32 Prediction of Width and Thickness of Injection Molded …
467
(Tercan 2018). This might help to increase the generalization abilities of the models
so that they are able to predict dimensions and final quality not only for the dogbone
specimens from HDPE but for a bigger variety of products and materials. It is also
planned to do extended optimization of the model’s parameters to possibly increase
the prediction quality.
32.6 Conclusion
Injection molding is one of the major processes for the manufacturing of products
from thermoplastic polymers. It is easily suited for the production of large amounts
of plastic parts of different shapes and sizes (Mathivanan et al. 2010). However,
the final quality of the plastic product depends on many factors, such as injection
molding machine, mold in use and parameter settings (Liu 2017). Choosing the
correct production parameters might be a complicated task leading to production of
large quantities of scrap. That is why it is important to predict the final part quality
based on the chosen process parameters.
This study presents an approach that involves the application of the machine
learning methods, such as REPTree decision tree, random forest, k-nearest neighbors
and multilayer perceptron ANN to create prediction models for the thickness and
width of the HDPE dogbone specimens.
The model assessment is based on the application of 10-folds cross-validation, as
well as correlation coefficient, mean absolute error and root square mean error. For
all the algorithms some of their parameters were varied to see how it would influence
the quality of the obtained prediction model.
The results of the study indicate that such an approach has the potential to facilitate the process of finding optimal production conditions and decrease amounts of
produced plastic scrap in the field of thermoplastics injection molding.
References
Altan M (2010) Reducing shrinkage in injection moldings via the Taguchi, ANOVA and neural
network methods. Mater Des 31(1):599–604
Berti G, Monti M (2013) A virtual prototyping environment for a robust design of an injection
moulding process. Comput Chem Eng 54:159–169
Breiman L (2001) Random forests. Mach Learn 45(1):5–32
Chen W-C et al (2008) A neural network-based approach for dynamic quality prediction in a plastic
injection molding process. Expert Syst Appl 35(3):843–849
Cheng J, Liu Z, Tan J (2013) Multiobjective optimization of injection molding parameters based
on soft computing and variable complexity method. Int J Adv Manuf Technol 66(5–8):907–916
Fernandes C et al (2018) Modeling and optimization of the injection-molding process: a review.
Adv Polym Technol 37(2):429–449
Fu J, Ma Y (2018) Computer-aided engineering analysis for early-ejected plastic part dimension
prediction and quality assurance. Int J Adv Manuf Technol 98(9–12):2389–2399
467
(Tercan 2018). This might help to increase the generalization abilities of the models
so that they are able to predict dimensions and final quality not only for the dogbone
specimens from HDPE but for a bigger variety of products and materials. It is also
planned to do extended optimization of the model’s parameters to possibly increase
the prediction quality.
32.6 Conclusion
Injection molding is one of the major processes for the manufacturing of products
from thermoplastic polymers. It is easily suited for the production of large amounts
of plastic parts of different shapes and sizes (Mathivanan et al. 2010). However,
the final quality of the plastic product depends on many factors, such as injection
molding machine, mold in use and parameter settings (Liu 2017). Choosing the
correct production parameters might be a complicated task leading to production of
large quantities of scrap. That is why it is important to predict the final part quality
based on the chosen process parameters.
This study presents an approach that involves the application of the machine
learning methods, such as REPTree decision tree, random forest, k-nearest neighbors
and multilayer perceptron ANN to create prediction models for the thickness and
width of the HDPE dogbone specimens.
The model assessment is based on the application of 10-folds cross-validation, as
well as correlation coefficient, mean absolute error and root square mean error. For
all the algorithms some of their parameters were varied to see how it would influence
the quality of the obtained prediction model.
The results of the study indicate that such an approach has the potential to facilitate the process of finding optimal production conditions and decrease amounts of
produced plastic scrap in the field of thermoplastics injection molding.
References
Altan M (2010) Reducing shrinkage in injection moldings via the Taguchi, ANOVA and neural
network methods. Mater Des 31(1):599–604
Berti G, Monti M (2013) A virtual prototyping environment for a robust design of an injection
moulding process. Comput Chem Eng 54:159–169
Breiman L (2001) Random forests. Mach Learn 45(1):5–32
Chen W-C et al (2008) A neural network-based approach for dynamic quality prediction in a plastic
injection molding process. Expert Syst Appl 35(3):843–849
Cheng J, Liu Z, Tan J (2013) Multiobjective optimization of injection molding parameters based
on soft computing and variable complexity method. Int J Adv Manuf Technol 66(5–8):907–916
Fernandes C et al (2018) Modeling and optimization of the injection-molding process: a review.
Adv Polym Technol 37(2):429–449
Fu J, Ma Y (2018) Computer-aided engineering analysis for early-ejected plastic part dimension
prediction and quality assurance. Int J Adv Manuf Technol 98(9–12):2389–2399