See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/336239156

Intelligent decision support for maintenance: an overview and future trends

Article · October 2019

DOI: 10.1080/0951192X.2019.1667033

CITATIONS READS

55 2,351

5 authors, including:

Chris Turner Christos Emmanouilidis

University of Surrey University of Groningen
108 PUBLICATIONS 3,428 CITATIONS 154 PUBLICATIONS 2,350 CITATIONS

SEE PROFILE SEE PROFILE

Tetsuo Tomiyama Ashutosh Tiwari

International Professional University of Technology in Tokyo The University of Sheffield
291 PUBLICATIONS 8,661 CITATIONS 312 PUBLICATIONS 9,456 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Chris Turner on 15 October 2019.

The user has requested enhancement of the downloaded file.

 Intelligent decision support for maintenance: An overview and future trends

1
C.J.Turner*, 2C. Emmanouilidis, 2T.Tomiyama, 3A.Tiwari, 2R.Roy
1Rik Medlik Building, Surrey Business School, University of Surrey, Guildford, Surrey, GU2 7XH, UK
2Building 50, Manufacturing Dept., Cranfield University, Bedford, Bedfordshire, MK43 0AL, UK
3Amy Johnson Building, Department of Automatic Control and Systems Engineering, University of
Sheffield, Portobello Street, Sheffield, S1 3JD, UK

Abstract: The changing nature of manufacturing, in recent years, is evident in industry’s

willingness to adopt network connected intelligent machines in their factory development
plans. A number of joint corporate/government initiatives also describe and encourage the
adoption of Artificial Intelligence (AI) in the operation and management of production lines.
Machine learning will have a significant role to play in the delivery of automated and
intelligently supported maintenance decision making systems. While e-maintenance practice
provides a framework for internet connected operation of maintenance practice the advent of
IoT has changed the scale of internetworking and new architectures and tools are needed.
While advances in sensors and sensor fusion techniques have been significant in recent
years, the possibilities brought by IoT create new challenges in the scale of data and its
analysis. The development of audit trail style practice for the collection of data and the
provision of a comprehensive framework for its processing, analysis and use should be a
valuable contribution in addressing the new data analytics challenges for maintenance created
by internet connected devices. This paper proposes that further research should be conducted
into audit trail collection of maintenance data and the provision of comprehensive framework
for its processing analysis and use, allowing future systems to enable ‘Human in the loop’
interactions.
Keywords: Machine learning, Industry 4.0, E-Maintenance, Intelligent Maintenance.

1. Introduction
Increasingly manufacturing industry is adopting network connected intelligent machines in
their factory development plans. This has created a new wave of interest in incorporating
advances in Artificial Intelligence (AI), which is described and encouraged by a number of
international government/industry initiatives. The Industry 4.0 movement is one such initiative,
between the German government and national industries, with a role to envisage and promote
the use of new technologies and organizational methods for manufacturing (German Federal
Government, 2016). Cyber Physical Systems (CPS) are a core theme of Industry 4.0,
encompassing the further integration between machines and computing resources, leveraged
in part by the Internet of Things (IoT). In addition, the enhanced information processing and
analysis opportunities provided by the ubiquity of sensor use in modern machinery to provide
data streams and resulting Big Data sets is seen to create potential for new products and new
types of manufacturing models. In the US the Industrial Internet Consortium is an initiative
setup between the US government and large industrial organizations. While having similarities
to the vision provided by Industry 4.0 there is a concentration on three core components:
Intelligent Production Machines, Analysis of Sensed Data, People and Machines working
together (Posada et al. 2015). The industrial Internet is also much more focused on the
visualisation of data at both global and local levels (Industrial Internet Consortium, 2017). The
central differentiator between the two visions is that while Industry 4.0 focusses on
manufacturing, the remit of the Industrial Internet is much wider bringing in other sectors of
the economy as well. In the opinion of the authors of this paper the central challenge is how
maintenance can best utilise the opportunities brought by this expansion of AI into the
manufacturing arena. The quality and provenance of data are important factors in data
management and a key success factor for when engaging in any form of analytics. With
maintenance rapidly adopting key Industry 4.0 technologies, such issues attain increased
importance for successful applications and services. The path towards Industry 4.0-enabled
maintenance has seen developments in Predictive Maintenance, Condition Based
Maintenance, Intelligent Maintenance, and E-Maintenance; leading to the introduction of IoT,
Context aware computing and Audit Trail concepts for maintenance. This paper offers a critical
overview of this evolving landscape, including Industry 4.0 applications in the area of
maintenance. The paper concludes by finding that while maintenance is increasingly adopting
Industry 4.0 technologies, issues related to data governance, provenance and quality
management, are already well appreciated in the Big Data literature; these topics deserve
equal attention in this application domain and to this end a discussion of the potential of using
the Audit Trail for maintenance data is provided.

2. Intelligent Decision Support

2.1 Condition Based and Predictive Maintenance

Condition Based Maintenance (CBM) is the standard term employed to describe maintenance
strategies determined on the basis of the actual condition of assets, as identified by the
application of condition monitoring programmes (ISO 13372:2012)(BS EN 13306:2017). While
this general viewpoint holds a central role in literature, details on how exactly CBM benefits
from individual technologies, methods, has also received significant attention in the literature.
In industrial practice, CBM involves the performance of maintenance tasks triggered by the
analysis and interpretation of monitored hardware parameters and the associated decision
making rules as an integrated process (Liyanage et al., 2009). Peng et al. (2010) describe
Condition Based Maintenance (CBM) as a “decision making strategy to enable real-time
diagnosis of impending failures and prognosis of future equipment health, where the decision
to perform maintenance is reached by observing the “condition” of the system and its
components” or additionally on the basis of prognostics about the anticipated future condition.
While the diagnostics and prognostics parts of CBM approaches have benefited from
incorporated machine intelligence in order to associate measured data and parameters with
current and future machinery conditions (Jardine et al., 2006) (Emmanouilidis et al., 2006), an
early criticism of machine intelligence use in CBM has been that the research has
concentrated on very specific cases with only limited attempts to deliver solutions with generic
applicability (Lee et al. 2006). As a response to this Lee et al. (2006) put forward a toolkit for
predictive CBM based on sensor data, capable of working with different manufacturing
machines and set ups. In their review of machinery diagnostics and prognostics Jardine et al.
(2006) indicated a number of research directions for Condition Based Monitoring (CBM)
systems for condition based Maintenance, including the development of a new generation of
sensors for on-line data collection in real time and investigation into the provision of predictive
techniques based on collected data. The next step in the maintenance data processing chain
is to produce action recommendations, as highlighted in the OSA-CBM architecture (MIMOSA,
2017). This elevates a CBM strategy to proactive maintenance. Within such an approach,
asset events and errors are decomposed in a process flow, arguing that in understanding the
conditions leading up to a fault, more accurate estimates of safe operating limits can be
identified (Radkowski and Jasinski, 2014). The importance of utilising a range of evidence
contained in multiple data sets and streams when making CBM related decisions is highlighted
by Niu et al. (2010). Such a data fusion approach can bring benefits through the combination
of many condition measurements into a consolidated description of maintenance needs for an
individual component or unit under observation. Niu et al. (2010) outline a maintenance system
that takes advantage of data fusion and the OSA-CBM standard, providing a platform for the
optimised exploration of maintenance decisions and predictions. Building on the availability of
Big Data Bousdekis et al. (2015) provides a review of Condition Based Maintenance (CBM)
and proposes a framework for maintenance decision making, utilising expert knowledge,
which is capable of recommending maintenance actions for implementation.
The question of how much capital to invest in maintenance practice has been addressed in
the field of vibration monitoring by Al-Najjar and Alsyouf (2004). These authors concluded that
a framework of performance measurement should be utilised to ensure value for money is
being obtained from maintenance activities. Further research has been conducted into
calculation of the likelihood of failure of assets and the appropriate stage at which to conduct
maintenance interventions. Yao et al. (2016) identify two types of failure under CBM where in
the first instance an asset may fail before the monitoring threshold is reached and the second
where the monitoring threshold is exceeded without asset failure. Goyal et al. (2016) provide
a more recent review that includes a number of machine intelligence methods for CBM and
predictive maintenance practice and note that while such techniques provide good offline
models research scope still exists in harnessing them for real time prediction and decision
making. Accorsi et al. (2017) add a set of models to aid the prediction of faults in production
systems and explore machine learning techniques such as decision trees utilised for the
classification and identification of abnormal operating conditions derived from production
machine data streams. Accorsi et al. (2017) go onto propose a framework for data mining and
modelling related to CBM.

Prognostic maintenance practice is based on the prediction of likely breakdowns in hardware

formed from the analysis of collected parameters and degradation trends. Liyanage et al.
(2009) identify three prognostic approaches:

• Model based: Centred on detailed knowledge of a system and its interlinkages; its use
is limited due to inherent complexity of modern industrial systems.
• Data-driven: Requires historical parameter collection from monitored assets; requires
pattern recognition and machine intelligence techniques to realise actionable decision
making outcomes.
• Hybrid: Is a combination of the two aforementioned approaches requiring a joint
analysis of both known information about a system in combination with sensed data
points.
The use of prognostic maintenance practice to estimate the remaining useful life of a
component has been investigated by Van Horenbeek and Pintelton (2013). This work takes
into account inter-component dependencies in the degradation calculation approach and
prognostic maintenance policy developed (Van Horenbeek and Pintelton, 2013). Prognostic
maintenance practice can benefit from advances in data capture and the availability of big
data for a range of applications. Lee et al. (2013) describe the use of a Digital Twin whereby
a machine may be represented in digital form utilising CAD models and sensor streams from
the machine. Lee et al. (2013) also describe the possibility for similar machines to
communicate with each other to check and compare status to form more accurate feedback
to the maintenance monitoring system along with the use of self-aware sensors with built in
decision making capabilities. Lee et al. (2013) conclude with an outline of a cloud based cyber
physical model of machine data capture, analysis and use. The capture and integration of
expert knowledge to support and validate predicted routines is very much an active subject of
research. The use of prognostics in condition based maintenance through the construction of
a hybrid model incorporating expert knowledge is investigated by Galar et al. (2015). In this
work a combination of discrete data and semantic feedback (provided by experts) is combined
to provide decision support in relation to issues of component degradation (Galar et al. 2015).
Baysian approaches to prediction in maintenance are not new; McNaught and Zagorecki
(2009) have explored the use of a Bayesian network approach to prognostic modelling of
equipment in terms of maintenance. A more recent technique drawing on Bayesian theory for
prediction is put forward by Desforges et al. (2017) is described as a support system for
maintenance planning activities with the aim of modelling fault prorogation in subsystems for
improved prediction. In addition the technique aims to reduce the downtime of systems
enabling further efficiencies in planning to take place (Desforges et al., 2017). Niu and Jiang
(2017) propose a technique for prognostic control at a component level within a system while
enabling the optimisation of the system as a whole at a global level. This enables the
development of a suitable overall maintenance interval schedule based on sub system level
health prognostics (Niu and Jiang, 2017). Ragab et al. (2017) put forward a way of pattern
selection from condition monitoring data to support prognostic maintenance, a method that
does not rely on expert judgement and statistical base assumptions on initial set up. A case
study on prognostic techniques relating the maintenance of railway infrastructure is presented
by Marugan and Marquez (2016). Binary Decision Diagrams are used with fault trees to
provide an Internet based decision making process for problem diagnosis in railway points.
Recent work by Belkacem et al. (2017) investigated the combined approach of integrating
diagnostic and prognostic maintenance policies to provide a dynamic maintenance system; a
technique these authors aim to extend and develop further, in terms of its scalability, in future
research. The practice of prognostic maintenance must of course be viewed within a wider
maintenance system composed of the latest hardware and software. Such a system is
envisaged within the field on E-maintenance, which is the subject of the following section of
this paper. Xia et al. (2018) provide a concise summary of predictive techniques in use for
maintenance practice in a range of digital manufacturing activities; these authors note that
innovative manufacturing techniques, such as 3D printing, bring new challenges to the way
maintenance is performed, demanding new research into how supply chains support products
manufactured in this way. Vafaei et al. (2019) have investigated CBM from the perspective of
providing an approach for an early warning system. In this work Vafaei et al. (2019) utilise a
fuzzy inferencing approach to enable a system capable of developing what-if scenarios based
on generated rulesets regarding the potential for break downs in monitored production lines.
In their survey paper on CBM and prognostic techniques in industry Sakib and Wuest (2018)
make the case that a combination of such techniques is increasingly seen as the most likely
future path for maintenance practice, involving a multiphase approach to problem detection,
diagnosis and corrective/mitigating actions.

2.2 E-maintenance
Incorporating predictive maintenance approaches within E-maintenance aims to integrate
developments in web enabled communication technologies with semantically described data
resources, sensing technologies and artificial intelligence algorithms to realise new
capabilities for remote and ubiquitous maintenance. Levrat et al. (2008) state that inherent in
the concept of e-maintenance is the remote monitoring and management of assets though
Internet-based technology. Levrat et al. (2008) go onto propose a framework for e-
maintenance encompassing issues such as infrastructure, business processes and
information architecture, noting that further research is needed in terms of unified standards
for e-maintenance and the communication protocols required for effective operation. A
particular feature of e-maintenance, enabled though its framework, is the facilitation of fault
prediction in order to pre-emptively schedule mitigating maintenance activity. Voisin et al.
(2010) proposes a prognosis business process as a formalisation of the predictive feature of
e-maintenance practice in their proposed model. Muller et al. (2008) provides a review of the
main research works in the area of e-maintenance, focussing on definitions of e-maintenance
ranging from a maintenance strategy to a type of maintenance planning. Muller et al. 2008)
state that the combination of the latest ICT developments, especially with regard to
prognostics, with maintenance practice has led to the emergence of e-maintenance. However,
a more accurate assessment would see e-Maintenance as an enabling factor for more efficient
maintenance, which would also include prognostics, rather than the other way round. The
utility of machine learning in the successful delivery of e-maintenance has been noted by Ucar
and Qiu (2005). These authors also note the rise of wireless communications technology
(networks, sensors etc.). Arnaiz et al. (2010) provide a review of communication technology
use in e-maintenance and point to two trends; that of the use of wireless web enabled
communication technologies and the miniaturisation of sensing devices. The potential value
of RFID and other associated smart tagging technologies is noted by Adgar et al. (2010) along
with the rise of ubiquitous computing, a movement describing the almost universal availability
of miniaturised computing power in a range of, often, portable devices (Arnaiz et al., 2010;
Krommenacker et al. 2010). The prominence of one particular approach has been identified
by several authors (Arnaiz et al., 2010; Campos, 2009; Vogel-Heuser et al. 2014); that of
Agent technology, where sometimes geographically distributed software modules are able to
cooperate in order to autonomously fulfil a given objective or set of objectives. When used
with machine learning techniques this approach is particularly relevant to the field of e-
maintenance. Overall, e-maintenance is considered an umbrella term to include a range of
enabling technologies which facilitate the whole data process chain in maintenance, from data
acquisition through sensor miniaturisation, smart tags, and sensor networks, to wireless
communications and mobile devices, all the way to web-based and semantic computing for
offering maintenance services and decision support, including technology enablers for
maintenance training (Holmberg et al., 2010).
Holgado et al. (2016) identify a range of functionalities provided by e-maintenance applications
listing 10 categories of tools. These focused on provided diagnostic and prognostic
functionality were rated more highly for usefulness than those than were based on model
simulation. A recent evaluation work comparing diagnostic and prognostic maintenance
policies is provided by Belkacem et al. (2017). The use of AR (Augmented Reality), where
animations and graphics are overlaid on actual scenes in real time, is identified by Azuma
(1997) and Azuma et al. (2001) as an aid to maintenance activities. Henderson and Fiener
(2011) explore the use of AR for engineer knowledge assistance in maintenance and repair
activities. These findings are interesting as Turner et al. (2016) envisages the development of
AR with simulation, allowing models of production systems to be fed with data in real time and
overlaid on the actual physical view of the plant/assets in question. Such a combination of
technologies could act as a context relevant visualisation to aid ‘in-field’ maintenance
decisions. Ceruti et al. (2019) examine the use of AR within case studies drawn from aviation
maintenance practice, concluding that such an approach can streamline part identification
tasks and on the job training and support of maintenance technicians.
Real world case studies of e-maintenance systems in action can be found in industries such
as aerospace and rail and road maintenance where sensed data about both static and mobile
assets may be collected and analysed to make decisions about present and future
maintenance actions. (Ben-Daya et al., 2016). Ben-Daya et al. (2016) charts the rise of e-
maintenance, from manual systems, and CBM into web connected systems.
Increasingly industry is witnessing the gradual introduction of Cyber Physical Systems (CPS).
Such CPS systems are composed of deeply interconnected hardware and software systems
with sensing capabilities and are often able to provide intelligent decision support and decision
making functionalities to users (NIST, 2013). Holgado et al. (2016) notes the importance of
CPS and the increasing potential of machines to interact with their maintenance systems and
influence the works carried out and their timing. As a tangent to this Ruiz-Arenas et al. (2014)
explores many of the e-maintenance issues that pertain to CPS systems themselves and
provide the case study of a CPS enabled greenhouse as an example. Penna et al. (2014) and
Botelho et al. (2014) describe an approach for the visualisation of CPS integration in
maintenance systems and the development of maintenance scenarios using 3D modelling
tools. The aforementioned visualisation approach focusses on Human Computer Interaction
issues taking into account and designing interfaces for the support of human operators within
the maintenance process (Penna et al., 2014). One of the central components of e-
maintenance is the ability to freely collect, exchange and process data. One approach to this
is through the use of semantic technology and ontology use. Nuñez and Borsato (2017)
explore the potential of semantic technologies to describe machine health management and
prognostic forecasting of potential failure. The semantic framework built by Nuñez and Borsato
(2017) is provided in the form of prototype software to allow experimentation with a wide
variety of plant and machinery. Zhou et al. (2017) provide a potential augmentation to the
aforementioned semantic framework through their research of fault diagnosis and provision of
a requisite knowledge model. This work also utilises a semantic approach and envisages the
use of pattern recognition approaches such as Neural Network to better identify and classify
a fault through data analysis (Zhou et al., 2017). Li et al. (2017) in a review of artificial
intelligence/machine learning use in manufacturing discuss programmes for proactive and
preventative maintenance that would be possible within an intelligent manufacturing system.
The interconnected nature of organisational management systems and manufacturing
production machines in combination with accessible rich data and information sets is leading
to this new role for machine learning in industry (Li et al. 2017). In applications with high
maintenance needs there is a requirement to coordinate the supply chain responsible for spare
parts delivery. This subject has been researched by Espíndola et al. (2012) who put forward
a conceptual approach to combining an intelligent maintenance system with supply chain
coordination and planning processes. In addition da Silva et al. (2014a) also investigate the
integration of parts supply chain and planning with an intelligent maintenance system touching
on the use of ontology to describe communication within the combined architecture; along with
Saalmann et al. (2016) who propose a multi-layer ontology incorporating existing semantic
approaches to supply chain and intelligent systems. A particular use of ontology is in the
potential integration of spare parts supply chains and the field of CBM, where the two entities
possess distinct knowledge sets and express their data and parameters in different levels of
granularity and importance (Saalmann et al., 2016). Saalmann et al. (2016) make the case for
a common terminology and utilise DPWS (Device Profile for Web Services) in order to obtain
metadata from physical devices (DPWS is a standardised middleware for exposing
parameters about and data from physical hardware devices).
One particular challenge relating to e-maintenance data is that of missing values, where
connectivity issues and faulty sensors can lead to incomplete data recordings (Loukopoulos
et al., 2017). The research of Loukopoulos et al. (2017) explores the process of imputing
missing values in data relating to the e-maintenance practice required for compressors used
in the oil industry. This work investigates the use of computational intelligence approaches
such as self-organising map (SOM) Neural Network learning, K nearest neighbours classifier
and Bayesian techniques. Lou found that while SOM and KNN produced reasonable results
the best result was produced by Multiple Imputation MI (an uncertainty method used to
introduce simulated data based on Bayesian theory). Beyond missing values, Liyanage et al.
(2009) makes the point that there is also a need to keep experts in the loop; networks of people
are required to enrich data from systems and sensors with their own contributed observations
or knowledge. Djurjanovic et al., (2003) outline a watchdog agent which has been designed to
convert sensor data into health management information and Liyanage et al. (2009) place this
within an overall e-Maintenance framework. In combination with decision management
functionality the agent is designed as a semi-autonomous software module capable of
interaction and coordination with enterprise maintenance and manufacturing systems.
In line with Muller et al. (2008) definition of e-maintenance as a combination of
‘telemaintenance principles with Web services and modern e-collaboration principles’ enabling
knowledge exchange and intelligence based on the ability to identify and collect relevant and
timely parameters, a more recent wave of technologies presents a step change and the
possibility for real time proactive maintenance. This paradigm addressed existing practices
that were much more focussed on the improved management of maintenance, from reactive
to proactive activities, through prognostics based on largely disconnected datasets with
potential for data quality and timeliness penalties. The increasing importance of maintenance
as a service in industry is highlighted by Akkermans et al. (2019); these authors highlight that
the product service methodology has evolved into the provision of smart maintenance services
to complement products, made possible by more accurate quantifications of service needs
and costs based on real-time analytics.
2.3 IOT
Developments within hardware have increasingly leveraged the availability of almost
ubiquitous network connectivity provided by internet based communication protocols. This has
now culminated in IoT whereby hardware from sensors to entire machines may be web
addressed as interactive objects providing raw and often intelligently filtered data points to
client software applications. Cloud implementations utilise both network technologies and big
data production capabilities of IoT connected hardware to provide new distributed
manufacturing forms and the opportunity for prognostic flexible maintenance based on
intelligent near real-time analysis of live operating environments. The utilisation of cloud
technologies to enable CBM is one of the more recent research strands within e-maintenance.
Karim et al. (2016) make the case for what they term ‘Maintenance Analytics’, where four time
related perspectives of practice are defined in the utilisation of data provided by industrial
application cloud platforms. In a rail related case study, ‘Rail Cloud’, Karim et al. (2016) find
that a systematic treatment of maintenance data is required with its synthesis and integration
required for decision making in order to support the next generation of digitally monitored plant
and machinery. In the research of Truong (2018) it is acknowledged that the inherent
complexity of modern machinery IoT enabled cloud platforms require additional analysis of
interactions between system components, and that this analysis also requires human
intervention and decision making capabilities. Truong (2018) go onto propose a system
capable of automatically recognising when human expertise is required and alerting the
correct expert for input of knowledge and decision making capability. While this research
outlines an architecture for distributed analytics processing Truong (2018) notes that much
research is still required in the correct mapping of analytics to domain knowledge derived from
experts an in the facilitation of the ‘Human in the Loop’. Mourtzis et al. (2016) propose a shop
floor monitoring approach that includes CBM functionality delivered via a cloud infrastructure.
The approach of Mourtzis et al. (2016) highlighted the possibility of near real time data
acquisition and monitoring for maintenance decision making. In later work Mourtzis et al.
(2018) provide a cloud based model for IoT sensor data collection from a manufacturing
production line; highlighting the potential of such a system in its ability to interconnect the shop
floor with enterprise IT software, these authors elude to the possibility of fine grain control and
prediction at the individual machine tool sensor pack level. Wang et al. (2017) provide an
example model of cloud based prognostic maintenance practice outlining the advantages of
local processing of data on mobile devices in order to manage the overall analytics load within
the system and reduce the communications bandwidth required. In their work Wang et al.
(2017) cite the need for further research in the development of distributed data analysis
practice and co-ordination of heterogeneous data streams and for improved security for data
communication. The question of enhanced security practice for cloud based CBM practice is
one explored by Tedeschi et al. (2017) who propose a structured approach to the assessment
of security requirements within a cloud based CBM system. In more recent work Bowden et
al. (2019) propose a ‘plug and Play’ end to end cloud architecture for predictive maintenance.
The architecture utilises Docker containers (an open source software that is used to ‘wrap’ up
unites of code into generically compatible container compliant with common software as a
service and platform as a service implementations) to provide flexibility in the implementation
and deployment of the completed analytics system. A case study is provided by these authors,
based on the monitoring of a Comau industrial robot, with initial results demonstrating a range
of predictive and near to real time alerting functionalities expected of a future industry 4.0
maintenance system (Bowden et al., 2019). Predictive maintenance architectures such as
Bowden et al. (2019) often rely on data streams provided by IoT compliant sensing packs
composed sensors and Edge data processing devices. It is the opinion of certain research
works that there are limitations in the utilisation of Cloud platforms for industrial applications
in that the sheer volume of data transfer that must take place between facility and Cloud
infrastructure means that more localised processing is necessary (Anaya et al. 2018). Patel et
al. (2017) also acknowledge the data transfer limitation of Cloud platforms along with the need
for near to real time processing of data, a factor that is also difficult to achieve in such
platforms.
In such circumstances the ability of Edge devices to pre-process data streams emanating from
production and machine tool sensors takes on additional importance; as such decentralised
processing of data using Edge devices is an active stream of research within Industrial IoT
analytics programs to support maintenance activities (Uhlmann et al. 2017). Parpala & Iacob
(2017) describe how IoT enabled Edge technologies can be used to allow data collection from
legacy machinery. This work also demonstrates a simple data communication interface to
complement the sensor and Edge device hardware implementations required (Parpala &
Iacob, 2017). Jantunen et al. (2018) provide a case study drawn from research of a proactive
maintenance approach within a power plant. This work examined the output of vibration
sensors monitoring flue gas blowers within a power plant; the research concluded that a six
month time difference between component replacement times suggested by use of this
approach and manual assessments of the same data (Jantunen et al., 2018). A wider
exploration of Edge computing in the manufacturing domain is provided by Wan et al. (2018)
who propose an architecture for IoT enabled production. In their maintenance based case
study Wan et al. (2018) the authors found that the packing of confectionary boxes by robots
could be performed autonomously with self-organisation and planning undertaken at the
production line level, made possible by the inherent advantages of co-located processing
provided by Edge computing devices. The connection and synergies achievable with the
combined use of IoT, predictive maintenance and 3D printing are elicited by Yamato et al.
(2017a). Such linkages are explored with regard to aircraft maintenance, and an analytics
platform is proposed (Yamato et al., 2016) along with a case examining the potential of sound
stream analysis in maintenance utilising edge devices (Yamato et al., 2017). In terms of
machine learning use with data streams Tran and Yang (2012) propose a platform for CBM
utilising intelligent techniques such as Principal Component Analysis (PCA) and Support
Vector Machine (SVM) in particular to extract features from data and then diagnose faults in
rotating machinery, respectively. In further studies involving the maintenance practice relating
to rotating machinery, Yunusa-Kaltungo and Sinha (2017) make the case that while analysis
of big data obtainable from such equipment is potentially transformative, in the case of
vibration based parameters more streamlined techniques can hold the potential for lower cost
and simplified e-maintenance practice. These authors provide an approach utilising
classification and optimisation techniques for use in the monitoring of such machines (Yunusa-
Kaltungo and Sinha, 2017). Kanawaday and Sane (2017) explore the use of a forecasting
method, AutoRegressive Integrated Moving Average (ARIMA), on data streams generated by
IoT sensorised production line machinery. This approach has been used to improve
maintenance planning and in future research may be adapted to predict remaining useful life
of a production machine and detect operational anomalies (Kanawaday and Sane, 2017). The
concept of tele maintenance (remote maintenance) is highlighted by Selcuk (2018) as a future
direction for prognostic maintenance, made possible by IoT connected sensors, intelligent
products and machines. This author also points to the emergence of maintenance as a full
integrated service provided to customers, leveraged through IoT technology (Selcuk, 2018). It
is also the case that Digital Twins, providing virtual replicas of real world production lines and
assets, may be used in IoT (Koulamas and Kalogeras, 2018) for connected predictive
maintenance practice (Qi & Tao, 2018) and administered from both inside and outside the
customer organisation. A number of interesting new business models for IoT based service
provision are outlined by Ju et al. (2016). These authors propose a generic framework for the
enablement of IoT business model development (Ju et al., 2016). Khan et al. (2017) also
provide a methodology for IoT sensing in industry, illustrated by a use case based on a process
to facilitate predictive maintenance within an organisation. It is clear that localised processing
of data streams can provide real benefits in terms of real time decision making and the
enablement of intelligent automation; for an additional commentary on the mining of streaming
data for maintenance activities Munir et al. (2018) provide a concise summary. At this point it
should be noted that newer IoT enabled maintenance techniques are perhaps not a complete
replacement for existing techniques such as root cause failure analysis and preventative
maintenance practice; existing and new techniques can be complimentary in their use, a point
made by Bengtsson and Lundstrom (2018).
2.3.1 Data fusion from multiple sensor outputs
The IoT opens up disparate physical plants and machinery to the potential for ubiquitous and
real time data connectivity. While much work still remains to be completed on the
establishment of unified data exchange standards and semantics progress has been made in
terms of data networking and management approaches for this recent paradigm shift in
connectivity.
2.3.1.1 Large scale data internetworking
Emmanoulidis et al. (2009) make the case for the take up of advanced communication
networks in conjunction with mobile computing solutions in order to support maintenance
activities. A reliance on locally available data and resources provided by LAN (local Area
Network) often means that organisations must undertake manual data mining tasks on
disconnected data sets in order to make planning decisions on maintenance activities
(Emmanoulidis et al., 2009). Sayafar et al. (2016) add that the real time optimisation of
maintenance activity planning, in part enabled through mobile networked devices, will lead to
universal access to vital asset data for involved workers. The production of data by intelligent
products provides another IoT enabled source of data. Intelligent products may produce data
while in operation ‘in the field’ or even while in production while being assembled in a factory.
McFarlane et al. (2012) investigate the state of the art in intelligent products point to the use
of RFID (Radio-frequency identification) tags to trace products through the supply chain and
also note the rise of IOT and its potential to network connect intelligent products. Cuthbert et
al. (2016) make the case for product intelligence in domestic appliances suggesting that low
cost electronics could be integrated into such products to enable health tracking for
maintenance purposes. Improvements in communications networks especially mobile
networks are helping to leverage interest in IOT. The 5G mobile standard promises
bandwidths capable of serving the requirements for the wireless connection of IOT devices
with greater energy efficiency (Andrews et al., 2014). Papakostas et al. (2016) outlines 5G in
a manufacturing context pointing to the possibility for ubiquitous connectivity and potential for
plug and play hardware on the shop floor.
2.3.1.2 Large scale data management and analytics
The volume of data sets and streams available with networked hardware in manufacturing
leads to changes in the way that data analysis takes place. Cloud technologies have been
assessed for this purpose and the concept of Cloud Manufacturing has been put forward as a
potential analytics solution. The Cloud Manufacturing paradigm is based on the use of
distributed Cloud Computing technologies for sustainable manufacturing while integrating
distributed Internet technologies such as IoT (Zhang et al., 2014). Sustainability in
maintenance practice is a theme explored by Franciosi et al. (2018) who surveyed literature
and found that proactive and predictive maintenance practices could lead to reduced
environmental impact in many cases, noting that improved end of life estimation and failure
modes that take account of emission/environmental damage due to machine breakdown hold
much potential. The distributed processing of data envisaged by Cloud Manufacturing is one
way to address the analytics need created by the challenge of continuous maintenance
particularly of high value long lifecycle products (Roy et al., 2016). Truong (2018) provide a
predictive analytics approach for maintenance utilising IoT and Big Data Cloud resources.
More efficient methods of maintenance are required as many high value products are sold as
product-service offerings whereby maintenance is delivered as part of the retail offer (Baines
et al., 2009). Cyber Physical Systems (CPS) in manufacturing are entities that both produce
and consume vast quantities of data in their operation. While encompassing such entities as
cybernetic extensions to humans CPS systems in terms of manufacturing are more likely to
be formed of the following components: production capable machines; sensing functionality
(both hardware and software); intelligent computer processing functionality. With CPS there
is a need for both local data processing (within the CPS hardware entities), for autonomous
operation within a shop floor perhaps, and remote analytics for monitoring and global
coordination. Gubbi et al. (2013) notes that both Cloud and IoT technologies are required to
fully enable CPS and link together intelligence at both local and global levels; it is this, in the
authors’ of this paper opinion, that will help to deliver the next generation of manufacturing
solutions including those focussed at the maintenance level. While realising interconnectivity
at a hardware and digital communications level is key to the latest maintenance practice an
often overlooked concept, though one that is gaining in acceptance, is that of context
awareness in relation to maintenance.
2.4 Context aware computing
The area of context awareness in systems has been growing over recent years. Dey et al.
(2001) describe the context awareness of systems as being provided through the intelligent
characterisation and interaction of computer applications with their surrounding environment.
In the view of Dey et al. (2001) context awareness acquisition by a system may be manual as
well as automatic. The research of Bettini et al. (2010) advises that applications should be
abstracted away from the context related functionality that they utilise, meaning that changes
in context data and models should not break the software systems built upon them. To this
end Bettini et al. (2010) survey the field of context modelling with the aim of identifying good
practice in order to reduce the complexity of context aware application development;
recommendations are also made as to the use of formal modelling techniques such as Object-
Role Modelling (ORM) in the development of context models. Blasch et al. (2012) highlights
the importance of context in the development of a data fusion architecture noting that the use
of technologies such as simulation in combination with context based information can provide
further efficiencies at the analysis stage. One key driver for increased interest in context
awareness is the rise of pervasive computing. Ye et al. (2012) identify pervasive computing
as a type of computing that through the use of sensors is able to interact with the world with
minimal or no human intervention. Perera et al. (2014) provide an overview and a taxonomy
of a lifecycle approach to context awareness for utilisation in conjunction with IoT linked
middleware, highlighting a progression from context acquisition through processing and
analysis to its eventual distribution though API (Application Programming Interfaces) and
appropriate data formats.
Hong et al. (2009) provide a classification framework for context aware systems in their survey
of the area identifying five distinct layers: Concept and research layer; Network layer;
Middleware layer; Application layer; User intrastate layer. Focusing on Enterprise Information
Systems, El-Kadiri at al. (2016) argue that the multi-networked nature of physical entity
supported IoT empowers physical products and assets to become intelligent; but in order to
cope with the breadth, depth, rate, and sheer volume of produced data a context aware
approach is needed. El-Kadiri at al. (2016) identify abstract context categories (e.g. user,
environment, system, service and social context) as relevant to a wide range of applications
but they indicate that such high-level context abstraction needs to be supplemented by
domain-specific context modelling, providing examples relevant to maintenance and asset
management.
Of most interest to this paper are the application and middleware layers. While benefitting from
the other layers the application and middleware layers contain the logic required to establish
context and provide intelligent processing and presentation of data to the user while holding
the potential as a platform for automated decision making. In the research put forward by
Perera et al. (2014) these authors also sought to examine appropriate data collection
frequency levels and establish responsible components for data collection and decision
making within context aware IoT systems. The importance of context in data fusion is
highlighted by a number of authors (Khaleghi et al. 2013, Fernández-de-Alba et al. 2015,
Snidaro et al. 2015, Linas et al. 2016). In particular the fusion of context related data is
interesting in the work of Fernández-de-Alba et al. (2015) who put forward a framework to
combine senor data from different sensors and platforms; the framework is demonstrated
through a case study that guides users to meetings within an organisation. An important
facilitator of context aware computing is semantic technology.
The survey of Snidaro et al. (2015) underlines the increasing role of machine learning in the
analysis and use of context based data, in addition these authors also note the need to provide
context processing functionality as part of a shared middleware layer that applications utilise
for information processing. Linas et al. (2016) go onto formalise the roles of context and
information fusion in their combined use. These authors also promote the JDL (Joint Directors
of Laboratories) data fusion model which defines five fusion levels their roles in applications
and the algorithmic approaches associated in their realisation. Smirnov et al. (2015) introduce
a number of context based knowledge fusion patterns. The seven patterns aim to encapsulate
the different context based effects that occur in decision support systems when integrating
new knowledge and changes in semantic mappings to related ontologies. The role of ontology
in the collection and analysis of context related data has been researched by a number of
authors (Perera et al. 2014; Sminov et al. 2015, Sminov et al. 2016, Linas et al. 2016). From
literature it is clear that semantics and metadata descriptions will play a significant part in the
development of context awareness, as raised by Sminov et al. (2015) the ability to arrive at
and distribute processed data with a recognised shared meaning will be key. Perera et al.
(2014) identify six wider research challenge areas for context aware computing for IoT that in
the opinion of the authors of this paper also relate to IoT enabled maintenance applications:
1. Automated configuration of Sensors
2. Context discovery
3. Acquisition, modelling, reasoning and distribution
4. Selection of sensors in sensing as a service model
5. Security, privacy and trust
6. Context sharing
Establishing and describing context for data is an important subject for the further development
of maintenance practice, though perhaps it still lacks a holistic containing formalisation to
enable its universal take-up in industry. It is the opinion of the authors of this paper that
underlying many of the aforementioned IoT, Context, and prognostic maintenance research
challenges outlined by Perera et al. (2014) it is perhaps that there is a need for an audit trail
framework to be applied to data collection and semantic description methods, particularly for
its use with maintenance applications taking into account their transactional nature and need
for integrated scheduling.
2.5 The audit trail for maintenance
The quality and provenance of data are important factors in data management and a key
success factor for engaging in any form of analytics. With maintenance rapidly adopting key
Industry 4.0 technologies, such issues attain increased important for successful applications
and services. Product and asset lifecycle data are increasingly acknowledge as a valuable
asset (Kubler et al., 2015). Therefore their own lifecycle needs to be appropriately managed
and this could become a key factor in establishing a credible audit trail for maintenance
activities and data. Lin et al. (2007) conducted a survey into data quality relating to asset
management information. The survey found that processes and software for asset related data
quality management were missing in a majority of organisations interviewed; in addition
organisations did not have a strategy in place regarding data quality (Lin et al., 2007). In a
review of standards relating to Asset Management, Koronios et al. (2007) noted the increasing
use of XML (eXtensible Markup Language) as a data description standard along with OPC UA
(OPC Unified Architecture) for industrial system intercommunication. The OPC UA standard,
while comprehensive in its specification, can be complex and expensive for an organisation to
implement. The work of Henßen and Schleipen (2014) examines the role that the
AutomationML mark-up language can play in simplifying the use of OPC UA models with
existing data sets and streams expressed in XML. According to Henßen and Schleipen (2014)
use of OPC UA directly is a complex task, utilising AutomationML mapping to OPC UA opens
up the opportunity of streamlined connectivity with OPC UA compliant systems and
manufacturing systems. Liyanage et al. (2009) mention the semantic web, ontology and use
of XML metadata descriptions for information exchange in e-maintenance. Grangel-Gonzalez
et al. (2016) take the semantic communication notion a step further by producing a metadata
software shell for Industry 4.0 components. The approach is based on RDF (Resource
Description Framework) and OWL (Web Ontology Language) and allows for new functionality,
described by ontological elements, to be integrated into the communication framework with
minimum disruption (Grangel-Gonzalez et al., 2016). In combination with machine intelligence
such a framework could acts as an enabling protocol for automation efforts in maintenance
activities and factory operations alike.
Many enterprise systems in organisations, such as ERP (Enterprise Resource Planning),
possess event logging capabilities. Such event logs may be mined in order to reconstruct a
chain of activities that have taken place within the organisation and administrated by the
system (Tiwari et al. 2008, IEEE Task Force on Process Mining 2011, Turner et al. 2012) and
then further analysed by automated techniques to provide optimised processes (Tiwari et al.
2010, Vergidis et al. 2015). Similar event logging based audit trails have been utilised in the
field of cyber threat detection within networked software systems. Bass (2002) details efforts
made in the development of intrusion detection systems utilising a data fusion approach. In
this work Bass (2002) highlights the use of pattern detection utilising templates. In later
research Vaughn et al. (2005) examine the possibility for automated cyber vulnerability
recognition where sensor data is used to trigger security warnings. The aim of automated
cyber security is also sought by Abreu et al. (2015) with the use of audit trail data. With this
work Abreu et al. (2015) and others such as Nehinbe (2014) employ machine learning
techniques to derive patterns and insights to, in principle, enable automated actions and
decisions to be made. Duncan and Whittington (2016) advise on the regular analysis of audit
trails in the effective securing of Cloud based systems. While useful in countering intrusions
into maintenance systems it is also the case that such approaches provide much of the rigor
and data management practise required to ensure quality and enforce standards within an
organisation and its supply chain and linked parties. The use of such audit trail techniques in
manufacturing has been much less evident though its use with IoT has in outline been
explored by Lomotey et al. (2018) in research exploring the need for visualisation of Internet
connected devices. In addition Lomotey et al. (2018) propose a provenance methodology to
allow for improved traceability and identification of routes through a network that specific data
points may take. Efforts towards a unified metadata syntax and model for provenance are
embodied in the work of Moreau et al. (2011) who put forward the Open Provenance Model
(OPM), enabling the unified and secure exchange of such data between networked systems
and entities. Park et al. (2011) also explore issues surrounding the location of provenance
data for an entity (local vs global storage) and the rights of access to the provenance data by
other network connected entities.
Use has been made of audit type data in industrial applications. A sensor fusion approach has
been used by Payan et al. (2016) in the development of proactive safety metrics for
Helicopters. In this research Payan et al. (2016) fuse the outputs of flight data monitoring to
form the basis for predictive safety measures, with the potential to advise preventative actions.
Such an approach may also inform the development of audit trail compilation and use to
enhance the scheduling and performance of maintenance activities. An approach to
combining multi sensor data has been put forward in the information fusion technique of Basir
and Yuan (2007), who utilise Dempster-Shafer theory evidence theory with an industrial case
based on engine testing on an automotive production line. Basir and Yuan (2007) found that
their approach was able to successfully address decision conflicts pertaining to engine fault
diagnosis with an improved level of accuracy.
With the use of such audit trail based intelligent data mining there arises the potential need to
explain the reasoning behind automated decisions to humans for the purposes of
evaluating/ensuring provenance of maintenance data. Duncan and Whittington (2016) make
a number of recommendations on how the audit trail for Cloud computing could be improved;
the following are an adaptation of a subset of those recommendations with relevance to the
maintenance field:

• Insufficient logging of data within Cloud environments and manufacturing systems,

data logging is not set to ‘on’ by default
• A proper regime of data log migration to data storage is required
• Further understanding on information flow within a manufacturing system is required
• Enhanced data security is required to safeguard collected audit trail data and digital
entry points to manufacturing systems from cyber attackers
It is the case that a ‘human in the loop’ is also required as their expert knowledge and overview
capability can be leveraged, in particular, to help ensure data and process security. A vital
step along the road to automation is the inclusion of human expertise along with standards
such as the MIMOSA open system architectures for CBM and EAI (Enterprise Application
Integration) (MIMOSA, 2017), which potentially provide a wider underlying structure for the
concept of maintenance audit trails.
Figure 1 illustrates the concept of the audit trail with an example drawn from railway
maintenance activities. In Figure 1 it can be seen that for a section of track there are a range
of maintenance activities that may involve: maintenance workers, feeding back reports via
mobile devices; rail maintenance vehicles with sensors; passenger trains fitted with track and
infrastructure monitoring sensors. In addition a number of trackside sensors may also stream
back data to a control centre concerning a range of environment specific parameters. IoT hubs
may be co-located with trackside equipment and within train vehicles. Edge devices on
standard passenger services may be linked to sensors and process and store data for
forwarding to the IoT hub (allowing for when the passenger service may be out of
communication range with the advantage of possibly reducing the amount of data to be
communication due to built-in intelligent processing and filtering stage). The OPC/UA (Open
Platform Communications – Universal Architecture) standard and message queuing telemetry
transport protocol (MQTT) would provide the data transfer format to and from IoT hubs.
 Figure 1: An audit trail drawn from rail maintenance activities and sensor streams for a
section of track
The scenario depicted in Figure 1 relates to the possibility that sensors have registered faults
with a Balise (track based forming part of an automatic train protection (ATP) system) and
trackside signals in a period of time after the section of track has been tamped (where the
ballast bed of the track is adjusted). In addition a bankside sensor has noted some occasional
subsidence in the past. All these data streams are recorded at a central control centre. The
use of data mining may establish a causal link between these events taking into account the
outlier measurement from the bankside sensor leading to the root cause of the fault. The audit
trail establishes the order of events via timestamps and the output from data mining/machine
learning. Such audit trails once established can help in the decision making and may also
advise trackside workers, undertaking maintenance in future scheduled activities, to make
additional checks based on the history of the track section.

3. Related Work and Discussion

It is clear that Industry 4.0 enabling technologies are changing attitudes towards digital
connectivity and automation in manufacturing, though it is also the case that there is a need
for a holistic understanding of the use of machine intelligence in the achievement of automated
and autonomous manufacturing visions of the near future. It is also clear from this review that
standardised data collection processes and intelligent analysis techniques are the subject of
current investigation by many research groups around the world. As part of this review, using
the search tool Scopus, it was possible to identify the amount of papers published in the period
2000 to 2018. Of interest were the findings for papers published involving the subject of E-
maintenance and that of predictive maintenance. Figure 2 shows that predictive maintenance
papers have shown a gradual increase of the period peaking in 2018. For the same time period
Figure 3 shows that E-Maintenance papers peaked in 2010 and then have stabilised at around
15-23 publications per year for the most recent 5 year period. From this review it was possible
to identify a number of works that best typify the sub areas highlighted in this paper. Table 1
summarises these papers in terms of intelligent decision challenges and approaches taken.
From Table 1 it is clear that there is a wide range of potential solutions and approaches to
intelligent maintenance, though it is the authors’ opinion this field would benefit from clear
processes to support audit trail style collection of data and clear framework for its processing,
analysis and use. There will be a necessity to capture and store data streams from production
machinery and the audit trails of decision making within Semantic technologies may provide
a way of describing maintenance data so it may be shared across the manufacturing
enterprise and potentially within the supply chain. The case for a metadata layer capable of
semantically describing maintenance data is made in part by Pistofidis et al. (2016) who put
forward a methodology for maintenance metadata management involving the incorporation of
expert knowledge. Such technology may be the cornerstone of evolving context awareness in
maintenance systems to enable automated decision making and scheduling for maintenance
activities.

Predictive Maintenance Papers Since 2000

1200

1000

800

600

400

200

Figure 2: Papers published involving predictive maintenance between 2000 – 2018 (Source:
Scopus)

E-Maintenance Papers Since 2000

50
45
40
35
30
25
20
15
10
5
0

Figure 3: Papers published involving the subject of e-maintenance between 2000 – 2018
(Source: Scopus)
It is clear that machine learning will have a significant role to play in the delivery of future
automated and intelligently supported maintenance decision making systems. Predictive
maintenance programmes will increasingly rely on machine learning techniques in order to
deliver proactive and dynamic maintenance plans. The extension of the manufacturing
enterprise to digitally link with its supply chain has gained another potential benefit in the ability
to order spare parts in advance of potential breakdowns, when predictive forecasting
techniques are employed.
It is clear that industry is still missing an overall framework for digital maintenance. Advances
in sensors and sensor fusion techniques have run-ahead of suitable processes and systems
capable of fully harnessing their outputs. This is evident in the introduction of Cyber Physical
Systems (CPS) within manufacturing, though progress is being made in the area of worker
interactions with shop floor machinery. Both simulation and visualisation technologies,
especially Mixed Reality, provide a new platform for enhanced machine assistance for human
engineers and raise the potential for maintenance related Cobot development. Much research
has concentrated on the further development and use of more readily available data streams
such as those provided by SCADA systems. Increasingly though papers may be found
exploring the role IoT can play in the provision of data streams from both new and even legacy
equipment. This move is especially evident in the area of Condition Based Maintenance (CBM)
and predictive approaches. Utilisation of Digital Twin systems to replicate industrial production
assets are now being introduced; a significant focus in Digital Twin development is that of
health monitoring and prediction of maintenance needs/breakdowns with many organisations
adopting this virtual representation for the specific goal of increased uptime. For many
organisations the use of existing systems in combination with new sensor technologies and
software will provide many of the potential advantages promised by e-maintenance and
visions such as Industry 4.0.

Table 1: Intelligent Decision Support Challenges and Approaches

Publication Approach outline Problem targeted

Prognostic maintenance

de Novaes et al. (2017) A review of prognostic Prognostic maintenance

techniques relating to practice relating to remote
remote field maintenance of field maintenance of wind
wind turbines including turbines
discussion of Hidden
Markov Model, Neural
Networks and use of
SCADA data.
Djurdjanovic et al. (2003) Describes ‘Watchdog agent’ Product performance
capable of sensor degradation prognostics
assessment and prediction
of machine performance.
Utilises Neural Network,
Hidden Markov Model,
Particle Filter
Fumagalli and Macchi E-maintenance platform E-maintenance platform for
(2015) utilising web services rapid integration of
approach and statistical maintenance activities
techniques.
Katsouros et al. (2013) Bayesian approach for Intelligent decision
maintenance action support/recommendation
recommendation based on
historical cases
Kiritsis (2011), Closed loop product Intelligent products and
lifecycle approach for product service model and
Intelligent products data usage for improved
Definition, Semantic model product lifecycle including
and ontology for data maintenance practice.
interchange respecting IoT
connectivity possibilities.
Lee et al. (2006) Tools for prognostics Machine health prognostics
utilising Self-Organizing
Map (SOM),Hidden Markov
Model, illustrated with case
studies
Lee et al. (2014) A position paper and case Machine health prognostics
study detailing. Discussion
of Self-Organizing Map
(SOM), Gaussian Mixture
Model
(GMM), Bayesian Belief
Network (BBN), Fuzzy Logic
for maintenance
Leitao et al. (2016) A survey of Smart Agent Intelligent Maintenance
use in combination with Systems and CPS
machine intelligence within Integration
manufacturing and
maintenance
Leite et al. (2017) Holistic prognostics to Maintenance Prognostics
support maintenance of
wind turbines utilising an
approach incorporating
Neural Network and
Particle Filter (maintaining
assets located in
inaccessible/remote sites)
Liyanage et al. (2009) An outline of an integrated Framework for E-
e-maintenance approach maintenance with prognostic
and description of the maintenance application.
Watchdog (Agent-Based
Real-time Remote
Machinery Prognostics and
Health
Management) toolbox
application. Utilisation of
Software agent and custom
algorithms.
McNaught and Zagorecki Development of a model for Prognostic modelling for
(2009), Prognostic maintenance reliability and maintenance
decision making and policy policy experimentation
evaluation utilising Bayesian
Networks
Niu and Jiang (2017) Custom algorithm for Health Prognostics and
dynamic maintenance maintenance optimisation
incorporating both local
health prognostics for
combination with global
optimisation approach
Papathanassiou et al. E-learning approach to E-learning for maintenance
(2013) support maintenance management
management proposing a
toolkit for training delivery.
Peng et al. (2010) Classification of common Machine health prognostics
prognostic models for – residual useful life
maintenance. Discussion of
neural network, Bayesian-
related methods, hidden
Markov models for
maintenance.
Selcuk (2018) Prognostic maintenance, Maintenance as an IoT
made possible by IoT enabled service
connected sensors,
intelligent products and
machines. The emergence
of maintenance as a full
integrated service provided
to customers, leveraged
through IoT technology.
Vafaei et al. (2019) Fuzzy inferencing for rule Maintenance decision
set composition and support for CBM.
maintenance scenario
generation.
Voisin et al. (2010) Custom algorithm for Generic prognostic
proactive maintenance, approach and methodology
framework for ‘predict and for maintenance decision
prevent ‘. support
Xia et al. (2018) Concise summary of Predictive maintenance
predictive techniques in use review of techniques
for maintenance practice in
a range of digital
manufacturing practices.
Yunusa-kaltungo and Sinha Big data and optimisation Classification and
(2017) techniques for simplified e- optimisation technique for
maintenance practice; maintenance prognostics
utilisation of classification
and optimisation for
production machine
monitoring.
Zhou et al. (2017) Custom algorithm statistic Performance Degradation
for performance degradation Prognostics
forecasting of power
 machinery based on historic
and monitored datasets

Remaining useful life prediction

Belkacem et al. (2017) Prognostic and diagnostic Remaining useful life

architecture for industrial prognostics and diagnostics
system maintenance using a
custom approach based on
hybrid automata modelling.
Desforges et al. (2017) Prognostic function for Proposes co-operative
maintenance planning planning between
support. Proposes production and maintenance
extensions to Object to promote their
Oriented Bayesian Networks synchronisation.
for Remaining Useful Life
calculation
Fan et al. (2015) Remaining Useful Life Remaining useful life
prediction and prognostics prediction and prognostics
approach for LED lights
utilising a particle filter
approach based on
Sequential Monte Carlo
(SMC) and Bayesian
techniques.
Galar et al. (2015) Residual useful life Residual useful life
predicted through data prediction for condition
aggregation and context based maintenance
awareness using fuzzy
approaches and clustering
Kanawaday and Sane Use of the forecasting Forecasting method utilising
(2017) method, AutoRegressive IoT data streams, with
Integrated Moving Average application to RUL.
(ARIMA), on data streams
generated by IoT sensorised
production line machinery.
Used to improve
maintenance planning with
possible future application to
RUL.
Marugan and Marquez Approach for the monitoring Diagnostic and prognostic
(2015) of rail track points utilising maintenance decision
binary decision diagram and making
fault tree analysis. An online
decision making system
based on this research is
envisaged as a next stage.
McNaught and Zagorecki Development of a model for Prognostic modelling for
(2009) Prognostic maintenance reliability and maintenance
decision making and policy policy experimentation
evaluation modelling with
utilising Bayesian Networks
Niknam et al. (2015) Prognostic based approach Residual life prognostics
for maintenance decision approach
 making based on remaining
life prediction using a
custom algorithm
Ragab et al. (2017) Pattern based approach Prognostic technique for
utilising a logical analysis of Condition based
data (LAD) and Kaplan– maintenance
Meier (KM) estimator;
provides an estimated
reliability curve for a given
monitored asset.
Van Horenbeek and Dynamic maintenance Prognostic approach for
Pintelon (2013) policy development based component lifetime
on prognostics using a extension
custom algorithm and
degradation model.
Zhou et al. (2017 ) Combination of forecasting Degradation trend
and statistic based method prognostics and fault
to provide a combined time diagnosis
and condition based
maintenance approach.

Intelligent products and assets

Barbosa et al. (2016) Envisages the combination Intelligent products in use

of both intelligent products data combined with Cyber
with Cyber Physical systems Physical Systems (CPS) for
(CPS), outlining the prognostic maintenance.
additional combined use of
agent based systems to
provide prognostic
maintenance and decision
making.
Brintrup et al. 2011 Intelligent agent platform Autonomous intelligent
based on a web service products in-use data
approach to leverage in field harvesting with maintenance
assets data generation. application.
Cuthbert et al. (2016) Framework for domestic Intelligent product data use
products health monitoring and maintenance with self-
describing the potential for repairing assets
automated self-repair of
assets based. A future
research challenge section
is set out by this paper
Dhall and Solanki (2017) IoT based predictive car Predictive Car maintenance
maintenance where the scheduling
connected car is able to
communicate health data so
that service scheduling is
streamlined.
Katsouros, et al. (2015) Embedded event detection Self-aware assets and
within intelligent products analysis of their generated
though feature extraction data streams
and statistical time series
data analysis.
Kiritsis (2011) Closed loop product Intelligent products and
lifecycle approach for product service model and
Intelligent products data usage for improved
Definition, Semantic model maintenance practice and
and ontology for data product lifecycle
interchange respecting IoT management.
connectivity possibilities.
McFarlane et al. (2013), A review of intelligent Review of intelligent
product and RFID tag data products field including
use. RFID technologies for active
participation of products in
their manufacture and in
field use
Meyer et al. (2009) A review of intelligent Intelligent products in-use
products giving reference to data and system interaction
maintenance practice that and interconnectivity, data
may be provision potential for
envisaged/leveraged prognostics and
through the paradigm. maintenance decision
Discussion of agent based support systems.
approach for potential
combination with machine
intelligence algorithms is
provided.
Parpala & Iacob (2017) Utilisation of IoT hardware IoT connectivity to legacy
and protocols to link legacy machine tools with software
machine tools to internet interface and edge
services for the provision of hardware implementation.
automated maintenance and
status alerts to production
line operators.
Selcuk (2018) Prognostic maintenance, Maintenance as an IoT
made possible by IoT enabled service.
connected sensors,
intelligent products and
machines. The emergence
of maintenance as a full
integrated service provided
to customers, leveraged
through IoT technology.
Truong (2018) Predictive analytics Analytics framework for IoT
approach for maintenance connectivity with cloud
utilising IoT and Big Data resources and big data for
Cloud resources. maintenance.
Wuest et al. (2018) Analysis of work and expert Future intelligent products
opinion synthesis regarding are projected to move
the present and future of beyond providing
intelligent products. information or decision
support by becoming more
proactive.

IoTStreaming Data and Intelligent Decision making

Accorsi et al. (2017) Model set to aid the Framework for data mining
prediction of faults in and modelling related to
production systems, CBM decision making.
framework for data mining
and modelling related to
condition based
maintenance.

Barbosa et al. (2016), Envisages the combination Integration of Cyber

of both intelligent products Physical Systems and
with Cyber Physical systems Intelligent Products for
(CPS), outlining the improved decision making in
additional combined use of maintenance practice
agent based systems to
provide prognostic
maintenance and decision
making.
Bowden et al. (2019) A ‘plug and Play’ end to end IoT enabled predictive
cloud architecture for maintenance utilising IoT
predictive maintenance, Edge devices and a cloud
utilising IoT Edge platform.
processing and open source
Docker containers. Case
study based on industrial
robot autonomous
operation.
Chen et al. (2018) IoT Edge computing for IoT Edge computing
manufacturing based architecture for
maintenance practice, case manufacturing based
study and architecture. maintenance practice.
Jantunen et al. (2018) IoT based framework for Industry 4.0 compliant IoT
maintenance, detailing a based framework for
case study drawn from maintenance.
research of a proactive
maintenance approach
within a power plant.
Karim et al. (2016) Knowledge discovery Knowledge discovery
framework for cloud based framework for cloud based
maintenance analytics maintenance analytics
platforms with a rail industry platforms
case study
Katsouros, et al. (2013) A Bayesian approach to Intelligent decision support
maintenance action for maintenance actions.
recommendation. The
technique utilises past
maintenance event data in
its classification of problem
types and probability
estimation.
Kiritsis (2011), Closed loop product Intelligent products and
lifecycle approach for product service model and
Intelligent products data usage for improved
Definition, Semantic model product lifecycle including
and ontology for data maintenance practice.
 interchange respecting IoT
connectivity possibilities.
Liyanage et al. (2009) An outline of an integrated Framework for E-
e-maintenance approach maintenance and with CBM
and description of the prognostic maintenance
Watchdog (Agent-Based application.
Real-time Remote
Machinery Prognostics and
Health
Management) toolbox
application. Utilisation of
Software agent and, custom
algorithms.
Mattila et al. (2016), Agent based approach to A distributed agent based
decision making utilising information architecture for
data from intelligent decision making utilising
products respecting the intelligent product data.
product life cycle.
Meyer et al. (2009), A review of intelligent Intelligent products in- use
products giving reference to data and system interaction
maintenance practice that and interconnectivity, data
may be provision potential for
envisaged/leveraged prognostics and
through the paradigm. maintenance decision
Discussion of agent based support systems.
approach for potential
combination with machine
intelligence algorithms is
provided.
Trappey et al. (2016), Underlying standards A review of standards
supporting IoT technology supporting IoT use in
with relevance to industry with relevance to
maintenance systems intelligent maintenance
including data linkages systems.
between hardware and
software systems for the
realisation of intelligent
maintenance practice.
Uhlmann et al. (2017) Maintenance analytics Maintenance analytics
approaches for approaches for
decentralised IoT Edge decentralised IoT Edge
platforms; Edge devices platforms.
pre-processing data streams
emanating from production
and machine tool sensors.

Data description and visualisation

Botelho et al. (2014) Maintenance skill capture Method for improved

for use in intelligent Intelligent decision making
maintenance systems for for maintenance with
improved automated inclusion of contextual data
decision making with regard represented by maintenance
to assets degradation
 monitoring using virtual skill capture from human
reality environment and operators.
simulation system.
Ceruti et al. (2019) AR based visualisation of Visualisation of spare parts
maintenance manual and maintenance
instructions and spare parts instructions using AR.
identification and
information provision within
an aviation maintenance
setting.
Fumagalli and Macchi E-maintenance platform E-maintenance platform for
(2015) utilising web services rapid integration of
approach and statistical maintenance activities
techniques.
Krempl et al. (2014) Mining of data streams open Open challenges for mining
challenges highlighting the of data streams provided by
potential role of sensors.
classification and machine
learning algorithms for pre-
processing of streamed
data.
Loukopoulos et al. (2017) Technique to impute Method to impute missing e-
missing values in data used maintenance data for use in
for maintenance prognostic prognostic maintenance
approaches using K Nearest systems.
Neighbour and Self-
Organising Map algorithms.
Nunez and Borsato (2017) Semantic framework Ontology based framework
expressed through OWL2 for data communication
language and between systems promoting
interoperability with P-prognostics for machine
SPARQL for decision maintenance.
making in prognostic
maintenance practice.
Penna et al. (2014) 3D visualisation and Interactive visualisation,
simulation tool using simulation and planning of
Augmented Reality to allow intelligent maintenance
for user editing of scenarios.
maintenance scenarios to
support intelligent
maintenance practice.
Turner et al. (2016) VR and AR visualisation and Visualisation of intelligent
DES simulation for potential maintenance decisions with
use in visualising the VR, AR and DES
decision making process for Simulation.
maintenance actions when
used in combination with
machine intelligence.
Zhong et al. (2017) Framework proposed for Mining RFID data for
mining and analytics of internet based intelligent
manufacturing shop floor manufacturing.
data utilising a custom
algorithm.
Conclusions
This paper has charted the evolution of intelligent decision support for maintenance practice
from Condition Based Maintenance (CBM) then prognostic use all the way to the e-
maintenance paradigm and the introduction of IoT and Cloud-enabled solutions. It is arguable
that the ability to digitally interconnect manufacturing plant and machinery provides many new
opportunities to raise productivity and efficiency within a production line and in itself leads to
a potential new era for intelligent maintenance though adoption of Industry 4.0 technologies.
Many challenges still remain in the provision of intelligent decision support for manufacturing
maintenance activities. While progress continues to be made in the area of prognostics for
maintenance and whole life considerations of manufacturing assets it is still the case that there
is a need for a holistic understanding of the use of machine intelligence in the achievement of
automated and autonomous manufacturing visions of the near future. The provision of
appropriate security measures for use in not just digital maintenance systems but throughout
the manufacturing organisation and its supply chain is a topic that will prompt much research
over the coming years. IoT technologies and their connectivity potential must be supported by
standards but also shared semantic descriptions, with OPC/UA (Open Platform
Communications – Universal Architecture) seen by many as a valid starting point in such
efforts. Further work remains to be completed on understanding data flows within
manufacturing and how digitisation of systems and information will impact maintenance
activities.
The quality and provenance of data are important factors in data management and a key
success factor for when engaging in any form of analytics. With maintenance rapidly adopting
key Industry 4.0 technologies, such issues attain increased importance in the delivery of
successful applications and services. It is put by the authors that clear processes to support
audit trail style collection of maintenance data and the provision of a comprehensive
framework for its processing, analysis and use should be important goals for the work that
must be completed in the near future for full enablement of digital maintenance practice. The
concept of ‘Human in the loop’ is also reinforced with the use of audit trails, allowing
streamlined access to decision making and the ability to mine decisions (and the reasoning
behind decisions for both machine assisted workers and managers). The ability to provide
procedural structure to data for reuse and communication within an Industry 4.0 maintenance
system will be vital for any future move towards semi or fully autonomous maintenance
activities.

References
Abreu, R., Bobrow, D.G., Eldardiry, H., Feldman, A., Hanley, J., Honda, T., de Kleer, J.,
Perez, A., Archer, D. and Burke, D., 2015. Diagnosing Advanced Persistent Threats: A
Position Paper. In DX@ Safeprocess, pp. 193-200.
Accorsi, R., Manzini, R., Pascarella, P., Patella, M. and Sassi, S., 2017. Data Mining and
Machine Learning for Condition-based Maintenance. Procedia Manufacturing, 11, pp.1153-
1161.
Adgar, A., Yau, A. and Addison, D., 2010. Smart tags. In E-maintenance, eds. (Holmberg, K.
et al.) Springer London. pp. 197-225
Akkermans, H., Zhu, Q., Fang, F., Lamper, L. and van de Kerkhof, R., 2019. Designing
Smart Services: A System Dynamics-Based Business Modeling Method for IoT-Enabled
Maintenance Services. In Proceedings of the 52nd Hawaii International Conference on
System Sciences.
Anaya, V., Fraile, F., Aguayo, A., García, O. and Ortiz, Á., 2018, Towards IoT Analytics. A
vf-OS Approach. In 2018 International Conference on Intelligent Systems (IS) (pp. 570-575).
IEEE.
Andrews, J.G., Buzzi, S., Choi, W., Hanly, S.V., Lozano, A., Soong, A.C. and Zhang, J.C.,
2014. What will 5G be ? IEEE Journal on selected areas in communications, 32(6), pp.1065-
1082.
Arnaiz, A., Iung, B., Adgar, A., Naks, T., Tohver, A., Tommingas, T. and Levrat, E., 2010.
Information and communication technologies within E-maintenance. In E-maintenance, eds.
(Holmberg, K. et al.) Springer London. pp. 39-60.
Azuma, R.T., 1997. A survey of augmented reality. Presence: Teleoperators & Virtual
Environments, 6(4), pp.355-385.
Azuma, R., Baillot, Y., Behringer, R., Feiner, S., Julier, S. and MacIntyre, B., 2001. Recent
advances in augmented reality. IEEE computer graphics and applications, 21(6), pp.34-47
Baines, T. S., Lightfoot, H. W., Benedettini, O., and Kay, J. M., 2009 ‘The servitization of
manufacturing: A review of literature and reflection on future challenges’, Journal of
Manufacturing Technology Management, Vol. 20, No. 5, pp. 547-567.
Bass, T., 2002. Multisensor Data Fusion for Next Generation Distributed Intrusion Detection
Systems. Ann Arbor, 1001, p.48113.
Basir, O. and Yuan, X., 2007. Engine fault diagnosis based on multi-sensor information
fusion using Dempster–Shafer evidence theory. Information Fusion, 8(4), pp.379-386.
Belkacem, L., Simeu-Abazi, Z., Dhouibi, H., Gascard, E. and Messaoud, H., 2017.
Diagnostic and prognostic of hybrid dynamic systems: Modeling and RUL evaluation for two
maintenance policies. Reliability Engineering & System Safety, 164, pp.98-109.
Ben‐Daya, M., Kumar, U. and Murthy, D.N., 2016. Computerized Maintenance Management
Systems and e‐Maintenance. Introduction to Maintenance Engineering: Modeling,
Optimization, and Management, pp.527-546.
Bengtsson, M. and Lundström, G., 2018. On the importance of combining “the new” with “the
old”–One important prerequisite for maintenance in Industry 4.0. Procedia Manufacturing,
25, pp.118-125.
Bettini, C., Brdiczka, O., Henricksen, K., Indulska, J., Nicklas, D., Ranganathan, A. and
Riboni, D., 2010. A survey of context modelling and reasoning techniques. Pervasive and
Mobile Computing, 6(2), pp.161-180.
Blasch E, Kessler O, Morrison J, Tangney J, White FE. Information fusion management and
enterprise processing. In Aerospace and Electronics Conference (NAECON), 2012 IEEE
National 2012 Jul 25 (pp. 204-211). IEEE.
Botelho, S.S., Rettberg, A., Duarte Filho, N., Hellingrath, B., Amaral, M., Espíndola, D.,
Pereira, C.E., Cordes, A.K., Ventura, R. and Frazzon, E., 2014. Towards intelligent
maintenance systems: rescuing human operator and context factors. IFAC Proceedings
Volumes, 47(3), pp.7110-7115.
Bousdekis, A., Magoutas, B., Apostolou, D. and Mentzas, G., 2015. A proactive decision
making framework for condition-based maintenance. Industrial Management & Data
Systems, 115(7), pp.1225-1250.
Bowden, D., Marguglio, A., Morabito, L., Napione, C., Panicucci, S., Nikolakis, N., Makris, S.,
Coppo, G., Andolina, S., Macii, A. and Macii, E., 2019. A cloud-to-edge architecture for
predictive analytics, Proceedings of the EDBT/ICDT 2019 Joint Conference, March 26, 2019,
Lisbon, Portugal.
Brintrup, A., McFarlane, D., Ranasinghe, D., Sánchez López, T. and Owens, K., 2011. Will
intelligent assets take off? Toward self-serving aircraft. IEEE Intelligent Systems, 26(3),
pp.66-75.
Campos, J., 2009. Development in the application of ICT in condition monitoring and
maintenance. Computers in Industry, 60(1), pp.1-20.
Ceruti, A., Marzocca, P., Liverani, A. and Bil, C., 2019. Maintenance in Aeronautics in an
Industry 4.0 Context: The Role of Augmented Reality and Additive Manufacturing. Journal of
Computational Design and Engineering, forthcoming.
Chen, B., Wan, J., Celesti, A., Li, D., Abbas, H. and Zhang, Q., 2018. Edge computing in
IoT-based manufacturing. IEEE Communications Magazine, 56(9), pp.103-109.
Chen F, Deng P, Wan J, Zhang D, Vasilakos AV, Rong X. 2015, Data mining for the internet
of things: literature review and challenges. International Journal of Distributed Sensor
Networks. Aug 30 2015.
Cuthbert, R., Giannikas, V., McFarlane, D. and Srinivasan, R., 2016. Repair Services for
Domestic Appliances. In Service Orientation in Holonic and Multi-Agent Manufacturing,
Springer International Publishing, pp. 31-39
da Silva, T.R. and Pereira, C.E., 2014. Building an Ontology for Intelligent Maintenance
Systems and Spare Parts Supply Chain Integration. IFAC Proceedings Volumes, 47(3),
pp.7843-7848
da Silva, T.R., Saalmann, P., Cordes, A.K., Giacomolli, A., Pereira, C.E. and Hellingrath, B.,
2014. Integration architecture of intelligent maintenance systems and spare parts supply
chain planning. Procedia CIRP, 25, pp.192-198.
Desforges, X., Diévart, M. and Archimède, B., 2017. A prognostic function for complex
systems to support production and maintenance co-operative planning based on an
extension of object oriented Bayesian networks. Computers in Industry, 86, pp.34-51.
Dey, A.K., Abowd, G.D. and Salber, D., 2001. A conceptual framework and a toolkit for
supporting the rapid prototyping of context-aware applications. Human-computer interaction,
16(2), pp.97-166.
Dhall, R. and Solanki, V., 2017. An IoT Based Predictive Connected Car Maintenance.
International Journal of Interactive Multimedia & Artificial Intelligence, 4(3).
Djurdjanovic, D., Lee, J. and Ni, J., 2003. Watchdog Agent—an infotronics-based
prognostics approach for product performance degradation assessment and prediction.
Advanced Engineering Informatics, 17(3-4), pp.109-125.
Duncan, R. A. K., & Whittington, M., 2016, Enhancing Cloud Security and Privacy: The
Power and the Weakness of the Audit Trail. In C. B. Westphall, Y. W. Lee, & S. Rass
(Eds.),CLOUD COMPUTING 2016 : The Seventh International Conference on Cloud
Computing, GRIDs, and Virtualization. (pp. 137). IARIA.
El Kadiri, S., Grabot, B., Thoben, K. D., Hribernik, K., Emmanouilidis, C., Von Cieminski, G.
and Kiritsis, D, 2016).Current trends on ICT technologies for enterprise information systems,
Computers in Industry. (79), pp. 14–33.
Emmanouilidis C, Jantunen E, MacIntyre J., 2006, Flexible software for condition monitoring,
incorporating novelty detection and diagnostics. Computers in Industry, V57, pp. 516–
527.Emmanouilidis, C., Liyanage, J.P. and Jantunen, E., (2009). Mobile solutions for
engineering asset and maintenance management. Journal of Quality in Maintenance
Engineering, 15(1), pp.92-105.
Espíndola D, Frazzon EM, Hellingrath B, Pereira CE. (2012) Integrating intelligent
maintenance systems and spare parts supply chains. IFAC Proceedings Volumes. 2012 May
23;45 (6):1017-22.
Fernández-De-Alba, J.M., Fuentes-Fernández, R. and Pavón, J., 2015. Architecture for
management and fusion of context information. Information Fusion, 21, pp.100-113.
Franciosi, C., Iung, B., Miranda, S. and Riemma, S., 2018. Maintenance for Sustainability in
the Industry 4.0 context: a Scoping Literature Review. IFAC-PapersOnLine, 51(11), pp.903-
908.
Fumagalli, L. and Macchi, M., 2015. Integrating maintenance within the production process
through a flexible E-maintenance platform. IFAC-PapersOnLine, 48(3), pp.1457-1462.
Galar, D., Thaduri, A., Catelani, M. and Ciani, L., 2015. Context awareness for maintenance
decision making: A diagnosis and prognosis approach. Measurement, 67, pp.137-150.
German Federal Government, The new High-Tech Strategy Innovations for Germany
[Online]. 2016, Available: https://www.bmbf.de/pub/HTS_Broschuere_eng.pdf accessed on
09/01/2017.
Goyal, D., Saini, A., Dhami, S.S. and Pabla, B.S., 2016, April. Intelligent predictive
maintenance of dynamic systems using condition monitoring and signal processing
techniques—A review. In Advances in Computing, Communication, & Automation
(ICACCA)(Spring), International Conference on (pp. 1-6). IEEE.
Gubbi, J., Buyya, R., Marusic, S. and Palaniswami, M., 2013. Internet of Things (IoT): A
vision, architectural elements, and future directions. Future generation computer systems,
29(7), pp.1645-1660.
IEEE Task Force on Process Mining (2011) Process Mining Manifesto. In F. Daniel, S.
Dustdar, and K. Barkaoui, (Eds.), BPM 2011 Workshops, Lecture Notes in Business
Information Processing (LNBIP), Vol 99, pp 169-194.
Henßen, R., and M. Schleipen, M., (2014) ‘Interoperability between OPC UA and
AutomationML’, Procedia CIRP, Vol. 25, pp.297-304.
Holgado, M., Macchi, M. and Fumagalli, L., 2016. Value-in-use of e-maintenance in service
provision: survey analysis and future research agenda. IFAC-PapersOnLine, 49(28), pp.138-
143.
Holmberg, K., Jantunen, E., Adgar, A., Mascolo, J., Arnaiz, A., and Mekid, S., 2010, E-
maintenance. Springer Science & Business Media, Berlin.
Hong, J.Y., Suh, E.H. and Kim, S.J., 2009. Context-aware systems: A literature review and
classification. Expert Systems with Applications, 36(4), pp.8509-8522.
Industrial Internet Consortium, 2017. The Industrial Internet of Things, Reference
architecture, Vol. G1:V1.80:20170131, Available at: http://www.iiconsortium.org/
Jantunen, E., Di Orio, G., Larrinaga, F., Becker, M., Albano, M. and Maló, P., 2018. A
framework for maintenance 4.0. In 2018 10th IMA International Conference on Modelling in
Industrial Maintenance and Reliability (MIMAR). Institute of Mathematics and its
Applications.
Jardine, A.K., Lin, D. and Banjevic, D., 2006. A review on machinery diagnostics and
prognostics implementing condition-based maintenance. Mechanical systems and signal
processing, 20(7), pp.1483-1510.
Ju, J., Kim, M.S. and Ahn, J.H., 2016. Prototyping business models for IoT service. Procedia
Computer Science, 91, pp.882-890.
Karim, R., Westerberg, J., Galar, D. and Kumar, U., 2016. Maintenance analytics–the new
know in maintenance. IFAC-PapersOnLine, 49(28), pp.214-219.
Khan, A., Pohl, M., Bosse, S., Hart, S.W. and Turowski, K., 2017. A Holistic View of the IoT
Process from Sensors to the Business Value. In IoTBDS (pp. 392-399).
Kanawaday, A. and Sane, A., 2017, November. Machine learning for predictive maintenance
of industrial machines using IoT sensor data. In 2017 8th IEEE International Conference on
Software Engineering and Service Science (ICSESS) (pp. 87-90). IEEE.
Katsouros, V., Koulamas, C., Fournaris, A.P. and Emmanouilidis, C., 2015. Embedded event
detection for self-aware and safer assets. IFAC-PapersOnLine, 48(21), pp.802-807.
Khaleghi, B., Khamis, A., Karray, F.O. and Razavi, S.N., 2013. Multisensor data fusion: A
review of the state-of-the-art. Information Fusion, 14(1), pp.28-44.
Kiritsis, D., 2011. Closed-loop PLM for intelligent products in the era of the Internet of things.
Computer-Aided Design, 43(5), pp.479-501.
Koulamas, C., and Kalogeras, A., (2018), Cyber-physical systems and digital twins in the
industrial internet of things, IEEE Computer, v51(11), pp. 95-98.
Koronios, A., Nastasie, D., Chanana, V. and Haider, A., 2007. Integration through
standards–an overview of international standards for engineering asset management. In
Fourth International Conference on Condition Monitoring, Harrogate, United Kingdom.
Krempl, G., Žliobaite, I., Brzeziński, D., Hüllermeier, E., Last, M., Lemaire, V., Noack, T.,
Shaker, A., Sievi, S., Spiliopoulou, M. and Stefanowski, J., 2014. Open challenges for data
stream mining research. ACM SIGKDD explorations newsletter, 16(1), pp.1-10.
Krommenacker, N., Lecuire, V., Salles, N., Katsikas, S., Giordamlis, C. and Emmanouilidis,
C., 2010. Wireless Communication. In E-maintenance (eds.) Holmberg, K. et al., Springer
London. pp. 247-272.
Kubler, S., Yoo, M.-J., Cassagnes, C., Framling, K., Kiritsis, D. and Skilton, M. 2015
Opportunity to Leverage Information-as-an-Asset in the IoT -- The Road Ahead, 2015 3rd
International Conference on Future Internet of Things and Cloud, pp. 64–71.
Lee, J., Ni, J., Djurdjanovic, D., Qiu, H. and Liao, H., 2006. Intelligent prognostics tools and
e-maintenance. Computers in industry, 57(6), pp.476-489.
Lee, J., Lapira, E., Bagheri, B. and Kao, H.A., 2013. Recent advances and trends in
predictive manufacturing systems in big data environment. Manufacturing Letters, 1(1),
pp.38-41.
Leite, G.D.N.P., Araújo, A.M. and Rosas, P.A.C., 2017. Prognostic techniques applied to
maintenance of wind turbines: a concise and specific review. Renewable and Sustainable
Energy Reviews, forthcoming.
Levrat, E., Iung, B. and Crespo Marquez, A., 2008. E-maintenance: review and conceptual
framework. Production Planning & Control, 19(4), pp.408-429.
Li, B.H., Hou, B.C., Yu, W.T., Lu, X.B. and Yang, C.W., 2017. Applications of artificial
intelligence in intelligent manufacturing: a review. Frontiers of Information Technology &
Electronic Engineering, 18(1), pp.86-96.
Lin, S., Gao, J., Koronios, A. and Chanana, V., 2007. Developing a data quality framework
for asset management in engineering organisations. International Journal of Information
Quality, 1(1), pp.100-126.
Liyanage, J.P., Lee, J., Emmanouilidis, C. and Ni, J., 2009. Integrated e-Maintenance and
intelligent maintenance systems. Handbook of maintenance management and engineering,
pp.499-539.
Llinas, J., Snidaro, L., García, J. and Blasch, E., 2016. Context and fusion: Definitions,
terminology. In Context-Enhanced Information Fusion (pp. 3-23). Springer International
Publishing.
Lomotey, R.K., Pry, J.C. and Chai, C., 2018. Traceability and visual analytics for the
Internet-of-Things (IoT) architecture. World Wide Web, 21, no. 1 pp. 7-32.Loukopoulos, P.,
Zolkiewski, G., Bennett, I., Pilidis, P., Duan, F. and Mba, D., 2017. Dealing with missing data
as it pertains of e-maintenance. Journal of Quality in Maintenance Engineering, forthcoming.
McFarlane, D., Giannikas, V., Wong, C. Y., & Harrison, M. 2012. Intelligent Products in the
Supply Chain - 10 Years On in '14th IFAC Symposium on Information Control Problems in
Manufacturing', pp. 655-660, Bucharest, Romania
McNaught, K.R. and Zagorecki, A., 2009, December. Using dynamic Bayesian networks for
prognostic modelling to inform maintenance decision making. In Industrial Engineering and
Engineering Management, 2009. IEEM 2009. IEEE International Conference on (pp. 1155-
1159). IEEE.
Meyer, G.G., Främling, K. and Holmström, J., 2009. Intelligent products: A survey.
Computers in industry, 60(3), pp.137-148.
MIMOSA Machinery Information Management Open Systems Alliance, 2017. Accessed on
05/09/17 available at: http://www.mimosa.org/
Moreau, L., Clifford, B., Freire, J., Futrelle, J., Gil, Y., Groth, P., Kwasnikowska, N., Miles, S.,
Missier, P., Myers, J. and Plale, B., 2011. The open provenance model core specification
(v1. 1). Future generation computer systems, 27(6), pp.743-756.
Mourtzis, D., Vlachou, E., Milas, N. and Xanthopoulos, N., 2016. A cloud-based approach for
maintenance of machine tools and equipment based on shop-floor monitoring. Procedia
CIRP, 41, pp.655-660.
Mourtzis, D. and Vlachou, E., 2018. A cloud-based cyber-physical system for adaptive shop-
floor scheduling and condition-based maintenance. Journal of manufacturing systems, 47,
pp.179-198.
Muller, A., Marquez, A.C. and Iung, B., 2008. On the concept of e-maintenance: Review and
current research. Reliability Engineering & System Safety, 93(8), pp.1165-1187.
Munir, M., Baumbach, S., Gu, Y., Dengel, A. and Ahmed, S., 2018. Data Analytics: Industrial
Perspective & Solutions for Streaming Data. by Last M., Kandel A., Bunke H., Series in
Machine Perception and Artificial Intelligence, World Scientific Publishing Co, 83(1).
Muñoz, C.Q.G., Marquez, F.P.G., Liang, C., Maria, K., Abbas, M. and Mayorkinos, P., 2015.
A new condition monitoring approach for maintenance management in concentrate solar
plants. In Proceedings of the Ninth International Conference on Management Science and
Engineering Management (pp. 999-1008). Springer, Berlin, Heidelberg.
Nehinbe, J.O., Solanke Olakunle, O. and Ige, N.J., 2014. An Analytic Method for Designing
Countermeasures against Computer Intrusions. Journal of Communication and Computer,
11, pp.10-21.
Nilsson, J. and Bertling, L., 2007. Maintenance management of wind power systems using
condition monitoring systems—life cycle cost analysis for two case studies. IEEE
Transactions on energy conversion, 22(1), pp.223-229.
NIST - National Institute of Standards and Technology, 2013. Foundations for Innovation in
Cyber-Physical Systems. Workshop Report, January 2013
[http://www.nist.gov/el/upload/CPS-WorkshopReport-1-30-13-Final.pdf]
Niu, G. and Jiang, J., 2017. Prognostic control-enhanced maintenance optimization for multi-
component systems. Reliability Engineering & System Safety. Forthcoming.
Niu, G., Yang, B.S. and Pecht, M., 2010. Development of an optimized condition-based
maintenance system by data fusion and reliability-centered maintenance. Reliability
Engineering & System Safety, 95(7), pp.786-796.
Nuñez, D.L. and Borsato, M., 2017. An Ontology-based Model for Prognostics and Health
Management of Machines. Journal of Industrial Information Integration. Forthcoming.
Papakostas, N., O'Connor, J. and Byrne, G., 2016, October. Internet of things technologies
in manufacturing: Application areas, challenges and outlook. In Information Society (i-
Society), 2016 International Conference on (pp. 126-131). IEEE.
Papathanassiou, N., Pistofidis, P. and Emmanouilidis, C., 2013. Competencies development
and self-assessment in maintenance management e-training. European Journal of
Engineering Education, 38(5), pp.497-511.
Park J, Nguyen D, Sandhu R. 2011. On data provenance in group-centric secure
collaboration. InCollaborative Computing: Networking, Applications and Worksharing
(CollaborateCom), 2011 7th International Conference on 2011 Oct 15 (pp. 221-230). IEEE.
Parpala, R.C. and Iacob, R., 2017. Application of IoT concept on predictive maintenance of
industrial equipment. In MATEC Web of Conferences (Vol. 121, p. 02008). EDP Sciences.
Patel, P., Ali, M.I. and Sheth, A., 2017. On using the intelligent edge for IoT analytics. IEEE
Intelligent Systems, 32(5), pp.64-69.
Payan, A.P., Gavrilovski, A., Jimenez, H. and Mavris, D.N., 2016. Review of Proactive
Safety Metrics for Rotorcraft Operations and Improvements Using Model-Based Parameter
Synthesis and Data Fusion. In AIAA Infotech@ Aerospace, 4-8 January 2016, San Diego,
California, USA. pp. 2133.
Peng, Y., Dong, M. and Zuo, M.J., 2010. Current status of machine prognostics in condition-
based maintenance: a review. The International Journal of Advanced Manufacturing
Technology, 50(1), pp.297-313.
Penna, R., Amaral, M., Espíndola, D., Botelho, S., Duarte, N., Pereira, C.E., Zuccolotto, M.
and Frazzon, E.M., 2014, July. Visualization tool for cyber-physical maintenance systems. In
Industrial Informatics (INDIN), 2014 12th IEEE International Conference on (pp. 566-571).
IEEE.
Perera, C., Zaslavsky, A., Christen, P. and Georgakopoulos, D., 2014. Context aware
computing for the internet of things: A survey. IEEE Communications Surveys & Tutorials,
16(1), pp.414-454.
Pistofidis, P., Emmanouilidis, C., Papadopoulos, A., and Botsaris, P.N., 2016. Management
of linked knowledge in industrial maintenance. Industrial Management & Data Systems,
116(8), pp.1741-1758.
Pliego Marugán, A. and García Márquez, F.P., 2016. A novel approach to diagnostic and
prognostic evaluations applied to railways: A real case study. Proceedings of the Institution
of Mechanical Engineers, Part F: Journal of Rail and Rapid Transit, 230(5), pp.1440-1456.
Posada, J., C. Toro, I. Barandiaran, D. Oyarzun, D. Stricker, R. De Amicis, R., E. B. Pinto, P.
Eisert, J. Dollner, and I. Vallarino, 2015. Visual computing as a key enabling technology for
industrie 4.0 and industrial internet, Computer Graphics and Applications, IEEE, 35(2),
pp.26-40.
Qi, Q. and Tao, F., 2018. Digital twin and big data towards smart manufacturing and industry
4.0: 360 degree comparison. Ieee Access, 6, pp.3585-3593.
Ragab, A., Yacout, S., Ouali, M.S. and Osman, H., 2017. Pattern‐based prognostic
methodology for condition‐based maintenance using selected and weighted survival curves.
Quality and Reliability Engineering International. Forthcoming.
Ruiz-Arenas, S., Horváth, I., Mejía-Gutiérrez, R. and Opiyo, E., 2014. Towards the
maintenance principles of cyber-physical systems. Strojniški vestnik-Journal of Mechanical
Engineering, 60(12), pp.815-831.
Saalmann, P., Zuccolotto, M., da Silva, T.R., Wagner, C., Giacomolli, A., Hellingrath, B. and
Pereira, C.E., 2016. Application potentials for an ontology-based integration of intelligent
maintenance systems and spare parts supply chain planning. Procedia CIRP, 41, pp.270-
275.
Sakib, N. and Wuest, T., 2018. Challenges and Opportunities of Condition-based Predictive
Maintenance: A Review. Procedia CIRP, 78, pp.267-272.
Selcuk, S., 2017. Predictive maintenance, its implementation and latest trends. Proceedings
of the Institution of Mechanical Engineers, Part B: Journal of Engineering Manufacture,
231(9), pp.1670-1679.
Smirnov, A., Levashova, T. and Shilov, N., 2015. Patterns for context-based knowledge
fusion in decision support systems. Information Fusion, 21, pp.114-129.
Smirnov, A., Levashova, T. and Shilov, N., 2016. Context-Aware Knowledge Fusion for
Decision Support. In Context-Enhanced Information Fusion Springer International
Publishing. pp. 125-154.
Snidaro, L., García, J. and Llinas, J., 2015. Context-based information fusion: a survey and
discussion. Information Fusion, 25, pp.16-31.
Syafar, F., Koronios, A. and Gao, J., 2018. Mobile Technologies in Asset Maintenance. In
Engineering Asset Management 2016 (pp. 245-253). Springer, Cham.
Tedeschi S., Emmanouilidis C., Farnsworth M., Mehnen J., Roy R. 2017. New Threats for
Old Manufacturing Problems: Secure IoT-Enabled Monitoring of Legacy Production
Machinery. In: Lödding H., Riedel R., Thoben KD., von Cieminski G., Kiritsis D. (eds)
Advances in Production Management Systems. The Path to Intelligent, Collaborative and
Sustainable Manufacturing. APMS 2017. IFIP Advances in Information and Communication
Technology, vol 513. Springer, Cham
Tiwari, A., Turner, C.J., and Majeed, B. 2008. A Review of Business Process Mining: State
of the Art and Future Trends, Business Process Management Journal, 14 (1), pp 5-22.
Tiwari, A., Vergidis, K. and Turner, C., 2010. Evolutionary multi-objective optimisation of
business processes. Soft Computing in Industrial Applications, pp.293-301.
Truong, H.L., 2018. Integrated Analytics for IIoT Predictive Maintenance using IoT Big Data
Cloud Systems. In 2018 IEEE International Conference on Industrial Internet (ICII) (pp. 109-
118). IEEE.
Turner, C.J., Hutabarat, W., Oyekan, J. and Tiwari, A., 2016. Discrete Event Simulation and
Virtual Reality Use in Industry: New Opportunities and Future Trends. IEEE Transactions on
Human-Machine Systems, 46(6), pp.882-894.
Turner, C.J., Tiwari, A., Olaiya, R. and Xu, Y., 2012. Business Process Mining: From Theory
to Practice, Business Process Management Journal, 18(3), pp.493 - 512.
Ucar, M. and Qiu, R.G., 2005. E-maintenance in support of e-automated manufacturing
systems. Journal of the Chinese institute of industrial engineers, 22(1), pp.1-10.
Uhlmann, E., Laghmouchi, A., Geisert, C. and Hohwieler, E., 2017. Decentralized data
analytics for maintenance in industrie 4.0. Procedia Manufacturing, 11, pp.1120-1126.
Vafaei, N., Ribeiro, R.A. and Camarinha-Matos, L.M., 2019. Fuzzy early warning systems for
condition based maintenance. Computers & Industrial Engineering, 128, pp.736-746.
Van Horenbeek A, Pintelon L. A prognostic maintenance policy-effect on component
lifetimes. In Reliability and Maintainability Symposium (RAMS), 2013. Proceedings-Annual
2013 Jan 28, IEEE, pp. 1-6.
 Vaughn RB, Farrell J, Henning R, Knepper M, Fox K. 2005. Sensor fusion and automatic
vulnerability analysis. In Proceedings of the 4th international symposium on Information and
communication technologies 2005 Jan 3, Trinity College Dublin. pp. 230-235.
Vergidis, K., Turner, C., Alechnovic, A. and Tiwari, A., 2015. An automated optimisation
framework for the development of re-configurable business processes: a web services
approach. International Journal of Computer Integrated Manufacturing, 28(1), pp.41-58.
Vogel-Heuser B, Diedrich C, Pantförder D, Göhner P. 2014. Coupling heterogeneous
production systems by a multi-agent based cyber-physical production system. In Industrial
Informatics (INDIN), 2014 12th IEEE International Conference on 2014 Jul 27, IEEE, pp.
713-719.
Voisin, A., Levrat, E., Cocheteux, P. and Iung, B., 2010. Generic prognosis model for
proactive maintenance decision support: application to pre-industrial e-maintenance test
bed. Journal of Intelligent Manufacturing, 21(2), pp.177-193.
Wang, J., Zhang, L., Duan, L. and Gao, R.X., 2017. A new paradigm of cloud-based
predictive maintenance for intelligent manufacturing. Journal of Intelligent Manufacturing,
28(5), pp.1125-1137.
Wuest, T., Schmidt, T., Wei, W., and Romero, D., 2018, Towards (pro-)active intelligent
products
Xia, T., Dong, Y., Xiao, L., Du, S., Pan, E. and Xi, L., 2018. Recent advances in prognostics
and health management for advanced manufacturing paradigms. Reliability Engineering &
System Safety, 178, pp. 255-268.
Yamato, Y., Fukumoto, Y. and Kumazaki, H., 2016. Proposal of real time predictive
maintenance platform with 3D printer for business vehicles. arXiv preprint arXiv:1611.09944.
Yamato, Y., Fukumoto, Y. and Kumazaki, H., 2017. Predictive maintenance platform with
sound stream analysis in edges. Journal of Information processing, 25, pp.317-320.
Yao, Y., Meng, C., Wang, C. and Jin, S., 2016. Preventive Maintenance Policies for
Equipment Under Condition Monitoring Based on Two Types of Failure Rate. Journal of
Failure Analysis and Prevention, 16(3), pp.457-466.
Ye, J., Dobson, S. and McKeever, S., 2012. Situation identification techniques in pervasive
computing: A review. Pervasive and mobile computing, 8(1), pp.36-66.
Yunusa-Kaltungo, A. and Sinha, J.K., 2017. Effective vibration-based condition monitoring
(eVCM) of rotating machines. Journal of Quality in Maintenance Engineering, 23(3), pp.279-
296.
Zhang, L., Luo, Y., Tao, F., Li, B. H., Ren, L., Zhang, X., Guo, H., Cheng, Y., Hu, A., and
Liu, Y., 2014. Cloud manufacturing: a new manufacturing paradigm, Enterprise Information
Systems, 8(2), pp. 167-187.
Zhong, R.Y., Xu, C., Chen, C. and Huang, G.Q., 2017. Big Data Analytics for Physical
Internet-based intelligent manufacturing shop floors. International Journal of Production
Research, 55(9), pp.2610-2621.
Zhou, Q., Yan, P. and Xin, Y., 2017. Research on a knowledge modelling methodology for
fault diagnosis of machine tools based on formal semantics. Advanced Engineering
Informatics, 32, pp.92-112.

View publication stats