Machine learning has become a core part of smart factories and Industry 4.0. In our work, we extend the use of machine learning for quality prediction of a thin glass product formed using a Non-isothermal Glass Moulding (NGM) process. As the form shape of a glass lens requires multiple variables to describe, Multi-Target Regression (MTR) is suitable for the same. Many MTR models are able to provide intuitive insights into the prediction target(s). We present a data pipeline that employs bootstrapping-inspired sampling for robust feature selection, modelling and validation for small dataset. The results demonstrate how MTR models can be used for prediction with dataset with high dimensional time series input and multiple targets.
The steadily growing thin glass market is driven by a vast amount of applications among which automobile interiors and consumer electronics are, such as 3D glass covers for displays, center consoles, speakers, etc. or as part of optics within head up-displays. Today, glass manufacturers are suffering from challenges brought about by the increases of shape complexity, accuracy and product variants while simultaneously reducing costs. The direct manufacturing method via grinding and polishing is no longer suitable because of its limited machinability for thin glasses in respect to fracture and its cost insufficiency due to the length of the process chain. Instead, replication-based technologies or thin glass forming become promising manufacturing methods to overcome the aforementioned technical and economic challenges. For instance, thermal slumping is only able to satisfy the most basic requirements and is in particular limited regarding the deformation degree and shape complexity of thin glass products. Technologies such as vacuum-assisted slumping or deep drawing are currently in development at the Fraunhofer Institute for Production Technology IPT and promise additional cost benefits. This paper introduces all potential process variants for thin glass forming. The suitability of different methods for process development, specifically process modeling based on either experimental-, simulation- or machine learning approaches (white box and black box models), will be addressed and discussed. Furthermore, process efficiency is examined on both an economic and technical level, where molding time, suitable geometries and accuracy are the focus. The methodologies presented in this paper aim at developing a guideline for glass manufacturers on determining the optimal strategy for the process development of thin glass production.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.