Analysis of feature extraction algorithms for quality prediction using machine learning in injection molding

When using machine learning for quality prediction in injection molding, feature processing is an important step in data preparation to improve the quality of model prediction. The objective of this study was to evaluate the prediction performance of different feature extraction algorithms compared to more common feature selection. Two test specimens, each with two recorded quality features, were produced in six different injection molding process states, resulting in 11.720 injection cycles as the data base. Depending on the process state, R² of up to 0.99 could be achieved. Nevertheless, the results show that feature selection is preferable for feature processing.


Citation style:
Could not load citation form.


License Holder:

© 2022 The Author(s)