal on Communications, 2021, 42(9):218-230.
［11］崔建明, 余茜, 刘铭. 基于WGAN的生成式信息隐写方法研究［J］. 河南理工大学学报(自然科学版), 2023, 42(3):146-152.
CUI Jianming, YU Xi, LIU Ming. An Innovative Generative Information Steganography Method Based on Wasserstein GAN［J］. Journal of Henan Polytechnic University (Natural Science), 2023, 42(3):146-152.
［12］KIM J, LEE J Y. Development of a Quality Prediction Algorithm for an Injection Molding Process Considering Cavity Sensor and Vibration Data［J］. International Journal of Precision Engineering and Manufacturing, 2023, 24(6):901-914.

[1]	陈昱, 项薇, 龚川. 基于数据挖掘的注塑产品质量在线故障检测及预测[J]. 中国机械工程, 2023, 34(14): 1749-1755.
[2]	余石龙, 鲍劲松, 李婕, 张启华. 基于误差注意力的晶圆制造数据异常检测[J]. 中国机械工程, 2020, 31(14): 1686-1692.
[3]	顾煜炯, 宋磊, 徐天金, 雷启龙. 变工况条件下的风电机组齿轮箱故障预警方法[J]. 中国机械工程, 2014, 25(10): 1346-1351,1405.
[4]	李翀, 刘思峰, 方志耕. 基于产品可靠性和时效性的闭环供应链系统策略研究[J]. 中国机械工程, 2012, 23(13): 1587-1592.
[5]	邓军, 余忠华, 吴昭同. 产品质量与过程质量的映射研究 [J]. 中国机械工程, 2010, 21(17): 2070-2074.

基于半监督学习双模型结构的注塑产品异常检测

Abnormal Detection of Injection Molding Products Based on Semi-supervised Learning Dual-model Structure

PDF

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 5

编辑推荐

Metrics