A Deep Learning Solution for Quality Control in a Die Casting
Process
Paula Mielgo
1
, Anibal Bregon
2
, Carlos J. Alonso-Gonz
´
alez
3
, Daniel L
´
opez
4
, Miguel A. Mart
´
ınez-Prieto
5
and Belarmino
Pulido
6
1,2,3,5,6
Department of Computer Science, University of Valladolid, Valladolid, Spain
paula.mielgo@uva.es, anibal.bregon@uva.es, calonso@uva.es, miguelamp@uva.es, b.pulido@uva.es
4
Factor
´
ıa de Motores, HORSE Spain, Valladolid, Spain
daniel.g.lopez@horse.tech
ABSTRACT
Industry 4.0 aims for a digital transformation of manufac-
turing and production systems, producing what is known
as smart factories, where information coming from Cyber-
Physical Systems (core elements in Industry 4.0) will be
used in all the manufacturing stages to improve productivity.
Cyber-physical systems through their control and sensor sys-
tems, provide a global view of the process, and generate large
amounts of data that can be used for instance to produce data-
driven models of the processes. However, having data is not
enough, we must be able to store, visualize and analyze them,
and to integrate induced knowledge in the whole production
process. In this work, we present a solution to automate the
quality control process of manufactured parts through image
analysis. In particular, we present a Deep Learning solution
to detect defects in manufactured parts from thermographic
images of a die casting machine at an aluminum foundry.
1. INTRODUCTION
In 2015, the foundational definition and main design princi-
ples of Industry 4.0 were presented by Hermann, Pentek, and
Otto (Hermann, Pentek, & Otto, 2016) as a guide for imple-
menting Industry 4.0. This definition encompasses the four
key components of Industry 4.0: Cyber-Physical Systems
(CPS), the Internet of Things (IoT), the Internet of Services
(IoS), and Smart Factories. The objective of Industry 4.0 is
the digital transformation of manufacturing and production
industries. Cyber-physical systems represent the foundation
of Industry 4.0, which frequently involve control systems,
embedded software, and a substantial array of data coming
from sensors and actuators. These systems generate a vast
Paula Mielgo et al. This is an open-access article distributed under the terms
of the Creative Commons Attribution 3.0 United States License, which per-
mits unrestricted use, distribution, and reproduction in any medium, provided
the original author and source are credited.
quantity of data, which must be integrated and analysed in
order to achieve the designation of “smart factories”.
The concept of Smart Manufacturing was introduced in the
United States to facilitate the deployment of emerging tech-
nologies in manufacturing, including the Industrial Internet
of Things (IIoT) and Artificial Intelligence (AI). Smart man-
ufacturing, also known as intelligent manufacturing, focus
on the adoption of these advanced information and manufac-
turing technologies to optimize the production (Zhong, Xu,
Klotz, & Newman, 2017). The main focus of this methodol-
ogy is to enhance the quality, traceability, and efficiency of
the production process. Each industrial revolution has been
accompanied by an increase in productivity, which has been
attributed to the introduction of new technologies, including
the steam engine, electricity, and digital technology. For the
fourth industrial revolution, the primary factor driving pro-
ductivity enhancement is the far-reaching impact of these vast
quantities of data, which influence not only production but
also other sectors, particularly engineering processes. This
allows for more effective decision-making processes. How-
ever, having data is not enough; it must be stored, visualised
and analysed, and the resulting knowledge must be integrated
into the entire production process. This can be achieved,
for instance, by producing data-driven models that can sub-
sequently be employed in a digital twin (which represents
another crucial component in the smart factory framework).
This is of particular importance in those smart factories where
there are few, if any, analytical models available, due to the
nature or complexity of the processes.
Artificial intelligence in general and Machine Learning (ML)
in particular play a pivotal role in this contemporary develop-
ment of smart manufacturing characterised by the production
of data-driven models. The integration of ML with the pro-
duction process facilitates the reduction of production time,
improvement of quality and the elimination of unnecessary
1
