Sarikaya, Erkut ; Elling, Magnus von ; Lu, Xu ; Weigold, Matthias (2023)
Continual learning based machining simulation for the prediction of NC signals.
In: Procedia CIRP, 120
doi: 10.1016/j.procir.2023.09.094
Artikel, Bibliographie
Kurzbeschreibung (Abstract)
This article presents a continual learning method for the prediction of NC signals of a machine tool. The regression model has been accomplished with a long short-term memory recurrent neural network consisting of a dynamic multi-head architecture to satisfy tool-specific learning. Additionally, a regularization based method has been used. The results demonstrate that catastrophic forgetting could be significantly reduced by applying the proposed continual learning method. In an experimental validation the model shows good results for the prediction of the spindle current despite high process diversity in a real production environment.
| Typ des Eintrags: | Artikel |
|---|---|
| Erschienen: | 2023 |
| Autor(en): | Sarikaya, Erkut ; Elling, Magnus von ; Lu, Xu ; Weigold, Matthias |
| Art des Eintrags: | Bibliographie |
| Titel: | Continual learning based machining simulation for the prediction of NC signals |
| Sprache: | Englisch |
| Publikationsjahr: | 2023 |
| Verlag: | Elsevier B.V. |
| Titel der Zeitschrift, Zeitung oder Schriftenreihe: | Procedia CIRP |
| Jahrgang/Volume einer Zeitschrift: | 120 |
| DOI: | 10.1016/j.procir.2023.09.094 |
| URL / URN: | https://www.sciencedirect.com/science/article/pii/S221282712... |
| Kurzbeschreibung (Abstract): | This article presents a continual learning method for the prediction of NC signals of a machine tool. The regression model has been accomplished with a long short-term memory recurrent neural network consisting of a dynamic multi-head architecture to satisfy tool-specific learning. Additionally, a regularization based method has been used. The results demonstrate that catastrophic forgetting could be significantly reduced by applying the proposed continual learning method. In an experimental validation the model shows good results for the prediction of the spindle current despite high process diversity in a real production environment. |
| Freie Schlagworte: | catastrophic forgetting, machine learning, machine tool, material removal, milling |
| Fachbereich(e)/-gebiet(e): | 16 Fachbereich Maschinenbau 16 Fachbereich Maschinenbau > Institut für Produktionsmanagement und Werkzeugmaschinen (PTW) 16 Fachbereich Maschinenbau > Institut für Produktionsmanagement und Werkzeugmaschinen (PTW) > TEC Fertigungstechnologie |
| Hinterlegungsdatum: | 15 Jan 2024 09:46 |
| Letzte Änderung: | 17 Jan 2024 11:12 |
| PPN: | 514763426 |
| Export: | |
| Suche nach Titel in: | TUfind oder in Google |
 |
Frage zum Eintrag |
Optionen (nur für Redakteure)
 |
Redaktionelle Details anzeigen |
Drucken
Impressum