Micro-injection moulding (mu IM) is a special injection moulding process to obtain plastic parts with at least one micrometrical feature. Direct quality control in mu IM is particularly complicated given the strongly nonlinear and time-varying relationship between the process parameters set on the machine, the processing variables, and the need for a complete raw material characterization. In this study, a machine learning model based on artificial neural networks (ANNs) was developed with the aim of predicting a selected quality indicator, namely the cavity filling length. To this purpose, data were acquired during spiral flow tests, generally adopted for assessing the material mouldability, performed on a well-characterized polypropylene in different moulding conditions. During the process, the evolution of hydraulic pressure and piston position were acquired adopting the sensors already available to the machine. The part length was measured at the end of each moulding test. Moreover, an undocumented maintenance activity was carried out before completing all the runs and has been considered to test the predictivity of the artificial intelligence model in case a restoring or repairing activity should not bring the system back to its initial working conditions. After training, validating, and testing the ANN on the collected experimental data, the model showed a coefficient of determination R2 of 88.6% on cavity filling prediction in conditions of mould temperature and injection pressure not previously tested, and an R2 of 83% on the prediction of the length of the successive piece. Moreover, it has been successfully validated also in intermediate moulding conditions between those used during the training phase.

Artificial neural network-based flow length prediction system under different machine settings for micro-injection moulding

Liparoti S.;Di Pasquale V.
;
Volpe V.;Adinolfi F.;Pantani R.
2024-01-01

Abstract

Micro-injection moulding (mu IM) is a special injection moulding process to obtain plastic parts with at least one micrometrical feature. Direct quality control in mu IM is particularly complicated given the strongly nonlinear and time-varying relationship between the process parameters set on the machine, the processing variables, and the need for a complete raw material characterization. In this study, a machine learning model based on artificial neural networks (ANNs) was developed with the aim of predicting a selected quality indicator, namely the cavity filling length. To this purpose, data were acquired during spiral flow tests, generally adopted for assessing the material mouldability, performed on a well-characterized polypropylene in different moulding conditions. During the process, the evolution of hydraulic pressure and piston position were acquired adopting the sensors already available to the machine. The part length was measured at the end of each moulding test. Moreover, an undocumented maintenance activity was carried out before completing all the runs and has been considered to test the predictivity of the artificial intelligence model in case a restoring or repairing activity should not bring the system back to its initial working conditions. After training, validating, and testing the ANN on the collected experimental data, the model showed a coefficient of determination R2 of 88.6% on cavity filling prediction in conditions of mould temperature and injection pressure not previously tested, and an R2 of 83% on the prediction of the length of the successive piece. Moreover, it has been successfully validated also in intermediate moulding conditions between those used during the training phase.
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11386/4894455
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo

Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 0
  • ???jsp.display-item.citation.isi??? 0
social impact