Data-driven deep learning approach for suggesting process parameters for the milling operations

This work introduces an approach that leverages neural networks to classify 3D models based on volumetric boundaries and systematically organizes machining knowledge into a library of operations to suggest suitable process parameters. The primary objective of this approach is to preserve valuable in...

Ausführliche Beschreibung

Gespeichert in:

Bibliographische Detailangaben
Veröffentlicht in:	Procedia CIRP Jg. 130; S. 307 - 312
Hauptverfasser:	Asghar, Eram, Ratti, Andrea, Tolio, Tullio
Format:	Journal Article
Sprache:	Englisch
Veröffentlicht:	Elsevier B.V 2024
Schlagworte:	3D convolutional neural networks class-specific autoencoders Manufacturing knowledge reuse removed volume geometry of tool passes 3D convolutional neural networks removed volume geometry of tool passes Manufacturing knowledge reuse class-specific autoencoders
ISSN:	2212-8271, 2212-8271
Online-Zugang:	Volltext
Tags:	Tag hinzufügen Keine Tags, Fügen Sie den ersten Tag hinzu!

Beschreibung
Zusammenfassung:	This work introduces an approach that leverages neural networks to classify 3D models based on volumetric boundaries and systematically organizes machining knowledge into a library of operations to suggest suitable process parameters. The primary objective of this approach is to preserve valuable insights from past machining experiences, and digitally organize and analyze them to improve the efficiency of forthcoming machining processes. This leads to better decision-making in process planning for the new machining components. The methodology revolves around the extraction of geometric and operational parameters from industrial part programs for each tool movement. Formalizing the available process information and creating a knowledge database for all the tool passes in the operations library which contains the relevant information for each tool pass, provides intricate details of the machining process. The proposed approach utilizes the available machining knowledge while producing new and existing parts. Utilizing 3D convolutional neural networks, the approach classifies tool pass geometries within a dataset based on the cutting tool used. This allows further identification and recognition of the geometric and operational similarity amongst the classified volumes through autoencoders. This process leads to the development of a repository of operations that captures the essential process design knowledge, thus fostering the reuse of available machining parameters. The proposed approach demonstrates its effectiveness by implementing it on actual machining data and creating a machining database that assists in making propositions related to operational parameters for similar geometric features during the process planning phase.
ISSN:	2212-8271 2212-8271
DOI:	10.1016/j.procir.2024.10.092