International Journal of Information Management Data Insights 2 (2022) 100107

Contents lists available at ScienceDirect

International Journal of Information Management Data

Insights
journal homepage: www.elsevier.com/locate/jjimei

Deep learning for manufacturing sustainability: Models, applications in

Industry 4.0 and implications
Anbesh Jamwal a, Rajeev Agrawal a,∗, Monica Sharma a,b
a
Department of Mechanical Engineering, Malaviya National Institute of Technology, Jaipur, 302017, India
b
Department of Management Studies, Malaviya National Institute of Technology, Jaipur, 302017, India

a r t i c l e i n f o a b s t r a c t

Keywords: Recent advancements and developments in artificial intelligence (AI) based approaches have shifted the man-
Artificial intelligence ufacturing practices towards the fourth industrial revolution, considered as Industry 4.0 practices. A positive
Deep learning impact of AI-based techniques on sustainability can be seen in manufacturing organisations’ at the system, prod-
Industry 4.0
uct and process levels. Adopting AI-based strategies in manufacturing improves decision making, productivity
Sustainability
and system performance. Despite sustainability and other benefits, the adoption of AI-based approaches in manu-
Sustainable manufacturing
Fault diagnosis facturing organisations is still limited due to employees’ knowledge and digital skills. In the present time, due to
Predictive maintenance the digitalisation of manufacturing activities, intelligent sensors, and supply chain activities, industries are facing
Quality management challenges with the generation of high volume, different variety and velocity of data. This data can be helpful for
manufacturing organisations to enhance their performance and sustainability. However, managing this big data
is still a significant challenge due to a lack of knowledge and limited literature. Deep learning (DL) based models
can be a suitable choice to provide advanced analytics tools for manufacturing data processing and analysis.
However, literature on the DL is still limited in the manufacturing context with its relationship to sustainability.
The present study discusses the evolution of DL approaches in manufacturing and different DL-based models. This
study also highlights how DL-based approaches can enhance the sustainability performance of industries. In the
study, primary research areas, i.e., fault diagnosis, quality management, and predictive maintenance, have been
discussed. Finally, a conceptual DL-based framework is proposed for the manufacturing industries to enhance
their sustainability performance in manufacturing activities.

1. Introduction The role of DL in manufacturing organisations is vital to collect,

analyse and interpret a large amount of data (Akter et al., 2021;
Artificial intelligence (AI) based algorithms are transforming the Ashok et al., 2022; Lee et al., 2020; Zhang et al., 2021). Unlike tradi-
service and manufacturing organisations into high-efficiency indus- tional ML approaches, DL is based on predictive analytics rather than
tries by improving productivity and reducing maintenance costs and linear.
production defects (Chien et al., 2020; Li et al., 2017). The adoption The past studies conducted on Industry 4.0 shows that 82% of manu-
of machine learning (ML) and deep learning (DL) has significantly facturing and service organisations have increased their efficiency by us-
increased in the last few years in the manufacturing sector due to the ing Industry 4.0 technologies, i.e., AI-based approaches and the Internet
generation of a large amount of data (Lee et al., 2018; Rawat et al., of things (IoT) (Akter et al., 2021). In the last few years, both developed
2021). Shop floor, supply chain activities and manufacturing processes and developing countries started the development of roadmaps for the
generate a variety of data in large volumes, which is not effectively Industry 4.0 transition (Mittal et al., 2018; Yadav et al., 2020). Indus-
used by industries (Yao et al., 2017). Extracting relevant information tries are now more focused on infrastructure development for adopting
is crucial and significant, which DL algorithms can do effectively. DL data science and IoT-enabled manufacturing practices (de Sousa Jab-
approaches are based on AI and ML, which imitates how humans gain bour et al., 2018; Kamble et al., 2020; Khan & Javaid, 2021). For ex-
a specific type of knowledge (Wang et al., 2018). DL approaches can ample, the German Federal Government introduced “Industry 4.0” as
be considered critical in data science, including predictive modelling the high technology plan for their industries in 2010 as a collaborative
and statistics (Wuest et al., 2016). Information management with effort of countries from the European Union. In 2011, the United States
AI and ML is an emerging research area in present time in various introduced the concept of “smart manufacturing” under the “Smart Man-
domains such as: manufacturing, management and social sciences. ufacturing Leadership Coalition”.

∗
Corresponding author.
E-mail address: [email protected] (R. Agrawal).

https://doi.org/10.1016/j.jjimei.2022.100107
Received 4 April 2022; Received in revised form 4 August 2022; Accepted 6 August 2022
2667-0968/© 2022 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/)
A. Jamwal, R. Agrawal and M. Sharma International Journal of Information Management Data Insights 2 (2022) 100107

Fig. 1. Role of data-driven approaches for sus-

tainable manufacturing (Wang et al., 2018).

Similarly, China has introduced their manufacturing strategy, which represents how data-driven strategies can be used in the manu-
“China 2025”, which is focused on promoting advanced manufactur- facturing practices in organisations (Wang et al., 2018). In this, it can
ing (Machado et al., 2020). The dynamic changes in the market drive be seen that computational approaches based on AI and DL algorithms
these plans and industrial revolutions. Currently, manufacturing indus- act as drivers for intelligent manufacturing or smart production in the
tries are equipped with the Internet and sensors to collect and transmit Industry 4.0 context, which have benefits in terms of quality, cycle time
data from machines to servers (Chauhan et al., 2021). This helps reduce reduction, and waste management (Chui et al., 2018). Generally, all the
the industry’s downtime and improve the automation level of manufac- data-driven approaches model the complex relationship between the in-
turing organisations (Sharma et al., 2020). put data provided by the system. A depth understanding of a physical
Sensors are used to collect the data, and advanced computational al- system is not required for the analysis.
gorithms based on AI, ML and DL are used for real-time process monitor- In recent years, many efforts have been made on AI-based tools
ing (Syafrudin et al., 2018). Industry 4.0 technologies positively impact for manufacturing applications (Sahu et al., 2021). However, now re-
product, process and system-level sustainability (Enyoghasi & Badur- search is more focused on the sustainability aspects of manufacturing
deen, 2021). Few researchers have highlighted “smart manufacturing” through the use of advanced computational algorithms (Bag et al., 2021;
as the new trend in manufacturing systems in which machines within Liu et al., 2020). (Cioffi et al., 2020) highlighted the role of cutting-edge
a system are connected through a wireless network (Wang et al., 2018; computational approaches, i.e., ML and AI-based algorithms, for sustain-
Yao et al., 2017). These machines are continuously monitored by in- able production. (Jamwal et al., 2021b) discussed the limitations and
telligent sensors, which capture data and share it to the cloud, where advantages of various ML approaches for sustainable manufacturing.
analysed by intelligent computational algorithms. The data collected Recent studies on AI and ML show that manufacturers are now more fo-
from the different activities within the shop floor is used to increase cused on adopting advanced computational algorithms for manufactur-
sustainability, system productivity, maintenance operations and prod- ing operations (Alexopoulos et al., 2020; Chien et al., 2020). However,
uct quality (Akdil et al., 2017; Dikhanbayeva et al., 2020). In past years, few studies have highlighted the importance of identifying challenges
many new technologies have been developed, which has boosted the related to data-driven approaches for decision-making in manufacturing
fourth industrial revolution wave (Carvalho et al., 2018). As discussed organisations (Cohen & Macek, 2021; Niri et al., 2022). In the manufac-
by (Sung, 2018), many technologies related to Industry 4.0 came into ex- turing sector, few studies have studied the limitation and advantages of
istence before the evolution of the term “Industry 4.0”. So, in Korea, the ML and AI-based approaches to manufacturing practices (Carvalho et al.,
term “Fourth industrial revolution” is more popular than the “Industry 2019). Compared to AI techniques, DL-based algorithms perform bet-
4.0”. The developments in AI, DL, blockchain technology (BT), cyber- ter in manufacturing-based applications such as condition monitoring,
physical systems (CPS), decentralised manufacturing systems (DMS), predictive maintenance and real-time process monitoring (Yang et al.,
cloud computing (CC) and the Internet of Things (IoT) are support- 2020). Generally, DL-based approaches allow automatic data process-
ing the innovations in manufacturing organisations (Bag et al., 2021; ing, which is collected through the sensors connected to machines and
Javaid et al., 2021; Lee et al., 2018). Now manufacturers can collect, transmitted to the cloud by the Internet (LeCun et al., 2015). In Industry
analyse and process the product life cycle data, ranging from raw mate- 4.0, DL offers a wide range of economic, social and environmental bene-
rial extraction to the disposal phase of the product (Ferrari et al., 2021). fits for manufacturing organisations, which helps achieve sustainability
ML and AI-based computational algorithms can help process and an- (Bai et al., 2020). AI and DL based approaches are providing solutions
alyze the large volume of data sets and provide critical insights to the for many applications in the industries. Despite of AI and DL benefits
manufacturers from these data sets (Verma et al., 2021; Yang et al., to industries the adoption level is limited in many industry sector due
2020). The valuable information collected from these datasets can be to lack of knowledge about and understanding about the benefits of AI
used for predictive maintenance, demand forecasting and condition and DL in manufacturing. Apart from this, there is limited literature
monitoring (Arinez et al., 2020). However, few AI-based algorithms can available on DL approaches and its benefits to achieve sustainability in
be used for generative design purposes (Wan et al., 2020). The role of manufacturing practices. To fill these research gaps, the main research
AI-based approaches in sustainable production is illustrated in Fig. 1, questions of this study are as follows:

2
A. Jamwal, R. Agrawal and M. Sharma International Journal of Information Management Data Insights 2 (2022) 100107

RQ1: What are the different deep learning models and their applica- (LeCun et al., 2015). In 1969, neural networks entered the dark phase
tions in manufacturing to achieve sustainability? of research and development when MIT professors argued that they
RQ2: What are the main research themes in deep learning for man- could not learn simple XOR functions (Zhao et al., 2019). The other MIT
ufacturing? professors’ findings dampened the research motivation for deep neural
RQ3: What are the main challenges for manufacturing organisations networks (DNN). During the analysis, it is found that the universal ap-
to adopt deep learning models? proximation theorem can solve any continuous problem using a single
hidden layer (Shrestha & Mahmood, 2019). The mathematical valida-
In the light of literature gaps related to DL methods, this study aims
tion of the universal approximation theorem has put a question on the
to discuss DL methods and applications in Industry 4.0 for manufactur-
validation of the DNN algorithm. DNN has a feature extraction bene-
ing and its relationship with sustainability. Based on the literature anal-
fit which differentiates DNN from other ML techniques (Miotto et al.,
ysis, we have provided an integrated DL-based sustainable production
2018). Generally, DNN is also a type of NN having a multilayer percep-
framework for manufacturing organisations to achieve sustainability. In
tron. DNN algorithms are trained to learn and analyse the representa-
this study, we have discussed the evolution of data-driven approaches
tions from more extensive data sets (Voulodimos et al., 2018). DL-based
and their applications to manufacturing organisations. Various DL mod-
approaches found many applications in the manufacturing sector, as
els have also been discussed with their advantages and limitations in
it doesn’t require manual design for information extraction from data
manufacturing.
sets (Wang et al., 2018). The advancement in DL can solve non-linear
The rest of the paper has eight sections. Section 2 discusses about
functions and more complex manufacturing problems efficiently. These
the overview and history of deep learning; Section 3 discusses about the
advancements in DL can be implemented by cheaper processing units
basic structures in deep learning; Section 4 discusses about the appli-
and big data for model training and representation (Shrestha & Mah-
cations of deep learning to sustainability; Section 5 discusses the chal-
mood, 2019). (Vinuesa et al., 2020) argued that AI-based approaches
lenges of deep learning adoption; Section 6 discusses the deep learning-
will continue to impact sustainability in healthcare, finance and manu-
based framework for sustainable production; Section 7 discuss the im-
facturing. Further, Fig. 2 shows the evolution of different AI-based mod-
plications and Section 8 discuss the conclusion and limitation of the
els with their timeline, which can be categorised into four main periods
study.
(1) Infacy period,(2) First upsurge period,(3) Second upsurge period and
2. Literature Review (4) Third boom period.

Due to evolution of Industry 4.0 practices, generation of data from 2.2. Comparison between traditional approaches of machine learning and
manufacturing and production activities has been increased which is deep learning approaches
challenging for organizations when finding an optimal solution of prob-
lem (Kotsiantis et al., 2006; Votto et al., 2021). These issues can be Traditional approaches of machine learning and deep learning algo-
addressed by application of bio-inspired algorithms which can provide rithms are trendy in Industry 4.0 for manufacturing organisations as it
solution based on incomplete information and limited computation fa- helps to increase efficiency and productivity through predictive main-
cility (Chakraborty & Kar, 2017; Kar, 2016; Kushwaha et al., 2021). tenance and condition monitoring (Mao et al., 2019). These approaches
Machine learning which is now presented as the sub-set of artificial in- are part of data-driven tools used to model the complex relationship
telligence can be referred as the intelligence. In the history of manufac- between any process’s input and output variables (Yao et al., 2017).
turing activities various intelligence approaches have been used which is Both ML and DL are critical enabling technologies for Industry 4.0, of-
time consuming and requires higher costs (Carvalho et al., 2019). In last ten overlapping in the literature (Davis et al., 2015). Few studies have
few years, machine learning tools have been used which have numerous investigated and highlighted the differences between both the terms ML
applications in manufacturing sector such as: quality control, optimiza- and DL with their similarities (Shinde & Shah, 2018). DL approaches
tion and forecasting. As a part of artificial intelligence machine learning are the advanced form of ML algorithms used when data is colossal and
approaches can learn and adapt any new changes in the production en- unstructured. Thus, DL approaches can solve more complex problems
vironment (Loyer et al., 2016). Therefore, machine learning approaches than ML (Salloum et al., 2020). DL approaches have a higher hierarchi-
are the area of interest why these are important in manufacturing sec- cal structure than ML and distinctive attributes in model development,
tor in many applications. Generally, machine learning algorithms are training and feature learning (Chauhan & Singh, 2018). DL techniques
designed in such a way that they can derive the required knowledge integrate both the model development and feature learning in one phase
from the large data sets (Herath & Mittal, 2022; Priore et al., 2001). by selecting different kernels or tuning the input variables or parameters
However, after evolution of Industry 4.0 practices manufacturing with the help of end-to-end optimisation (Nosratabadi et al., 2020).
data has moved into the era of “Big Data” in which machine learning ap-
proaches have been evolved into deep learning techniques (Cioffi et al.,
3. Basic structures in deep learning
2020). These techniques can handle large amount of data and more pow-
erful than traditional machine learning approaches with wide range of
3.1. Auto encoders (AE)
application areas such as: condition monitoring, predictive maintenance
and accurate forecasting (Wang et al., 2018).
AE are the types of ANN used to learn the data encoding in an unsu-
2.1. Evolution of deep learning pervised manner. The main aim of the AE is to understand the encod-
ing (lower-dimensional representation) for the higher-dimensional data
Deep learning approaches are the subset of ML-based on neural net- (Khan & Yairi, 2018). Generally, AE is a neural network with unsuper-
work layers. These layers are used to learn the representation of a large vised algorithms representing and learning the input data. Based on the
volume of data with multiple levels of abstraction (Guo et al., 2016). input data set, it reduces the data dimensionality helps and recreate the
In the past few years, DL methods have improved state of art in ob- original data set (Shrestha & Mahmood, 2019). Figure 3 (A) represents
ject detection, speech recognition and manufacturing decision-making the autoencoder nodes.
(Li et al., 2020). DL-based network drives many applications and ser-
vices, including information, analytics without human intervention and 3.2. Restricted Boltzmann machine (RBM)
predictive maintenance (Nikolakis et al., 2020).
Frank Rosenblatt first introduced neural networks in 1957, con- RBM is a type of NN in which an unsupervised learning algorithm
sisting of two processing layers for recognising simple patterns can be applied (See Figure: 3[B]). This DL approach can build non-linear

3
A. Jamwal, R. Agrawal and M. Sharma International Journal of Information Management Data Insights 2 (2022) 100107

Fig. 2. Different AI models with their timeline.

Fig. 3. (A) Autoencoder Nodes (Khan &

Yairi, 2018) (B) Restricted Boltzmann machine
(Zhao et al., 2019).

generative models from a more extensive data set consisting of unla- of manufacturing activities. Today, many DL-based frameworks can pro-
beled data (Zhao et al., 2019). The main principle of RBM is to train vide a better level of abstraction with the programming solution and
the network to increase log or product functions. RBM algorithms can simplify complex problems (Hansen & Bøgh, 2021). In this study, we
learn the probability distribution of a function over the different inputs have discussed the eight most popular frameworks for manufacturing
(Ardabili et al., 2019). RBM consists of a two-layer network in which activities. Each framework has its limitation and purpose in manufac-
each visible layer of the network is connected to the hidden layer. There turing. Further, Table 1 shows the different frameworks and their de-
is no connection between the different units in the same layers. scription.

3.3. Long Short term memory (LSTM) 4. Applications of deep learning to achieve sustainability in
manufacturing
LSTM was introduced by (Hochreiter & Schmidhuber, 1997) as the
implementation of RNN 1997. Unlike other feed-forward architectures, AI-based tools, i.e., DL and ML algorithms, are the ally that sus-
LSTM can capture the knowledge from previous stages of data and train tainable development needs to advise, execute and design the future
the model for the manufacturing applications in which state awareness of upcoming generations (Vinuesa et al., 2020). These days, advanced
or memory is required. LSTM was introduced to address the limitation data analytic tools based on AI help to use the resource in the in-
of DNN (LeCun et al., 2015). The architecture of LSTM is presented in dustry more efficiently, waste reduction and reduce carbon emissions
Fig. 4. (Carvalho et al., 2019). AI is contributing to sustainability in the sup-
AI and DL are no more mere buzzwords in manufacturing organi- ply chain also. Applying AI tools to supply chain practices helps to re-
sations (Li et al., 2017). In the last few years, developing and devel- duce carbon emissions and supply chain optimisation (Belhadi et al.,
oped nations have realised the need to adopt AI-based approaches in 2021). The optimal supply chain design with AI-based tools minimises
manufacturing activities (Yadav et al., 2020). Using these approaches in mobility and the negative impact of the supply chain on the environ-
business activities has brought a significant shift in how manufacturing ment (Abdirad & Krishnan, 2020). UN 2030 sustainability agenda has
and service-based organisations invest in AI-based methods (Sahu et al., motivated the manufacturing organisation to adopt renewable energy
2021). Implementation of DL approaches is quite challenging in the case sources as the primary energy source. Renewable energy sources have

4
A. Jamwal, R. Agrawal and M. Sharma International Journal of Information Management Data Insights 2 (2022) 100107

Fig. 4. LSTM architecture (LeCun et al., 2015).

Table 1
Popular Deep learning frameworks and their benefits for manufacturing organisations.

Deep learning tool Tool type Description Benefits in manufacturing

TensorFlow Open Source TensorFlow is ap popular DL framework which 1. Manufacturing graphs visualisation and queues using TensorBoard
(Chung et al., 2017) supports languages, e.g., R, C++ and Python, to 2. Robust GUI support for various manufacturing applications
create DL models. TensorFlow has natural
language processing, forecasting, text
classification and summarization capabilities.
Torch/Pytorch Open-Source Torch/Pytorch provides Lua-based DL frameworks. 1. It helps to facilitate the exchange the manufacturing data with the
(Shams et al., 2017) Generally, Torch/Pytorch employs CUDA along external libraries
with C or C++ libraries for data processing. 2. Better performance in rapid prototyping
3. It provides a cleaner interface and is easier to use in case of
manufacturing data

Deeplearning4J Commercial Deeplearning4J is a DL library for Java virtual 1. It can process a large amount of data quickly
(Parvat et al., 2017) machines (JVM). However, it supports other JVM 2. It includes both single-thread and multi-thread deep learning
languages such as Kotlin, Clojure and Scala. It frameworks
supports deep networks through RNN, CNN, RBM, 3. It can bring the entire Java ecosystem to execute DL models
and DBM.
Microsoft cognitive Open-Source Microsoft cognitive toolkit is an open-source 1. Results are more accurate
toolkit platform which efficiently uses the CNN and 2. Easy to use as compared to other interfaces
(Chung et al., 2017) training for the text-based data 3. Detailed documentation with command interfaces

Accors.Net Open-source Accors.Net uses C# as a writing language for the 1. Data clustering, classification, data hypothesis and data
(Parvat et al., 2017) deep learning algorithms. It provides a combined distribution, Real-time defect detection
machine learning framework for image and audio
processing.
Keras Open source Keras is a python based framework which supports 1. Data classification, text generation and Speech recognition
(Han et al., 2020) recurrent networks and convolutional networks. 2. It supports multiple deep learning backends.
Unlike other frameworks, Keras provide a
simplistic framework for users for quick
prototyping.
ONYX Open-source ONNX is an open-source deep-learning ecosystem 1. Provides flexibility, interoperability, compatible libraries and
(Han et al., 2020) that Microsoft develops and can switch between runtimes
different platforms. ONNX provides in-built 2. Performance maximisation across the hardware
operator definition and standard data types.
Unlike other platforms, ONNX is more flexible for
DL-based models.

some limitations related to efficiency compared to non-renewable en- agriculture supply chain, which have reported the benefits of using AI-
ergy sources (Hidayatno et al., 2019). AI-based approaches are also based models to enhance sustainability (Collins et al., 2021).
helping manufacturing industries enhance the efficiency of renewable At present, manufacturing activities rely on data-driven technolo-
energy sources (Jena et al., 2020). The main application areas are gies. The advanced data-driven technologies of Industry 4.0 support
energy production, prediction and fault diagnosis of energy plants manufacturing innovations and sustainability in manufacturing organ-
(Carvalho et al., 2019). Beyond the energy sector, AI-based tools have isations (Raj et al., 2020). Computational intelligence can be con-
other applications in industries like food manufacturing, healthcare and sidered an essential factor in achieving manufacturing sustainability

5
A. Jamwal, R. Agrawal and M. Sharma International Journal of Information Management Data Insights 2 (2022) 100107

Fig. 5. AI-based models applications to Indus-

try 4.0 and sustainability (Jamwal et al., 2021).

(Ram & Tyagi, 2020). The manufacturers are now focusing on adopt- include non-destructive testing approaches such as corona detection, oil
ing AI-based tools such as DL and ML with sustainable manufac- analysis, acoustic, sound level measurement, thermal imaging, predic-
turing practices for better decision-making and optimisation (Hansen tive maintenance and operations data in real-time using smart sensors
& Bøgh, 2021). Optimising manufacturing systems can be critical (Butte et al., 2018). Currently, manufacturing organisations are using
in manufacturing organisations due to many industry constraints re- IoT-based and AI-based predictive maintenance services for machines
lated to resources, energy and carbon emission policies (Enyoghasi & and equipment maintenance operations (Zonta et al., 2020). (Chen et al.,
Badurdeen, 2021). Manufacturing organisations have already accepted 2021) proposed an RUL-based model with the geographical informa-
ML approaches and are widely used in the design, product life cy- tion system data for automobile maintenance. The M-LSTM network-
cle management, production activities, manufacturing operations and based deep learning system is designed for predictive maintenance in
sustainability-related aspects (Jung et al., 2021). The typical scenario the study. (Berghout et al., 2021) proposed a deep learning approach
of ML approaches in manufacturing organisations is presented in Fig. 6. for condition monitoring in the navel industry. The study focuses on
However, few studies have highlighted the importance of data min- extracting and learning essential factors and patterns generated from a
ing and analytics in manufacturing in the literature (Collins, 2020). It small amount of data to predict health estimation. It is found that the
has been found that data analysis is a critical issue for manufacturing proposed algorithms are more effective and have a high level of accu-
organisations. Various data in different formats are generated from ma- racy in terms of generalisation and approximation as compared to other
chines that can be used for multiple industry maintenance applications variants of machine learning approaches.
(Kim, 2017). The main application area of AI tools is fault detection, de- (Nguyen & Medjaher, 2019) proposed a dynamic maintenance
cision support system, quality control and maintenance (Shivajee et al., framework for failure prognostics in manufacturing organisations. The
2019). The importance of AI-based tools, i.e., ML and DL applications, study presents an extended short-term memory-based network, which
has been briefly discussed by (Wang et al., 2018). It is found that ML provides the probabilities that a system can fail in different time frames.
approaches can help reduce energy consumption, cost reduction, waste A real-time industry case scenario validates the proposed algorithm and
reduction and remote maintenance, which positively impact the sustain- framework. The proposed algorithm shows better performance than the
ability of manufacturing (Vinuesa et al., 2020). From the social sustain- other algorithms. (Namuduri et al., 2020) stated that predicting fail-
ability perspective, DL helps reduce employee workload by reducing hu- ures in manufacturing organisations using the smart sensors and data
man efforts in various process monitoring operations. Also, few studies collected from these sensors helps to reduce machine downtime and
have reported on the DL applications in real-time warnings, improving maintenance operations. The use of machine learning and deep learn-
workplace operator safety (Yang et al., 2020). The effect of various tech- ing tools is vital to predicting failures based on the available historical
nologies and DL approaches on sustainability is presented in Fig. 5. data. (Butte et al., 2018) stated that predictive maintenance operations
However, in past literature, many deep learning-based models have are essential in the microelectronics industries to reduce the cycle time
been proposed, which have a wide range of applications in manufactur- and cost due to unplanned machine downtime and maintenance. Big
ing and positively impact sustainability performance. Table 2 shows the data analytics can develop a predictive maintenance strategy for man-
various deep learning models’ applications to the manufacturing sector. ufacturing organisations (Kushwaha et al., 2021). In the study, a deep
neural network is proposed, and the results reveal that the proposed ap-
4.1. Research themes in deep learning for manufacturing applications proach is more effective in predictive maintenance than the other ma-
chine learning algorithms.
4.1.1. Predictive maintenance with deep learning
Predictive maintenance uses deep learning tools and predictive ana- 4.1.2. Reliability analysis with deep learning
lytics to estimate the corrective maintenance schedule, which should In manufacturing organisations, reliability can be defined as the
be done before product failure (Nguyen & Medjaher, 2019). Predic- probability of a system performing its functionalities under uncertain-
tive maintenance in manufacturing organisations is essential as it helps ties. Due to the availability of a large volume of manufacturing data is
reduce maintenance costs and machine downtime (Sajid et al., 2021 important to do an accurate reliability analysis of system performance
Wang & Wang, 2017). Predictive maintenance in Industry 4.0 era aims (Chen et al., 2020). However, traditionally analytical approaches such as
to schedule the maintenance of a product or machine at the most ef- first-order and second-order reliability methods have been extensively
ficient and convenient time, which allows the optimise the lifespan of used for the reliability analysis in manufacturing organisations in which
a machine or equipment (Çınar et al., 2020). The architecture of pre- limit state function can be approximated by the reliability estimation or
dictive maintenance includes the data storage and acquisition, condi- Taylor expansion (Rajawat et al., 2022). Generally, the manufacturing-
tion monitoring approaches, data transformation, decision support sys- related issues are non-linear and based on non-linear functions. These
tem, prognostics, asset health evaluation and human interface layer approaches suffer the problem of non-convergence during the sampling
(Carvalho et al., 2019). Generally, predictive maintenance technologies methods, i.e., line sampling, importance sampling and Monte Carlo sim-

6
A. Jamwal, R. Agrawal and M. Sharma International Journal of Information Management Data Insights 2 (2022) 100107

Table 2
Deep learning models for manufacturing applications

S.No. Model Name Training model/Equation Application in Sustainability Reference

manufacturing Impact
√
1. D-CNN 𝜐𝑡 = 𝛽𝑡 𝜐𝑡−1 + (1 − 𝛽1 )𝑔𝑡2 Cost estimation in Economic ( ) (Imoto et al., 2018;
√
𝑚𝑡 = 𝛽1 𝑚𝑡−1 + (1 − 𝛽1 )𝑔𝑡 manufacturing Environment ( ) Ning et al., 2020)
√
𝜃𝑡+1 = 𝜃𝑡 − √ 𝛼 𝑚̂ 𝑡 Social ( )
𝜐̂ 𝑡 +𝜀
∑ √
2. R-CNN 𝐿𝑏𝑜𝑥 (𝑡𝑢 , 𝜐) = 𝑠𝑚𝑜𝑜𝑡 ℎ𝐿1 (𝑡𝑢𝑖 − 𝑉𝑗 ) Spectroscopic Economic ( ) (Bhuvaneswari et al., 2021)
𝑖∈{𝑥,𝑦,𝑤,ℎ}
√
∑ [ ] analysis in Environment ( )
√
𝐿𝑚𝑎𝑠𝑘 = 𝑚12 𝑦𝑖𝑗 .𝑙𝑜𝑔 𝑦̂𝑘𝑖𝑗 + (1 − 𝑦𝑖𝑗 ) log(1 − 𝑦̂𝑘𝑖𝑗 ) manufacturing Social ( )
1≤𝑖,𝑗≤𝑚
𝑒𝑥𝑗 √
3. CNN 𝑆𝑗 (𝑋) = ∑𝑛_𝑐𝑙𝑎𝑠𝑠 𝑥𝑙 Crack detection in Economic ( ) (Alipour et al., 2019)
𝑖=1
𝑒 √
𝑁 manufacturing Environment ( )
∑ √
𝐿(𝑦𝑖 , 𝑦̂𝑖 ) = − 𝑁1 𝑦𝑖 log(𝑦̂𝑖 ) Social ( )
𝑖
𝑛𝑢𝑛 𝑛𝑏𝑘 ( ) √
∑ 1
4. CNN 𝐿𝑖𝑛𝑑𝑖 = 𝑁𝑐 𝑎𝑟
𝑉 {𝑔(P𝑘 ), 𝑆 𝑛 , 𝑆 𝑏 } 1 Additive Economic ( ) (Li et al., 2020)
√
{{𝑘=1 } }𝑛𝑢𝑛 manufacturing Environment ( )
𝑛𝑏𝑘 √
𝑃𝑘 ∈ 𝑥𝑢𝑛𝑖,𝑗 quality assessment Social ( )
𝑗=1 𝑖=1
{ }𝑁𝑖𝑛𝑑𝑖 { }𝑁𝑖𝑛𝑑𝑖
𝑛 𝑏
𝑆 𝑛 = 𝑔(𝑃𝑘,𝑚 ) , 𝑆 𝑏 = 𝑔(𝑃𝑘,𝑚 )
𝑚=1 𝑚=1 √
5. RBM, DBM and AE Bayesian belief network, tanh functions, Deionizing, Fault diagnosis Economic ( ) (J. Wang et al., 2018)
√
contractive and sparse autoencoder stochastic gradient Environment ( )
√
descent Social ( )
√
6. RNN and CNN Point-wise thermal concatenation and thermal imaging Distortion prediction Economic ( ) (Alipour et al., 2019)
√
Environment ( )
√
[ ] Social ( )
∑𝑄 ∑𝑄
𝑞𝑖 − 12 (𝑄2 +𝑄) 𝑁𝑄+𝑄−2 𝑞𝑖 √
7. D-CNN and 𝑂𝑆 = 2 × 2 1 − 𝑖=1
= 𝑖=1
Crystal structure Economic ( ) (Ryan et al., 2018)
1
2
(2𝑁−𝑄+1)𝑄− 12 (𝑄2 +𝑄) 𝑁𝑄−𝑄2 √
Virtual-AE Environment ( )
√
Social ( )
𝑁
∑ √
8. Virtual-AE 𝑆𝑖𝑃 = 1
cos(𝜙𝑎𝑢𝑔 𝑛𝑜𝑛−𝑎𝑢𝑔
𝑖 , 𝜙𝑗 ), 𝑁 = 5 Virtual material Economic ( ) (Bhuvaneswari et al., 2021)
𝑁
𝑗=1
√
synthesis Environment ( )
√
Social ( )
1 ∑𝑁
| 𝐴 𝑖 −𝐹𝑖 | √
9. MFNN 𝑀𝐴𝑃 𝐸 = | 𝐴𝑖 | Mechanical Economic ( ) (Sun et al., 2019)
𝑁
𝑖=1 | | √
properties Environment ( )
√
Social ( )
√
10. CNN, RNN and Stochastic gradient descent, transfer learning and Material degradation Economic ( ) (Alipour et al., 2019)
√
LSTM large-scale visual representation Environment ( )
√
Social ( )

ulation approaches (Bichon et al., 2008). Due to a large amount of data novel modelling approach is proposed to analyse the reliability of man-
employing sampling methods requires high computational resources, ufacturing processes. The proposed method is adequate to evaluate and
which is a significant constraint in the case of SMEs (small and medium predict product reliability for smart manufacturing. (Chen et al., 2021)
enterprises) in developing nations (Mittal et al., 2018). To overcome proposed a deep learning-based system reliability method in stochastic
such issues, many studies have addressed the reliability issues with flow manufacturing. Capability assessment of a manufacturing system
the help of deep learning approaches in literature (Wang et al., 2018). is significant in the manufacturing industries. Industries use automated
(Chen et al., 2020) proposed a deep learning-based framework for re- machines for mass production, resulting in complex network formula-
liability analysis using the distance-based adaptive sampling method. tion for reliability analysis. Due to the present manufacturing scenario,
The Monte Carlo simulation approach in the study is also used with the it is challenging to calculate system reliability in shorter times. The
Gaussian process for the probability analysis in manufacturing organi- study proposes a neural network-based deep learning approach, which
sations. The proposed approach is validated with the three case studies can predict the system reliability with a 0.002 Root mean square error.
to investigate the effectiveness and efficiency of the proposed approach. (Zhang et al., 2020) stated that product reliability can be improved by
(Chen et al., 2020) stated that deep integration of cyber physical-based identifying weak links in the manufacturing and production activities.
manufacturing systems has increased the data generated from the shop The study proposes a Rough set Bayesian network model to improve the
floor activities. The role of data is important for predictive maintenance system reliability. The study results show that the proposed approach is
operations in Industry 4.0. The study proposes a reliability analysis more effective than the three other approaches for system reliability.
model based on the time series. The study uses, TensorFlow enabled
deep learning model for the reliability analysis. The proposed model is 4.1.3. Quality assurance with deep learning
validated and tested in the automobile industry. (Tripathi et al., 2021) Industry 4.0 practices have opened many opportunities for manufac-
stated that data mining approaches are now widely accepted in manu- turing organisations to reduce production losses or errors (Ammar et al.,
facturing industries. In the study, a detailed review of CRISP-data min- 2021; Haleem & Javaid, 2019; Müller, 2019). In manufacturing, indus-
ing approaches is done. A framework is proposed to ensure the reliability tries’ quality can be considered a primary concern which has a signif-
and robustness of data-driven knowledge in manufacturing industries. icant impact on the performance and sustainability of an organisation
(Zhang et al., 2021) stated that reliability can be considered the central (Horváthová et al., 2019). Industry 4.0 tools such as deep learning have
aspect in manufacturing organisations, which helps reduce maintenance played an important role in quality assurance in the past few years.
costs and increase efficiency. Due to the wave of Industry 4.0, industries Deep learning approaches based on computer vision can detect the de-
are now adopting reconfigurability in their manufacturing systems. The fect in the product in real-time with minimum error (Li et al., 2020).
study proposes conceptual maps for the reliability analysis of the re- Deep learning approaches enable manufacturers to fix the quality is-
configurable manufacturing systems. (Zhang et al., 2020) stated that sue at the early stages of production, which helps to save rework costs.
work in progress and reliability degradation in manufacturing organisa- Deep learning models represent the data models that use manufacturing-
tions leads to loss of reliability. In the study, a quality variation-centred related data from a large volume of data sets and make predictions

7
A. Jamwal, R. Agrawal and M. Sharma International Journal of Information Management Data Insights 2 (2022) 100107

Fig. 6. Quality management system in Indus-

try 4.0.

(Ozdemir & Koc, 2019). Generally, deep learning methods use neural ment model (Bai et al., 2017). Traditional test automation for quality
network-based approaches for quality assurance (Wang et al., 2018). assurance is inadequate for quality management in the Industry 4.0 era.
Currently, quality management systems (QMS) in Industry 4.0 are dif- Industries are adopting Deep learning-based approaches to make bet-
ferent from the conventional management system, which is presented ter predictions and quality control on larger data sets generated from
in Fig. 6. different manufacturing activities.
(Wang et al., 2018) proposed a cost-efficient and reliable quality
control method for additive manufacturing processes. The proposed AI- 4.1.4. Fault assessment with deep learning
based approach shows an accuracy of 98.9% on the destructive testing Manufacturing organisations can face unpredictive failures in their
results. It is stated that deep convolutional neural networks in qual- manufacturing operations and activities (Azamfar et al., 2020). Abnor-
ity assurance can minimise the quality issues and throughput times. mal operating conditions of machines can cause these failures, degrada-
(Gaikwad et al., 2020) additive manufacturing technologies are ben- tion of manufacturing systems, excessive loads, overheating issues, cor-
eficial to achieving manufacturing sustainability and complex geome- rosive environment and fracture or wear of machine tools (Tang et al.,
try shapes. Machine learning-based assessments in additive manufac- 2019). Unpredictive failures in manufacturing organisations can lead
turing help improve the quality control in the additive manufacturing to loss of operating hours, manufacturing costs, waste generation,
processes. (Zhang & Gao, 2021) stated that Industry 4.0 practices aim higher carbon emissions, downtimes and lower productivity (Arellano-
to improve the reliability and throughput of manufacturing processes. Espitia et al., 2020). However, digital solutions of Industry 4.0 for smart
Deep learning approaches have opened many opportunities to achieve factories can monitor the machine’s condition and shop floor activi-
this goal. In the study, data curation techniques such as imputation, de- ties in real-time, which can identify the defect or failures in real-time
noising, balancing, and outlier detection have been briefly discussed to (Kim, 2017). This helps diagnose the root cause of fault or negligence
improve the effectiveness of smart manufacturing processes. (Bhandari in the early stages so that downtimes can be reduced. In the last few
& Park, 2021) Many manufacturing organisations have introduced the years, the use of smart sensors has increased in manufacturing-related
concept of smart factories, which aims to promote high-efficiency, low- operations (Javaid et al., 2021; Wang & Gao, 2020). These intelligent
cost operations by applying AI and IoT technologies in manufacturing- sensors can catch the real-time data from the manufacturing activi-
related facilities. For surface roughness evaluation, deep learning meth- ties and help investigate the fault assessment or diagnosis (Shao et al.,
ods play an important role. The study adopted a CNN+LSTM neural 2017). In past studies, CNN has been used for defect diagnosis appli-
network-based approach, which reports an accuracy of 85.185%. It is cations such as rotor, gearbox, and bearing (Wang et al., 2018). (He
said that the proposed system is fast, efficient and economical compared & He, 2017) stated that bearing is one of the critical components in
to conventional techniques. (Kim et al., 2022) proposed a CNN-based manufacturing. Effective fault diagnosis of bearing is important to op-
stitch detection method for sewing operations. The proposed approach erate manufacturing activities normally and efficiently. The study pro-
shows an accuracy of 92.3% in defect detection. (Abou Tabl et al., 2021) poses a DL-based approach for fault diagnosis of bearing applications.
stated that manufacturing systems are generating a large amount of data The results show better performances at the lower rotating bearing
due to technological advancements. speeds compared to higher bearing speeds than the previous methods
Data extraction and data processing are significant concerns for man- reported in the literature. (Jia et al., 2018) stated that traditional meth-
ufacturing organisations. In the study, a CNN-based is proposed to pre- ods are less automatic for fault diagnosis than intelligent approaches.
dict the defective items in the production line. The proposed algorithm DL tools can be considered the potential tool for extracting mechanical
shows up to 96.75% accuracy in decision-making. (Jamwal et al., 2021) signals in manufacturing activities. Conventional encoders have prob-
stated that manufacturing organisations are applying ML and AI-based lems with feature extraction, which a local connection network can
in industries to improve product quality, employee safety and efficiency. overcome.
Due to digital transformation, m manufacturing organisations have ex- The proposed approach is validated with the dataset of a gearbox,
perienced the Industry 4.0 benefits such as agility, better customer expe- and the results show better performance than traditional approaches.
rience, profitability, sustainability and efficiency in production systems. (Hoang & Kang, 2019) stated that DL tools have many applications in
Digitalisation in industries runs on the Agile development framework the manufacturing sector for fault diagnosis. The paper uses a systematic
having shorter release cycles, or delivery needs to be done in a shorter literature review approach to find the research progress in fault diagno-
time (Mittal et al., 2018). There is a need to adopt new quality assurance sis of bearing. It is found that RBM, AE, and CNN are primarily used
strategies presented in Fig. 7. tools in fault diagnosis. (Zhao et al., 2016) developed a prognostic and
There are two driving forces for the QMS in the Industries a) Con- fault diagnosis-based system and compared it with traditional models. It
tinuous quality assurance and b) Quick response to market fluctuations. is found that conventional models have limitations related to expression
The quality assurance team needs proper training in the agile develop- capacity, which can be overcome with DL approaches.

8
A. Jamwal, R. Agrawal and M. Sharma International Journal of Information Management Data Insights 2 (2022) 100107

Fig. 7. Requirement for quality assurance

with deep learning.

Fig. 8. Challenges for deep learning adoption.

5. Challenges for manufacturing organisations for adoption of on the literature review, a DL-based conceptual framework is proposed
deep learning for manufacturing organisations to enhance their sustainability perfor-
mance (See Fig. 9). In the proposed framework, there are five main lay-
Deep learning approaches are helping manufacturing organisations ers, i.e., (1) Sustainable, connected processes, (2) IIOT layer, (3) Big
to automate manufacturing processes and increase their efficiency and data layer, (4) Deep learning layer (5) Application layer.
productivity (Wang et al., 2018). Manufacturing organisations are keen In the sustainable, connected processes layer, five principal compo-
on adopting deep learning algorithms and face many challenges. The nents have been considered, i.e., (1) Products, (2) Equipment, (3) Pro-
significant issues are management mindset, business goal alignment cess, (4) People (5) Environment. Few studies have reported that most of
and lack of knowledge regarding the algorithms (Wuest et al., 2016). the emissions dare one in the production stage. In this layer, industries
Further, Fig. 8 shows the main challenges manufacturing organisations need to consider sustainable manufacturing processes. Smart IoT-based
face in adopting deep learning-based approaches in their manufacturing sensors can be used for sustainability metrics. Employee involvement
practices. plays a vital role in sustainability practices. Sustainability awareness
and training programs will help industries to execute DL-based practices
6. Deep learning-based framework for manufacturing efficiently. This layer is connected with the IIoT (Industrial Internet of
sustainability things). Previous studies (Mittal et al., 2018) reported that shop floor
digitalisation is better in a large enterprise compared to SMEs (Small
Recent advancements in AI-based approaches, i.e., ML and DL, have and Medium enterprises). (Jamwal et al., 2021d; Yadav et al., 2020)
shifted manufacturing to a new paradigm considered Industry 4.0. Due reported that MNCs (Multinational Companies) now invest in emerg-
to data-driven processes in Industry 4.0, data generated from the man- ing economies due to cheap labour costs. The SMEs sector is the major
ufacturing and supply chain activities have been increased. However, contributor to the economy of developing nations. SMEs are still strug-
DL approaches are suitable for solving these complex issues in a large gling with shop floor digitalisation. To experience the benefits of DL ap-
volume of datasets. But, in the literature sustainability aspect of DL ap- proaches in manufacturing industries need to make sure about the shop
proaches for manufacturing sustainability is still missing. This study floor digitalisation. In the third layer, data is stored in a larger volume
presents the general structure of DL approaches, applications to man- of datasets called the Big data analytics layer. This layer has three main
ufacturing industries and their relationship with sustainability. Based components, i.e., (1) Process level, (2) Enterprise level and (3) Supply

9
A. Jamwal, R. Agrawal and M. Sharma International Journal of Information Management Data Insights 2 (2022) 100107

Fig. 9. Proposed Deep learning-based framework to enhance manufacturing sustainability.

chain activities. In this layer, data of different activities is stored, which 7.1. Theoretical contributions of the study
is further processed and analysed with DL approaches. DL layer consists
of four main components: The present study discusses the opportunities and research progress
on the deep learning approaches for sustainable production. Moreover,
(1) Descriptive analytics: It helps to analyse what has happened? this article discusses the various models and their applications in sus-
What are the main areas which caused the environmental im- tainable production. The uniqueness of this study is the identification
pacts? Which processes need to be optimised? of different research themes in deep learning for sustainable produc-
(2) Diagnostic analytics: This helps to find the root cause of the prob- tion, i.e., predictive maintenance, quality management and condition
lem. This layer will help to find out what the main reasons for monitoring. Based on the literature review, we have also presented a
the emissions were? Why has the sustainability performance of deep learning-based framework for sustainable products, which indus-
the organisation been reduced? tries can adopt in future to address the various sustainability issues in
(3) Predictive Analytics: This helps predict the effect of parameters Industry 4.0 context.
that need to be optimised to improve the sustainability perfor-
mance of the manufacturing activity. 7.2. Implications for researchers
(4) Prescriptive analytics: This helps determine what action to take
to improve the sustainability performance of manufacturing ac- Based on the literature review and proposed framework following
tivity. What steps can reduce the environmental impacts and en- research implications can be drawn for researchers for future research
hance the organisation’s economic performance? studies:

In the application layer, different applications of DL approaches have 1. The present study discusses the role of deep learning approaches in
been presented, i.e., product life cycle management, shop floor manage- manufacturing. However, this study also discusses the different deep
ment, quality management, fault diagnosis and predictive maintenance. learning models to achieve sustainable production. The literature on
deep learning in manufacturing shows that researchers have paid
7. Research implications less attention in the past in the manufacturing area.
2. The present study’s findings also indicate that there are many op-
The research and reports available in the field of deep learning ap- portunities for sustainable production through the deep learning ap-
proaches for sustainable production support the view that there are a proaches. The adoption of Industry 4.0 practices is helping industries
lot of opportunities for deep learning-based tools in Industry 4.0 to to achieve sustainable production. However, it will be more interest-
support sustainable production. This has motivated many researchers ing to investigate the impact of deep learning approaches on sustain-
to work in the area of deep learning approaches. In the present study, able production for small and medium enterprises. This study pro-
we have conducted a state of the art-based review to address the vari- vides the guidelines for the manufacturing industries on how deep
ous research opportunities in this area. Based on the literature review learning approaches can contribute to sustainable production.
following are some implications for policymakers, practitioners and re- 3. In future studies, energy consumption-related issues, predictive
searchers to drive future research. maintenance and quality management-related matters can be ad-

10
A. Jamwal, R. Agrawal and M. Sharma International Journal of Information Management Data Insights 2 (2022) 100107

dressed for small-scale industries as small-scale industries are the Scholar, ScienceDirect and InderscienceOnline, can be followed in fu-
major contributors to developing nations’ economies. ture studies. In future studies, bibliometric analysis can be done for the
4. There is a lack of empirical investigations in eh area of deep learn- statistical analysis of the documents related to this area. Moreover, in
ing approaches. In future studies, empirical analyses can be done future studies, applications of deep learning approaches in the various
to find the challenges, enabling factors and impact of deep learning industry sectors in the different regions of the world can be explored,
techniques for sustainable production. which will help provide a holistic overview and understanding of the
5. In future studies, the impact of deep learning approaches on the potential of deep learning approaches for sustainable production.
existing production systems can be addressed. How deep learning
approaches can transform the current predictive maintenance and
quality management practices can be addressed in future studies. References

Abdirad, M., & Krishnan, K. (2020). Industry 4.0 in logistics and supply chain manage-
7.3. Managerial contributions to study ment: A systematic literature review. EMJ - Engineering Management Journal, 00(00),
1–15. 10.1080/10429247.2020.1783935.
This review article delivers many important insights related to deep Abou Tabl, A., Alkhateeb, A., & ElMaraghy, W. (2021). Deep learning method based on big
data for defects detection in manufacturing systems Industry 4.0. International Journal
learning, which will be helpful for policymakers and practitioners to
of Industry and Sustainable Development, 2(1), 1–14.
understand the importance of deep learning approaches in Industry 4.0 Akdil, K. Y., Ustundag, A., & Cevikcan, E. (2017). Maturity and readiness model
for sustainable production. The main concern of developing Industry 4.0 for Industry 4.0 strategy. Springer Series in Advanced Manufacturing, 61–94.
practices is the loss of manual jobs. In this study, it is also discussed how 10.1007/978-3-319-57870-5_4.
Akter, S., McCarthy, G., Sajib, S., Michael, K., Dwivedi, Y. K., D’Ambra, J., &
deep learning can create job opportunities in the industries in future. Shen, K. N. (2021). Algorithmic bias in data-driven innovation in the age of AI. Inter-
The managers of manufacturing organisations can address the various national Journal of Information Management, 60, Article 102387.
research issues and challenges in the practical industrial scenario, which Alexopoulos, K., Nikolakis, N., & Chryssolouris, G. (2020). Digital twin-driven supervised
machine learning for the development of artificial intelligence applications in manu-
will help them experience more sustainability topics. The main ques- facturing. International Journal of Computer Integrated Manufacturing, 33(5), 429–439.
tion which arises from the present stud is whether the adoption of deep Alipour, M., Harris, D. K., & Miller, G. R. (2019). Robust pixel-level crack detection using
learning approaches can be helpful for industries to achieve sustainable deep fully convolutional neural networks. Journal of Computing in Civil Engineering,
33(6), Article 04019040.
production. Also, how the proposed framework can be beneficial for the Ammar, M., Haleem, A., Javaid, M., Walia, R., & Bahl, S. (2021). Improving material
enterprises to overcome the challenges in adopting the deep learning- quality management and manufacturing organizations system through Industry 4.0
based approaches. However, there is a need to conduct an extensive technologies. Materials Today: Proceedings, 45, 5089–5096.
Ardabili, S., Mosavi, A., Dehghani, M., & Várkonyi-Kóczy, A. R. (2019). Deep learn-
survey to investigate whether the deep learning approaches are con- ing and machine learning in hydrological processes climate change and earth sys-
tributing to sustainable production. The practitioners and policymakers tems a systematic review. In International Conference on Global Research and Education
can address these issues in future studies. (pp. 52–62).
Arellano-Espitia, F., Delgado-Prieto, M., Martinez-Viol, V., Saucedo-Dorantes, J. J., & Os-
ornio-Rios, R. A. (2020). Deep-learning-based methodology for fault diagnosis in elec-
8. Conclusion tromechanical systems. Sensors, 20(14), 3949.
Arinez, J. F., Chang, Q., Gao, R. X., Xu, C., & Zhang, J. (2020). Artificial intelligence in
advanced manufacturing: Current status and future outlook. Journal of Manufacturing
The present study is focused on presenting a state-of-the-art review of Science and Engineering, 142(11), Article 110804.
the deep learning approaches for sustainable production. In this study, Ashok, M., Madan, R., Joha, A., & Sivarajah, U. (2022). Ethical framework for artificial
intelligence and digital technologies. International Journal of Information Management,
the role of different deep learning models for sustainable production,
62, Article 102433.
various research themes and a framework for sustainable production Azamfar, M., Li, X., & Lee, J. (2020). Deep learning-based domain adaptation method for
is discussed with implications for the practitioners, policymakers and fault diagnosis in semiconductor manufacturing. IEEE Transactions on Semiconductor
Manufacturing, 33(3), 445–453.
researchers. In this study, three major digital scientific databases, i.e.,
Bag, S., Gupta, S., & Kumar, S. (2021). Industry 4.0 adoption and 10R advance manu-
IEEE, Scopus and Web of Science, are considered for the state-of-the- facturing capabilities for sustainable development. International Journal of Production
art review. The analysis shows that research in deep learning in the Economics, 231. 10.1016/j.ijpe.2020.107844.
manufacturing area is still limited. Industries are still struggling with Bag, S., Pretorius, J. H. C., Gupta, S., & Dwivedi, Y. K. (2021). Role of institutional pres-
sures and resources in the adoption of big data analytics powered artificial intelli-
the lack of knowledge and benefits of the deep learning approaches for gence, sustainable manufacturing practices and circular economy capabilities. Tech-
sustainable production. Our study shows how deep learning approaches nological Forecasting and Social Change, 163. 10.1016/j.techfore.2020.120420.
can contribute to quality management, predictive maintenance, reliabil- Bai, C., Dallasega, P., Orzes, G., & Sarkis, J. (2020). Industry 4.0 technologies assess-
ment: A sustainability perspective. International Journal of Production Economics, 229.
ity analysis and fault diagnosis. The relationship between deep learning 10.1016/j.ijpe.2020.107776.
techniques and sustainability is also discussed. The study’s main con- Bai, Y., Li, C., Sun, Z., & Chen, H. (2017). Deep neural network for manufacturing qual-
tribution is developing a deep learning-based framework for sustain- ity prediction. In 2017 Prognostics and System Health Management Conference (PH-
M-Harbin) (pp. 1–5).
able production, which will benefit the manufacturing organisations to Belhadi, A., Kamble, S., Fosso Wamba, S., & Queiroz, M. M. (2021). Building supply-chain
achieve sustainable production in the Industry 4.0-based business mod- resilience: An artificial intelligence-based technique and decision-making framework.
els. However, the proposed framework is not empirically tested, which International Journal of Production Research, 1–21.
Berghout, T., Mouss, L.-H., Bentrcia, T., Elbouchikhi, E., & Benbouzid, M. (2021). A deep
can be done in future studies by conducting the age survey across the
supervised learning approach for condition-based maintenance of naval propulsion
various industry sectors. The study also reveals that many deep learning systems. Ocean Engineering, 221, Article 108525.
approaches have been developed in the past few years, making new re- Bhandari, B., & Park, G. (2021). Development of a surface roughness evaluation method
from light and shade composition using deep learning. IEIE Transactions on Smart
search scopes in manufacturing. The advantages of various deep learn-
Processing & Computing, 10(3), 189–198.
ing models in manufacturing are also discussed in the present study. Bhuvaneswari, V., Priyadharshini, M., Deepa, C., Balaji, D., Rajeshkumar, L.,
Adopting deep learning approaches has sustainability benefits in pre- & Ramesh, M. (2021). Deep learning for material synthesis and manu-
dictive maintenance, reducing machine downtime, quality management facturing systems: A review. Materials Today: Proceedings, 46, 3263–3269.
10.1016/j.matpr.2020.11.351.
and fault diagnosis. Bichon, B. J., Eldred, M. S., Swiler, L. P., Mahadevan, S., & McFarland, J. M. (2008).
Efficient global reliability analysis for nonlinear implicit performance functions. AIAA
Journal, 46(10), 2459–2468.
8.1. Limitations Butte, S., Prashanth, A., & Patil, S. (2018). Machine learning based predictive maintenance
strategy: A super learning approach with deep neural networks. 1–5.
This study has a limitation related to considering the digital scientific Carvalho, N., Chaim, O., Cazarini, E., & Gerolamo, M. (2018). Manufacturing in the fourth
industrial revolution: A positive prospect in sustainable manufacturing. In K. H. Sh-
databases for the article collection. In this study, we have considered pitalni M. Seliger G.,. Wertheim R.,. Fischer A. (Ed.), Procedia Manufacturing (Vol. 21,
only three significant databases. Other scientific databases, i.e., Google pp. 671–678). Elsevier B.V. https://doi.org/10.1016/j.promfg.2018.02.170

11
A. Jamwal, R. Agrawal and M. Sharma International Journal of Information Management Data Insights 2 (2022) 100107

Carvalho, T. P., Soares, F. A., Vita, R., Francisco, R., da, P., Basto, J. P., & Al- Jamwal, A., Agrawal, R., Sharma, M., & Giallanza, A. (2021). Industry 4.0 technologies
calá, S. G. (2019). A systematic literature review of machine learning methods applied for manufacturing sustainability: A systematic review and future research directions.
to predictive maintenance. Computers & Industrial Engineering, 137, Article 106024. Applied Sciences, 11(12), 5725.
Chakraborty, A., & Kar, A. K. (2017). Swarm intelligence: A review of algorithms. Na- Jamwal, A., Agrawal, R., Sharma, M., Kumar, A., Kumar, V., & Garza-Reyes, J. A. (2021).
ture-Inspired Computing and Optimization, 475–494. Machine learning applications for sustainable manufacturing: A bibliometric-based
Chauhan, C., Singh, A., & Luthra, S. (2021). Barriers to industry 4.0 adoption and its review for future research. Journal of Enterprise Information Management.
performance implications: An empirical investigation of emerging economy. Journal Jamwal, A., Agrawal, R., Sharma, M., Kumar, V., & Kumar, S. (2021). Developing A sus-
of Cleaner Production, 285. 10.1016/j.jclepro.2020.124809. tainability framework for Industry 4.0. Procedia CIRP, 98, 430–435.
Chauhan, N. K., & Singh, K. (2018). A review on conventional machine learning vs deep Javaid, M., Haleem, A., Singh, R. P., Khan, S., & Suman, R. (2021). Blockchain technol-
learning. In 2018 International Conference on Computing, Power and Communication ogy applications for Industry 4.0: A literature-based review. Blockchain: Research and
Technologies (GUCON) (pp. 347–352). Applications, Article 100027.
Chen, B., Liu, Y., Zhang, C., & Wang, Z. (2020). Time series data for equipment reliability Javaid, M., Haleem, A., Singh, R. P., Rab, S., & Suman, R. (2021). Significance of sensors
analysis with deep learning. IEEE Access, 8, 105484–105493. for industry 4.0: Roles, capabilities, and applications. Sensors International, 2, Article
Chen, C., Liu, Y., Sun, X., Di Cairano-Gilfedder, C., & Titmus, S. (2021). An integrated 100110.
deep learning-based approach for automobile maintenance prediction with GIS data. Jena, M. C., Mishra, S. K., & Moharana, H. S. (2020). Application of Industry 4.0 to
Reliability Engineering & System Safety, 216, Article 107919. enhance sustainable manufacturing. Environmental Progress and Sustainable Energy,
Chen, Y.-F., Lin, Y.-K., & Huang, C.-F. (2021). Using deep neural networks to evaluate the 39(1). 10.1002/ep.13360.
system reliability of manufacturing networks. International Journal of Performability Jia, F., Lei, Y., Guo, L., Lin, J., & Xing, S. (2018). A neural network constructed by deep
Engineering, 17(7). learning technique and its application to intelligent fault diagnosis of machines. Neu-
Chien, C.-F., Dauzère-Pérès, S., Huh, W. T., Jang, Y. J., & Morrison, J. R. (2020). Artificial rocomputing, 272, 619–628.
intelligence in manufacturing and logistics systems: Algorithms, applications, and case Jung, H., Jeon, J., Choi, D., & Park, J.-Y. (2021). Application of machine learning tech-
studies. International Journal of Production Research, 58(9), 2730–2731. niques in injection molding quality prediction: Implications on sustainable manufac-
Chui, K. T., Lytras, M. D., & Visvizi, A. (2018). Energy sustainability in smart cities: Artifi- turing industry. Sustainability, 13(8), 4120.
cial intelligence, smart monitoring, and optimization of energy consumption. Energies, Kamble, S. S., Gunasekaran, A., Ghadge, A., & Raut, R. (2020). A performance measure-
11(11), 2869. ment system for industry 4.0 enabled smart manufacturing system in SMMEs- A re-
Chung, Y., Ahn, S., Yang, J., & Lee, J. (2017). Comparison of deep learning frameworks: view and empirical investigation. International Journal of Production Economics, 229.
About theano, tensorflow, and cognitive toolkit. Journal of Intelligence and Information 10.1016/j.ijpe.2020.107853.
Systems, 23(2), 1–17. Kar, A. K. (2016). Bio inspired computing–a review of algorithms and scope of applica-
Cioffi, R., Travaglioni, M., Piscitelli, G., Petrillo, A., & De Felice, F. (2020). Artificial intel- tions. Expert Systems with Applications, 59, 20–32.
ligence and machine learning applications in smart production: Progress, trends, and Khan, I. H., & Javaid, M. (2021). Role of Internet of Things (IoT) in adoption of Industry
directions. Sustainability (Switzerland), 12(2). 10.3390/su12020492. 4.0. Journal of Industrial Integration and Management, Article 2150006.
Çınar, Z. M., Abdussalam Nuhu, A., Zeeshan, Q., Korhan, O., Asmael, M., & Khan, S., & Yairi, T. (2018). A review on the application of deep learning in system health
Safaei, B. (2020). Machine learning in predictive maintenance towards sustainable management. Mechanical Systems and Signal Processing, 107, 241–265.
smart manufacturing in industry 4.0. Sustainability, 12(19), 8211. Kim, H., Jung, W.-K., Park, Y.-C., Lee, J.-W., & Ahn, S.-H. (2022). Broken stitch detec-
Cohen, S., & Macek, J. (2021). Cyber-physical process monitoring systems, real-time big tion method for sewing operation using CNN feature map and image-processing tech-
data analytics, and industrial artificial intelligence in sustainable smart manufactur- niques. Expert Systems with Applications, 188, Article 116014.
ing. Economics, Management & Financial Markets, 16(3). Kim, J. H. (2017). A review of cyber-physical system research relevant to the emerging IT
Collins, C., Dennehy, D., Conboy, K., & Mikalef, P. (2021). Artificial intelligence in infor- trends: Industry 4.0, IoT, big data, and cloud computing. Journal of Industrial Integra-
mation systems research: A systematic literature review and research agenda. Inter- tion and Management, 2(03), Article 1750011.
national Journal of Information Management, 60, Article 102383. Kotsiantis, S. B., Zaharakis, I. D., & Pintelas, P. E. (2006). Machine learning: A review of
Collins, K. (2020). Cyber-physical production networks, real-time big data analytics, and classification and combining techniques. Artificial Intelligence Review, 26(3), 159–190.
cognitive automation in sustainable smart manufacturing. Journal of Self-Governance Kushwaha, A. K., Kar, A. K., & Dwivedi, Y. K. (2021). Applications of big data in emerging
and Management Economics, 8(2), 21–27. 10.22381/JSME8220203. management disciplines: A literature review using text mining. International Journal
Davis, R., Sessions, B.-O., & Check, A. R. (2015). Industry 4.0. Digitalisation for Produc- of Information Management Data Insights, 1(2), Article 100017.
tivity and Growth. European Parliament, Members’ Research Service. LeCun, Y., Bengio, Y., & Hinton, G. (2015). Deep learning. Nature, 521(7553), 436–444.
de Sousa Jabbour, A. B. L., Jabbour, C. J. C., Foropon, C., & Filho, M. G. (2018). When 10.1038/nature14539.
titans meet – Can industry 4.0 revolutionise the environmentally-sustainable manu- Lee, J., Davari, H., Singh, J., & Pandhare, V. (2018). Industrial Artificial Intelligence for
facturing wave? The role of critical success factors. Technological Forecasting and Social industry 4.0-based manufacturing systems. Manufacturing Letters, 18, 20–23.
Change, 132, 18–25. 10.1016/j.techfore.2018.01.017. Lee, J., Lee, Y. C., & Kim, J. T. (2020). Fault detection based on one-class deep learning
Dikhanbayeva, D., Shaikholla, S., Suleiman, Z., & Turkyilmaz, A. (2020). Assessment of for manufacturing applications limited to an imbalanced database. Journal of Manu-
industry 4.0 maturity models by design principles. Sustainability (Switzerland), 12(23), facturing Systems, 57, 357–366.
1–22. 10.3390/su12239927. Li, B., Hou, B., Yu, W., Lu, X., & Yang, C. (2017). Applications of artificial intelligence in
Enyoghasi, C., & Badurdeen, F. (2021). Industry 4.0 for sustainable manufacturing: Op- intelligent manufacturing: A review. Frontiers of Information Technology & Electronic
portunities at the product, process, and system levels. Resources, Conservation and Re- Engineering, 18(1), 86–96.
cycling, 166. 10.1016/j.resconrec.2020.105362. Li, S., Ma, Y., Lu, X., Qiao, F., & Liu, J. (2020). Adaptive scheduling for smart shop floor
Ferrari, A. M., Volpi, L., Settembre-Blundo, D., & García-Muiña, F. E. (2021). Dynamic life based on deep Q-network. In 2020 IEEE 16th International Conference on Automation
cycle assessment (LCA) integrating life cycle inventory (LCI) and Enterprise resource Science and Engineering (CASE) (pp. 618–623).
planning (ERP) in an industry 4.0 environment. Journal of Cleaner Production, 286. Liu, J., Chang, H., Forrest, J. Y.-L., & Yang, B. (2020). Influence of artificial intelligence
10.1016/j.jclepro.2020.125314. on technological innovation: Evidence from the panel data of china’s manufacturing
Gaikwad, A., Giera, B., Guss, G. M., Forien, J.-B., Matthews, M. J., & Rao, P. (2020). sectors. Technological Forecasting and Social Change, 158, Article 120142.
Heterogeneous sensing and scientific machine learning for quality assurance in laser Loyer, J.-L., Henriques, E., Fontul, M., & Wiseall, S. (2016). Comparison of Ma-
powder bed fusion–A single-track study. Additive Manufacturing, 36, Article 101659. chine Learning methods applied to the estimation of manufacturing cost of jet
Guo, Y., Liu, Y., Oerlemans, A., Lao, S., Wu, S., & Lew, M. S. (2016). Deep learning for engine components. International Journal of Production Economics, 178, 109–119.
visual understanding: A review. Neurocomputing, 187, 27–48. 10.1016/j.ijpe.2016.05.006.
Haleem, A., & Javaid, M. (2019). Additive manufacturing applications in industry 4.0: A Machado, C. G., Winroth, M. P., & Ribeiro da Silva, E. H. D. (2020). Sustainable manufac-
review. Journal of Industrial Integration and Management, 4(04), Article 1930001. turing in Industry 4.0: An emerging research agenda. International Journal of Produc-
Han, J., Shihab, E., Wan, Z., Deng, S., & Xia, X. (2020). What do programmers discuss tion Research, 58(5), 1462–1484. 10.1080/00207543.2019.1652777.
about deep learning frameworks. Empirical Software Engineering, 25(4), 2694–2747. Mao, S., Wang, B., Tang, Y., & Qian, F. (2019). Opportunities and challenges of artifi-
Hansen, E. B., & Bøgh, S. (2021). Artificial intelligence and internet of things in small and cial intelligence for green manufacturing in the process industry. Engineering, 5(6),
medium-sized enterprises: A survey. Journal of Manufacturing Systems, 58, 362–372. 995–1002.
He, M., & He, D. (2017). Deep learning based approach for bearing fault diagnosis. IEEE Miotto, R., Wang, F., Wang, S., Jiang, X., & Dudley, J. T. (2018). Deep learning for
Transactions on Industry Applications, 53(3), 3057–3065. healthcare: Review, opportunities and challenges. Briefings in Bioinformatics, 19(6),
Herath, H., & Mittal, M. (2022). Adoption of artificial intelligence in smart cities: A com- 1236–1246.
prehensive review. International Journal of Information Management Data Insights, 2(1), Mittal, S., Khan, M. A., Romero, D., & Wuest, T. (2018). A critical review of smart man-
Article 100076. ufacturing & Industry 4.0 maturity models: Implications for small and medium-sized
Hidayatno, A., Destyanto, A. R., & Hulu, C. A. (2019). Industry 4.0 technology implementa- enterprises (SMEs). Journal of Manufacturing Systems, 49, 194–214.
tion impact to industrial sustainable energy in Indonesia: A model conceptualization. Müller, J.M. (2019). Contributions of Industry 4.0 to quality management—A SCOR per-
In H. B (Ed.), 5th International Conference on Power and Energy Systems Engineering, spective. In Y. F. Ivanov D. Dolgui A. (Ed.), IFAC-PapersOnLine (Vol. 52, Issue 13, pp.
CPESE 2018: 156 (pp. 227–233). Elsevier Ltd. 10.1016/j.egypro.2018.11.133. 1236–1241). Elsevier B.V. https://doi.org/10.1016/j.ifacol.2019.11.367
Hoang, D.-T., & Kang, H.-J. (2019). A survey on deep learning based bearing fault diag- Namuduri, S., Narayanan, B. N., Davuluru, V. S. P., Burton, L., & Bhansali, S. (2020). Deep
nosis. Neurocomputing, 335, 327–335. learning methods for sensor based predictive maintenance and future perspectives for
Hochreiter, S., & Schmidhuber, J. (1997). Long short-term memory. Neural Computation, electrochemical sensors. Journal of The Electrochemical Society, 167(3), Article 037552.
9(8), 1735–1780. Nguyen, K. T., & Medjaher, K. (2019). A new dynamic predictive maintenance framework
Horváthová, M., Lacko, R., & Hajduová, Z. (2019). Using industry 4.0 concept – digital using deep learning for failure prognostics. Reliability Engineering & System Safety, 188,
twin – to improve the efficiency of leather cutting in automotive industry. Quality 251–262.
Innovation Prosperity, 23(2), 1–12. 10.12776/QIP.V23I2.1211. Nikolakis, N., Siaterlis, G., & Alexopoulos, K. (2020). A machine learning approach for

12
A. Jamwal, R. Agrawal and M. Sharma International Journal of Information Management Data Insights 2 (2022) 100107

improved shop-floor operator support using a two-level collaborative filtering and Syafrudin, M., Alfian, G., Fitriyani, N. L., & Rhee, J. (2018). Performance analysis of IoT-
gamification features. Procedia CIRP, 93, 455–460. based sensor, big data processing, and machine learning model for real-time moni-
Ning, F., Shi, Y., Cai, M., Xu, W., & Zhang, X. (2020). Manufacturing cost estimation based toring system in automotive manufacturing. Sensors (Basel, Switzerland), 18(9), 2946
on a deep-learning method. Journal of Manufacturing Systems, 54, 186–195. PubMed. 10.3390/s18092946.
Niri, M. F., Liu, K., Apachitei, G., Román-Ramírez, L. A., Lain, M., Widanage, D., & Tang, S., Yuan, S., & Zhu, Y. (2019). Deep learning-based intelligent fault diagnosis meth-
Marco, J. (2022). Quantifying key factors for optimised manufacturing of Li-ion bat- ods toward rotating machinery. IEEE Access, 8, 9335–9346.
tery anode and cathode via artificial intelligence. Energy and AI, 7, Article 100129. Tripathi, S., Muhr, D., Brunner, M., Jodlbauer, H., Dehmer, M., & Em-
Nosratabadi, S., Mosavi, A., Duan, P., Ghamisi, P., Filip, F., Band, S. S., Reuter, U., mert-Streib, F. (2021). Ensuring the robustness and reliability of data-driven
Gama, J., & Gandomi, A. H. (2020). Data science in economics: Comprehensive re- knowledge discovery models in production and manufacturing. Frontiers in Artificial
view of advanced machine learning and deep learning methods. Mathematics, 8(10), Intelligence, 4, 22.
1799. Verma, S., Sharma, R., Deb, S., & Maitra, D. (2021). Artificial intelligence in marketing:
Ozdemir, R., & Koc, M. (2019). A quality control application on a smart factory prototype Systematic review and future research direction. International Journal of Information
using deep learning methods. In 2019 IEEE 14th International Conference on Computer Management Data Insights, 1(1), Article 100002.
Sciences and Information Technologies (CSIT): 1 (pp. 46–49). Vinuesa, R., Azizpour, H., Leite, I., Balaam, M., Dignum, V., Domisch, S., Felländer, A.,
Parvat, A., Chavan, J., Kadam, S., Dev, S., & Pathak, V. (2017). A survey of deep-learning Langhans, S. D., Tegmark, M., & Nerini, F. F. (2020). The role of artificial intelligence
frameworks. In 2017 International Conference on Inventive Systems and Control (ICISC) in achieving the sustainable development goals. Nature Communications, 11(1), 1–10.
(pp. 1–7). Votto, A. M., Valecha, R., Najafirad, P., & Rao, H. R. (2021). Artificial intelligence in
Priore, P., De la Fuente, D., Gomez, A., & Puente, J. (2001). A review of machine tactical human resource management: A systematic literature review. International
learning in dynamic scheduling of flexible manufacturing systems. In Artificial Journal of Information Management Data Insights, 1(2), Article 100047.
Intelligence for Engineering Design, Analysis and Manufacturing: 15 (pp. 251–263). Voulodimos, A., Doulamis, N., Doulamis, A., & Protopapadakis, E. (2018). Deep learning
10.1017/s0890060401153059. for computer vision: A brief review. Computational Intelligence and Neuroscience, 2018.
Raj, A., Dwivedi, G., Sharma, A., Lopes de Sousa Jabbour, A. B., & Rajak, S. (2020). Bar- Wan, J., Li, X., Dai, H.-N., Kusiak, A., Martínez-García, M., & Li, D. (2020). Artificial-in-
riers to the adoption of industry 4.0 technologies in the manufacturing sector: An telligence-driven customized manufacturing factory: Key technologies, applications,
inter-country comparative perspective. International Journal of Production Economics, and challenges. In Proceedings of the IEEE: 109 (pp. 377–398).
224(August 2019), Article 107546 –107546. 10.1016/j.ijpe.2019.107546. Wang, J., Ma, Y., Zhang, L., Gao, R. X., & Wu, D. (2018). Deep learning for smart manu-
Rajawat, A. S., Bedi, P., Goyal, S. B., Shaw, R. N., & Ghosh, A. (2022). Reliability analysis facturing: Methods and applications. Journal of Manufacturing Systems, 48, 144–156.
in cyber-physical system using deep learning for smart cities industrial IoT network Wang, K., & Wang, Y. (2017). How AI affects the future predictive maintenance: A primer
node. In AI and IoT for Smart City Applications (pp. 157–169). Springer. of deep learning. In International Workshop of Advanced Manufacturing and Automation
Ram, S. K., & Tyagi, R. D. (2020). Artificial intelligence and computational sustainability. (pp. 1–9).
Sustainability, 627–649. 10.1002/9781119434016.ch29. Wang, P., & Gao, R. X. (2020). Transfer learning for enhanced machine fault diagnosis in
Rawat, S., Rawat, A., Kumar, D., & Sabitha, A. S. (2021). Application of machine learn- manufacturing. CIRP Annals, 69(1), 413–416.
ing and data visualization techniques for decision support in the insurance sector. Wuest, T., Weimer, D., Irgens, C., & Thoben, K.-D. (2016). Machine learning in manufac-
International Journal of Information Management Data Insights, 1(2), Article 100012. turing: Advantages, challenges, and applications. Production & Manufacturing Research,
Ryan, K., Lengyel, J., & Shatruk, M. (2018). Crystal structure prediction via deep learning. 4(1), 23–45. 10.1080/21693277.2016.1192517.
Journal of the American Chemical Society, 140(32), 10158–10168. Yadav, G., Kumar, A., Luthra, S., Garza-Reyes, J. A., Kumar, V., & Batista, L. (2020).
Sahu, C. K., Young, C., & Rai, R. (2021). Artificial intelligence (AI) in augmented reality A framework to achieve sustainability in manufacturing organisations of develop-
(AR)-assisted manufacturing applications: A review. International Journal of Production ing economies using industry 4.0 technologies’ enablers. Computers in Industry, 122.
Research, 59(16), 4903–4959. 10.1016/j.compind.2020.103280.
Sajid, S., Haleem, A., Bahl, S., Javaid, M., Goyal, T., & Mittal, M. (2021). Data science Yang, J., Li, S., Wang, Z., Dong, H., Wang, J., & Tang, S. (2020). Using deep learning
applications for predictive maintenance and materials science in context to Industry to detect defects in manufacturing: A comprehensive survey and current challenges.
4.0. Materials Today: Proceedings, 45, 4898–4905. Materials, 13(24), 5755.
Salloum, S. A., Alshurideh, M., Elnagar, A., & Shaalan, K. (2020). Machine learning and Yao, X., Zhou, J., Zhang, J., & Boër, C.R. (2017). From intelligent manufacturing to smart
deep learning techniques for cybersecurity: A review. In AICV (pp. 50–57). manufacturing for industry 4.0 driven by next generation artificial intelligence and
Shams, S., Platania, R., Lee, K., & Park, S.-J. (2017). Evaluation of deep learning frame- further on. 311–318.
works over different HPC architectures. In 2017 IEEE 37th International Conference on Zhang, A., He, Y., Han, X., Li, Y., Yang, X., & Zhang, Z. (2020). Modeling Product Manufac-
Distributed Computing Systems (ICDCS) (pp. 1389–1396). turing Reliability with Quality Variations Centered on the Multilayered Coupling Op-
Shao, S.-Y., Sun, W.-J., Yan, R.-Q., Wang, P., & Gao, R. X. (2017). A deep learning approach erational Characteristics of Intelligent Manufacturing Systems. Sensors, 20(19), 5677.
for fault diagnosis of induction motors in manufacturing. Chinese Journal of Mechanical Zhang, D., Pee, L. G., & Cui, L. (2021). Artificial intelligence in E-commerce fulfillment:
Engineering, 30(6), 1347–1356. 10.1007/s10033-017-0189-y. A case study of resource orchestration at Alibaba’s smart warehouse. International
Sharma, R., Jabbour, C. J. C., & Lopes de Sousa Jabbour, A. B. (2020). Sustainable man- Journal of Information Management, 57, Article 102304.
ufacturing and industry 4.0: What we know and what we don’t. Journal of Enterprise Zhang, J., & Gao, R. X. (2021). Deep learning-driven data curation and model interpreta-
Information Management. 10.1108/JEIM-01-2020-0024. tion for smart manufacturing. Chinese Journal of Mechanical Engineering, 34(1), 1–21.
Shinde, P. P., & Shah, S. (2018). A review of machine learning and deep learning appli- Zhang, T., Homri, L., Dantan, J., & Siadat, A. (2021). Conceptual maps of reliability anal-
cations. In 2018 Fourth International Conference on Computing Communication Control ysis applied in reconfigurable manufacturing system. 136–145.
and Automation (ICCUBEA) (pp. 1–6). Zhang, W., Wang, X., Cabrera, D., & Bai, Y. (2020). Product quality reliability analysis
Shivajee, V., Singh, R. K., & Rastogi, S. (2019). Manufacturing conversion cost reduction based on rough Bayesian network. International Journal of Performability Engineering,
using quality control tools and digitization of real-time data. Journal of Cleaner Pro- 16(1).
duction, 237. 10.1016/j.jclepro.2019.117678. Zhao, G., Zhang, G., Ge, Q., & Liu, X. (2016). Research advances in fault diagnosis and
Shrestha, A., & Mahmood, A. (2019). Review of deep learning algorithms and architec- prognostic based on deep learning. In 2016 Prognostics and System Health Management
tures. IEEE Access, 7, 53040–53065. Conference (PHM-Chengdu) (pp. 1–6).
Sun, W., Li, M., Li, Y., Wu, Z., Sun, Y., Lu, S., Xiao, Z., Zhao, B., & Sun, K. (2019). The use Zhao, Z.-Q., Zheng, P., Xu, S., & Wu, X. (2019). Object detection with deep learning: A re-
of deep learning to fast evaluate organic photovoltaic materials. Advanced Theory and view. IEEE Transactions on Neural Networks and Learning Systems, 30(11), 3212–3232.
Simulations, 2(1), Article 1800116. Zonta, T., da Costa, C. A., da Rosa Righi, R., de Lima, M. J., da Trindade, E. S., &
Sung, T. K. (2018). Industry 4.0: A Korea perspective. Technological Forecasting and Social Li, G. P. (2020). Predictive maintenance in the Industry 4.0: A systematic literature
Change, 132, 40–45. review. Computers & Industrial Engineering, Article 106889.

13