Learning quality characteristics for plastic injection molding processes using a combination of simulated and measured data

被引:43
作者
Finkeldey, Felix [1 ]
Volke, Julia [2 ]
Zarges, Jan-Christoph [2 ]
Heim, Hans-Peter [2 ]
Wiederkehr, Petra [1 ]
机构
[1] TU Dortmund Univ, Chair Software Engn, Virtual Machining, D-44227 Dortmund, Germany
[2] Univ Kassel, Inst Mat Engn, Polymer Engn, D-34125 Kassel, Germany
关键词
Injection molding; Artificial intelligence; Machine learning; Predictive models; Simulation; OPTIMIZATION; PREDICTION; MODEL; REGRESSION; SELECTION; MACHINE; SYSTEM;
D O I
10.1016/j.jmapro.2020.10.028
中图分类号
T [工业技术];
学科分类号
08 ;
摘要
During the initial sampling of injection molds, the determination of suitable process parameter values to achieve a desired quality of the resulting parts, can be a time-consuming and demanding task. This is due to the complex viscoelastic properties of injection molding processes. Conducting technological investigations and using simulation techniques are popular approaches to support the design of the regarded process. However, while the former approach can require extensive research efforts, it can be difficult to design simulations and validate their prediction accuracy, especially when few process measurements are available as a baseline. In addition, the knowledge obtained by both, simulation and technologically based approaches, is only valid for the analyzed process configurations. In contrast, models based on machine learning (ML) approaches can provide forecasts for previously unseen data and can be evaluated quickly. Unfortunately, a high amount of data is required to train such models reasonably. In this contribution, a novel ML-based methodology to predict quality characteristics of an injection molding process for different process parameter values using an intelligent combination of simulation data and measurements, is presented.
引用
收藏
页码:134 / 143
页数:10
相关论文
共 47 条
[1]  
Bergstra J, 2012, J MACH LEARN RES, V13, P281
[2]   A virtual prototyping environment for a robust design of an injection moulding process [J].
Berti, Guido ;
Monti, Manuel .
COMPUTERS & CHEMICAL ENGINEERING, 2013, 54 :159-169
[3]  
Bhat HA, 2014, Int. J. Res. Eng. Technol, V3, P366
[4]   Random forests [J].
Breiman, L .
MACHINE LEARNING, 2001, 45 (01) :5-32
[5]  
Breiman L, 1996, MACH LEARN, V24, P123, DOI 10.1007/BF00058655
[6]  
Breiman L, 1984, Classification and Regression Trees, V1st, DOI DOI 10.1201/9781315139470
[7]  
Breiman Leo, 1997, Technical report
[8]   The origins of the Gini index: extracts from VariabilitA e MutabilitA (1912) by Corrado Gini [J].
Ceriani, Lidia ;
Verme, Paolo .
JOURNAL OF ECONOMIC INEQUALITY, 2012, 10 (03) :421-443
[9]   XGBoost: A Scalable Tree Boosting System [J].
Chen, Tianqi ;
Guestrin, Carlos .
KDD'16: PROCEEDINGS OF THE 22ND ACM SIGKDD INTERNATIONAL CONFERENCE ON KNOWLEDGE DISCOVERY AND DATA MINING, 2016, :785-794
[10]   A neural network-based approach for dynamic quality prediction in a plastic injection molding process [J].
Chen, Wen-Chin ;
Tai, Pei-Hao ;
Wang, Min-Wen ;
Deng, Wei-Jaw ;
Chen, Chen-Tai .
EXPERT SYSTEMS WITH APPLICATIONS, 2008, 35 (03) :843-849