Open Access3D Part Machining Time Prediction with Parameter Extraction, Deep Learning and 3D Data Augmentation
Author(s) -
Umut Nazmi Aktan,
Mehmet Dikmen
Publication year - 2025
Publication title -
ieee access
Language(s) - English
Resource type - Magazines
SCImago Journal Rank - 0.587
H-Index - 127
eISSN - 2169-3536
DOI - 10.1109/access.2025.3572621
Subject(s) - aerospace , bioengineering , communication, networking and broadcast technologies , components, circuits, devices and systems , computing and processing , engineered materials, dielectrics and plasmas , engineering profession , fields, waves and electromagnetics , general topics for engineers , geoscience , nuclear engineering , photonics and electrooptics , power, energy and industry applications , robotics and control systems , signal processing and analysis , transportation
In the aviation industry, an aircraft consists of thousands of detail parts in different types of materials. The sizes of these parts can range from centimeters to meters, and their complexity can also vary greatly. Therefore, there is a need to produce detail parts in a wide variety of geometries that constitute the product. In addition, these detail parts, especially in the prototype phase, often require a workshop-type of production rather than mass production. To optimize the production process and its cost, the production times of these parts must be estimated in advance. There is currently a great need for expert effort on this matter. Experts examine each detail part and calculate their predictions. In this study, multiple innovations are proposed to automate these time-consuming expert examinations by using special software. To that end, a regression model that can predict machining times using 3D models of detail parts used in the aerospace industry is proposed. On the other hand, the relatively small amount of datasets created by the detail parts used in aircraft also poses an issue in the deep learning regression problem. To solve this problem, a special data augmentation and a series of parameter extraction and concatenation methods are introduced in this study. The solutions developed with the proposed procedure have significantly contributed to the prediction performance. In experiments after adapting these methods to regression, the prediction error is significantly reduced.