=Paper=
{{Paper
|id=Vol-2191/paper22
|storemode=property
|title=Data Generation with a Physical Model to Support Machine Learning Research for Predictive Maintenance
|pdfUrl=https://ceur-ws.org/Vol-2191/paper22.pdf
|volume=Vol-2191
|authors=Patrick Klein,Ralph Bergmann
|dblpUrl=https://dblp.org/rec/conf/lwa/KleinB18
}}
==Data Generation with a Physical Model to Support Machine Learning Research for Predictive Maintenance==
https://ceur-ws.org/Vol-2191/paper22.pdf
Data Generation with a Physical Model to
Support Machine Learning Research for
Predictive Maintenance
Patrick Klein and Ralph Bergmann
Business Information Systems II
University of Trier
54286 Trier, Germany
[kleinp][bergmann]@uni-trier.de,
http://www.wi2.uni-trier.de
Abstract. Today, manufacturing machines are continuously equipped
with various sensors, whose data enable to derive a comprehensive pic-
ture of the current state of each machine. Predictive maintenance ap-
proaches make use of this data in order to predict the occurrence of
possible failures before they actually occur, thereby significantly reduc-
ing production and service costs. The application of machine learning to
sensor data streams is an essential part of data-driven predictive main-
tenance in order to find the patterns in the data that are indicators
of upcoming faults. Thus, research on machine learning for predictive
maintenance is a recent and very challenging field. However, there are
currently no appropriate data sets available that can be used for this
kind of research. In this paper we therefore propose an approach for the
generation of predictive maintenance data by using a physical Fischer-
technik model factory equipped with several sensors. Different ways of
reproducing real failures using this model are presented as well as a gen-
eral procedure for data generation.
Keywords: Data Generation · Machine Learning · Predictive Mainte-
nance · Industry 4.0.
1 Introduction
Today, we are facing the beginning of a transformation towards the fourth in-
dustrial revolution, called Industry 4.0, which is characterized by manufactur-
ing and service innovations based on cyber-physical systems, big data and an
intensive use of methods from artificial intelligence [15]. In particular, manu-
facturing environments are equipped with various sensors and actuators that
are fully connected for the purpose of building an industrial Internet of Things,
with the aim of using data in real time for decision making. An important field
of service innovation is related to diagnosis and maintenance of manufacturing
machines. For this purpose, manufacturing machines are equipped with various
sensors, whose data enable to derive a comprehensive picture of the current state
2 P. Klein and R. Bergmann
of each machine. Based on this data, occurring problems can be diagnosed and
more importantly, upcoming problems can be predicted prior to their occurrence.
In particular, predictive maintenance (PredM) aims at foreseeing a breakdown
of the system to be maintained by detecting early signs of failure in order to
make maintenance work more proactive. It has been adopted by various sectors
in manufacturing and service industries in order to improve reliability, safety,
availability, efficiency and quality as well as to protect the environment [20].
For PredM to work, knowledge is required about characteristic data patterns
that are indicators of specific faults that have occurred or that are likely to occur
in the future. Due to the large number of potential faults as well as the large
variety of production machinery and components used, the manual definition
of such patterns is not feasible. Instead, machine learning (ML) is required to
automatically derive such patterns from available data. Patterns could describe
healthy states as well as states which are in the transition phase towards an
upcoming breakdown. The application of ML to PredM, however, comes along
with a variety of challenges, in particular related to the complexity and hetero-
geneity of data, the lack of labelled data, the need for transfer learning, as well
as the necessity of deriving models that enable explainable decision support.
Thus, PredM is an ideal application area for research in ML due to its complex-
ity and practical relevance. However, appropriate and realistic data is required
to conduct such research. While plenty of data sets are available for research
purposes in ML, there is a lack of data that can be used immediately for PredM
applications. Also it is nearly impossible (at least for Universities) to get real
data from industry due to the serious confidentiality issues involved.
In this paper we therefore address the issue of obtaining data appropriate
for ML research in PredM. The contribution of this paper is manyfold. First, we
present a brief overview of PredM and the involved research challenges for ML
(Sect. 2). Then, we characterize the required data to address these challenges,
analyze existing data sets as well as methods for the generation of new research
data (Sect. 3). The main contribution of the paper is the presentation of an
approach for the generation of PredM data based on a physical model of a
specific production environment implemented based on a Fischertechnik (FT)
model factory (Sect. 4). We further describe various ways for injecting faulty
behaviour in a defined way into the FT model factory, in order to collect the
respective data that can be used to learn the related patterns for prediction. We
describe the current state of realization as well as our planned future work (Sect.
5).
2 Predictive Maintenance and Machine Learning
2.1 Predictive Maintenance
Industrial maintenance involves all measures that are required to ensure or to
re-establish the proper functioning of industrial machinery. The goal is to pre-
vent the occurrence of failures that could lead to breakdowns or downtimes of
machines or that could lead to safety concerns. For example, a common cause
Data Generation with a Physical Model for Predictive Maintenance 3
of failure is wear, which is a gradual damage or deformation of material due to
forces, which occurs in many mechanical components, in particular in bearings,
O-rings, or gears. In general, systematic maintenance procedures increase the
availability of machines, reduce costs, and enable to schedule required mainte-
nance actions. Traditional, preventive maintenance involves the systematic in-
spection of machines following a fixed time schedule or a fixed mileage, which is
based on the simplified assumption that failures mostly occur after a certain and
known operating time or effort. However, it often happens that failures occur
before the scheduled maintenance activity or that maintenance actions are per-
formed although they are not yet necessary. Thus, PredM aims at overcoming
the fixed time schedule approach by introducing methods that are able to indi-
vidually predict upcoming failures. The goal is to perform maintenance actions
only when they are really necessary, i.e., not too early and not too late. For
companies, PredM has the advantage that maintenance costs can be reduced
significantly by better utilization of capacities and by avoiding downtimes in
manufacturing.
PredM is based on forecasting failures based on the current state captured
by various sensors, such as vibration, temperature, humidity, or acoustic sensors.
In addition, parameters characterizing the current state in the production pro-
cess (e.g. position sensors or switches as well as the activity state of actuators)
are relevant. Machines in real production environments may have hundreds of
various sensors producing data streams with high frequency. As there is a wide
variety of factories, plants, and other systems that depend on reliability, it is
not economically feasible to build a customized solution for each plant. Thus,
research is focusing on universal parts of machines, such as bearings, gearboxes,
motors, valves, pumps, compressors, and so on to find universal solutions.
2.2 Machine Learning for Predictive Maintenance
The increasing number of sensor data streams makes manual monitoring and
analysis impossible, which is why ML and especially deep learning are suitable
for PredM data processing [13, 25]. They are mostly applied to the typical PredM
tasks [11] such as Remain Useful Life (RUL), Root Cause Analysis also referred
to as Fault Diagnosis (FD), Fault Prediction (FP), and Maintenance Strategy
Optimization (MSO). The prediction of RUL values for components is probably
the most prominent application which is a regression task with multivariate time
series as input, however, sometimes it is performed as a classification task in
which the RUL values are discretized from larger ranges to classes. For instance,
Babu et al. [4] applied a convolutional neural network for RUL prediction and
Yuan et al. [23] compare different recurrent neural network architectures for
RUL and FD of an aircraft turbofan engine. Furthermore, FP is used to predict
upcoming incorrect functioning which is not caused by wear, for instance, a
future incorrect positioning of a robot arm in a manufacturing process or to
predict defects on a production line [24]. Finally, MSO supports decision-making
in questions when, what, and how an upcoming maintenance task should be
carried out.
4 P. Klein and R. Bergmann
2.3 Research Challenges
These tasks lead to a number of challenges and requirements for the application
of ML methods in PredM. The first issue is the large number of distributed,
heterogeneous sensor data streams with non-uniform time stamps. It is difficult
to determine the subset of relevant data streams for the detection of a failure
as well as the time frame in which these data streams produce characteristic
patterns that are an indication of this failure. Quite often it is difficult to label
correctly the occurrence of a certain failure, as maintenance protocols are usu-
ally the only source of information about when which failure has occurred. This
leads to huge problems related to the data preparation prior to the use of ML
algorithms. In addition, failures are usually the exception, which makes the data
sets highly unbalanced. Although the overall volume of data is huge, the number
of different failure cases for a certain type of failure is rather small. This leads
to the need for transfer learning, in order to be able to transfer a learned failure
model from one machine component to a different, but similar component. Fi-
nally, the ability to explain a certain prediction is also very important in PredM
in order to enable a human operator to assess and verify an automatically pro-
posed maintenance action. All these different topics make PredM an interesting
field of application for ML research.
3 Research Data for Machine Learning in Predictive
Maintenance
3.1 Requirements on Research Data for Predictive Maintenance
Research
For conducting ML research for PredM (as well as for other kinds of Industry 4.0
applications of ML), it is necessary to have data available that is to some degree
comparable to the data in industrial settings. Thus, data sets are necessary which
are composed of various data streams with different characteristics (according
to the type of sensors used in industry) together with related data about the
current status of the production component or process. For learning to predict
failures, the data streams must contain patterns, similar to those that occur
in real situations, for example as a result of wear. Also the data sets must be
at least partially labelled with the respective fault to be predicted. Ideally, we
need large data sets describing several instances of the same fault and data sets
describing the same fault in various different but similar components (e.g. wear
in the bearings of different motors) to investigate transfer learning approaches.
3.2 Existing Data Sets
Eker et al. [5] benchmarked six common run-to-failure data sets for their appli-
cation to data-driven prognostics and found only two of them applicable. One
of them [22] consists of only 68 run-to-failure measurements and the other one
[19] consists of between 100 and 260 train examples for four different settings
Data Generation with a Physical Model for Predictive Maintenance 5
and failure types. The major sources for industrial prognostic data sets are the
NASA Prognostics Data Repository with 16 data sets1 as well as a collection
consisting of a dozen data sets from previously organized data competitions by
Prognostic and Health Management Society2 (of engineering systems).
The available data sets known to us have a variety of issues which makes
them not suitable for ML research on PredM. The data sets are either too
small for ML, several of them have anonymized features, and nearly all sets
are insufficiently labelled for the mentioned PredM tasks. Further, the data sets
usually only consider individual components or working station cells, but do not
provide a comprehensive snapshot of a factory’s sensor data.
3.3 Approaches for Data Generation
Since there is no sufficient or adequate data from real industrial factories pub-
licly available for research purposes, it is desirable to collect or generate them.
Sensor data generation without the real production environment at hand can be
categorized into four groups: 1. fully synthetical, 2. synthetical based on previous
data, 3. synthetical based on a virtual simulation model, and finally 4. based on
a simplified physical model.
Fully Synthetic Data Generation Fully synthetic data generation means
that sensor data is generated by an algorithm based on given parameters. The
resulting streams are based on a statistical structure and can contain concept
drifts (changing of underlying statistical properties over time). Typical parame-
ters are the data generating distribution (e.g. Gaussian), noise rate, data dimen-
sionality, and generation periodicity. For instance, Hahsler et al. [9] provide a
software framework for generation and analysis of fully synthetical data streams.
Synthetic Data Generation Based on Previous Data Another way to
generate sensor data is to learn the underlying properties of an existing data
distribution in order to generate new data. This can be done by training a gen-
erative and discriminative neural model by learning either explicitly the param-
eters of the distribution [2] or implicitly with a generative adversarial network
for time series [7].
Synthetic Data Generation Based on a Virtual Simulation Model A
further approach is the creation of a virtual simulation model with the properties
of the real model and use this for data generation. This approach, for example,
has been applied to aircraft gas turbines [19] and to create a virtual factory [12]
including detailed machine level data streams for testing machine health data
analytics applications.
1
https://ti.arc.nasa.gov/tech/dash/groups/pcoe/prognostic-data-repository/
2
https://www.phmsociety.org/
6 P. Klein and R. Bergmann
Data Collection Based on a Simplified Physical Model Instead of using
a virtual model of a factory or machine, there is also the possibility of using a
real but simplified physical model. Regarding the level of abstraction and the
constituents of the model, such models can be divided into two categories.
The first category are models which are equipped with real industrial com-
ponents leading to minor abstractions. Examples of such factories are Learning
Factories [1] such as AutFab [21] or the SmartFactory3 particularly established
for Industry 4.0 research. Also small physical models for the generation of specific
faults, such as bearing faults [18] exist.
The second category consists of models with a higher level of abstraction,
which are build using non-industrial components. The advantage of this ap-
proach is the significantly low cost involved in building such a model. There
are several platforms which enable the simple cost-efficient construction of such
models. Among them the most popular are Lego Mindstorms4 and Fischertech-
nik (FT)5 . Examples are the Smart-LEGO Factory6 at DFKI, the FT plant
model implemented for applicability validation of Industry 4.0 components [16]
and a FT punching workstation built to demonstrate how a generic client can
access data generated from the workstation [3].
4 Generating Research Data for Predictive Maintenance
by FT Model Factory
As publicly available data sets for research on ML for PredM are rare and lim-
ited to single components or working cells, we now describe an approach for
the generation of such data based on a simplified physical model. The aim is
to provide a cost-effective way to generate data according to the requirements
sketched in Sect. 3.1. This requires constructing a physical model of a factory,
attaching appropriate sensors and related data collection hard- and software, as
well as developing means for simulation faults.
4.1 A Physical Model Factory for Data Generation
Our Industry 4.0 factory model is built based upon the FT Factory Simulation1
as shown in Fig. 1. It has been selected due to its superior robustness compared
to Lego Mindstorms. As FT is also used in University education for automa-
tion engineers, the available FT components are already closer to real industrial
components than those of Lego. The FT factory model that we use consists
of four modules: a sorting line with color detection, a multi-processing station
with oven and milling machine, a high-bay warehouse, and a vacuum gripper
3
http://www.smartfactory.de/
4
https://www.lego.com/en-us/mindstorms
5
https://www.fischertechnik.de/en
6
https://www.dfki.de/web/aktuelles/dfki-cebit-2016/smart-lego
1
https://www.fischertechnik.de/en/service/elearning/simulating/fabrik-simulation-
24v
Data Generation with a Physical Model for Predictive Maintenance 7
Fig. 1. The FT Factory Simulation Model
robot. Each module is operated by its own controller based on an ARM Cortex
A8 CPU with various analog and digital input/output ports running under a
LINUX kernel. Overall, the model consists of nine light barriers, ten switches,
twelve motors and three compressors. Moreover, we enhanced the model with
six three-axis acceleration sensors that are mounted on motors and compressors
for vibration measuring and four differential pressure sensors are measuring the
pressure generated from the three compressors. These sensors are connected to
a separate Raspberry Pi controller. Furthermore, two micro-electro-mechanical
systems (MEMS) each with a gyroscope, an accelerometer, and a geomagnetic
sensor will be installed on the Robotic Vacuum Gripper and the High-Bay’s
storage and dispensing machine. RFID-Tags for product identification and a
reading device are also planned. Further, we will extend the oven model with a
heating pad in order to change the color of thermo-colored product materials.
This process will be monitored by a thermal imaging camera.
All controllers are connected via an Ethernet network and communicate via
remote procedure calls. The overall control software for the entire production
process is distributed over the controllers, each of which is in charge of a cer-
tain module of the factory. For processing the generated data, we selected the
SMACK stack [8] as a Lambda architecture [10] implementation because it is of-
ten used for Big Data applications in industry. Thus, we set up each controller as
a producer to the high throughput distributed messaging system Apache Kafka
[14]. Apache Cassandra was installed as a database for batch processing and we
further plan to use Apache Spark for stream processing and ML research.
8 P. Klein and R. Bergmann
The overall manufacturing process is designed as a cycle, meaning that data
can be generated without manual interference. The process starts from the High
Bay where workpieces are dispensed and transported to the Multi Processing
Station. After processing, they are sorted by color, transported by the Robotic
Vacuum Gripper and finally stored in the High Bay where the process repeats.
4.2 Reproduction of Failures
By using the FT model along with the developed software, the manufacturing
process is executed in a continuous loop. As FT blocks are quite robust and all
physical connections are very stable, problems occur quite rarely and hence the
model is able to run properly over a very long period of time. However, in order
to be able to produce data for predictive maintenance, faults must occur such
that the resulting data can be collected. As such faults do not occur naturally
(within an acceptable time limit) realistic faults must be artificially infused into
the model.
Figure 2 describes the interplay between reality, our FT model, the creation
of faults, and finally the data generation. In general, reality defines which failure
types are measurable and reasonable. Our FT model is a smaller and simplified
representation of reality and due to this it certainly restricts our ability to re-
produce realistic defects. Also life expectancy for components in real machines is
months to years and degradation processes are very slow. Thus we have to com-
press the time dimension, i.e., we have to significantly shorten the time during
which a certain type of fault causes its typical effects. Based on these limitations
we define plausible defects that can be simulated by our physical model such
that data is generated that can be used for learning and evaluating prognostic
models on predictive maintenance.
In general, there are several ways in which behaviour can be generated similar
to a failure in reality.
Fig. 2. Methodology and process for reproduction of failures.
Data Generation with a Physical Model for Predictive Maintenance 9
Amending the Model by Additional Actuators In order to produce an
abnormal behaviour of the physical model, additional actuators can be inte-
grated whose activities cause certain disturbances. For example, workpieces can
be pushed from the conveyor, a pressure line can be virtually broken by inserting
a pressure valve, or an additional motor can be inserted to produce an additional
mechanical load on a drive shaft.
Adapting the Controller Software for Actuator Based on knowledge
about how certain failures (e.g. motor problems due to wear) have an impact on
an actuator (reduced or unstable revolution speed), the controller software can
be designed such that it controls the actuator in a way that it behaves as if it
would exhibit the failure. For example, the motor supply voltage can be reduced
following a certain pattern or the frequency for the pulse-width modulation of
motor power supply can be lowered to increase vibration.
Simulating Defective Sensors Faults related to a defect of any kind of sensor
are also quite likely. They could also have a significant impact on the production
process, in case the sensor is used within the control procedure of the machine.
For example, a defective position switch might cause problems, as a gripper is
not able to adjust itself to the correct position. Defective sensors can be easily
simulated as part of the control software by manipulating the value they produce.
4.3 Data Generation Process
The just described ways of generating faulty behaviour have to be embedded
into an overall generation process for maintenance data. Therefore, the following
data generation process has been developed, allowing to generate a large number
of labelled maintenance data sets automatically. This process runs in a loop
consisting of four steps (see Fig. 2):
1. Selection of the particular error (e.g. motor failure due to wearing) to be
produced in the current run, including the relevant parameters (which motor,
degree of wearing, failure pattern curve, time horizon of wear process, etc.).
2. Configuration of the controller software to run the factory in a mode, in
which the respective fault reproduction is enabled.
3. Start of the controller software to run the production process. During the
run of the factory, all data is collected and stored in an Apache Cassandra
data base and labelled with the respective failure being produced.
4. After the failure has occurred, the factory model is reset to a defined initial
state compensating for any inconsistencies that might have resulted from the
insertion of the failure.
4.4 Example Case
In reality, bearing faults are the biggest failure source with almost 40% to 50%
of electrical motors. They result in vibration signatures of higher amplitudes,
10 P. Klein and R. Bergmann
increased noise, and also in a reduced motor torque and thus motor speed [17].
Bearing faults do not occur suddenly, but develop over time. To simulate this
kind of failure in the FT model, we first run the model in the regular mode (to
collect data unaffected by faults) and then we slowly reduce the motor speed
over time and in addition decrease the frequency for the pulse-width modulation
of motor power. Depending on which motor is affected, the reduced speed, for
example, leads to an increased duration of the movings of the gripper or increases
the time for transport of the workpiece on the conveyor. This leads to longer
delays until the respective signals from position switches arrive. In addition the
acceleration sensors record an increased vibration amplitude on one axis. This
data is recorded along with the data of all other sensors of the factory model.
Besides the sensors directly affected by the reduced motor speed, other sensors
might also be affected as an indirect consequence of the reduced speed. After
several runs of the data generation process with different variations of the failure
parameters, appropriate data is available for learning failure patterns and to
address the research challenges of ML for PredM.
5 Conclusion and Future Work
In this paper, we address the problem of data generation to enable ML research
for PredM. We surveyed currently available data sets and present several ap-
proaches for data generation. We then present an new approach for PredM data
generation based on a FT factory model. As of today, the mechanical and electri-
cal side of the model is nearly completely realized, the sensor data is collected,
processed, and stored using the SMACK-Stack as described. First failures, as
the example case just described, are implemented.
Future work will first of all address the implementation of additional failure
scenarios based on the approaches described in Sect. 4.2. This work is quite
difficult as it requires at least a basic understanding of typical mechanical faults
and their consequences in order to be able to reproduce them on the model.
In general it would be desirable to find ways of validating to which degree the
data we are collecting is realistic compared to real production data. However,
this will be clearly difficult as data from comparable real components, such as
provided in [6] for rolling bearings, is hardly available. However, we assume
that for the development of ML methods for PredM the exact reproduction of
patterns from real industrial factories is not required, as the goal of ML methods
is to find patterns according to the production environment at hand. Thus, we
are confident that the developed FT factory model is an appropriate means to
perform laboratory research on ML in a well controlled environment. We also
plan to publish the gained data sets such that they could be used by other
researchers as well.
References
1. Abele, E., Metternich, J., Tisch, M., Chryssolouris, G., Sihn, W., ElMaraghy, H.,
Hummel, V., Ranz, F.: Learning factories for research, education, and training.
Data Generation with a Physical Model for Predictive Maintenance 11
Procedia CiRp 32, 1–6 (2015)
2. Alzantot, M., Chakraborty, S., Srivastava, M.: Sensegen: A deep learning architec-
ture for synthetic sensor data generation. In: Pervasive Computing and Commu-
nications Workshops (PerCom Workshops), 2017 IEEE International Conference
on. pp. 188–193. IEEE (2017)
3. Angione, G., Barbosa, J., Gosewehr, F., Leitão, P., Massa, D., Matos, J., Peres,
R.S., Rocha, A.D., Wermann, J.: Integration and deployment of a distributed and
pluggable industrial architecture for the perform project. Procedia Manufacturing
11, 896–904 (2017)
4. Babu, G.S., Zhao, P., Li, X.L.: Deep convolutional neural network based regression
approach for estimation of remaining useful life. In: International conference on
database systems for advanced applications. pp. 214–228. Springer (2016)
5. Eker, Ö.F., Camci, F., Jennions, I.K.: Major challenges in prognostics: study on
benchmarking prognostic datasets (2012)
6. El-Thalji, I., Jantunen, E.: Fault analysis of the wear fault development in rolling
bearings. Engineering Failure Analysis 57, 470–482 (2015)
7. Esteban, C., Hyland, S.L., Rätsch, G.: Real-valued (medical) time series generation
with recurrent conditional gans. arXiv preprint arXiv:1706.02633 (2017)
8. Estrada, R., Ruiz, I.: Big data smack. Apress, Berkeley, CA (2016)
9. Hahsler, M., Bolaos, M., Forrest, J.: Introduction to stream: An extensible
framework for data stream clustering research with r. Journal of Statisti-
cal Software, Articles 76(14), 1–50 (2017). https://doi.org/10.18637/jss.v076.i14,
https://www.jstatsoft.org/v076/i14
10. Hausenblas, M., Bijnens, N.: Lambda architecture, http://lambda-
architecture.net/
11. Hegedűs, C., Varga, P., Moldován, I.: The mantis architecture for proactive mainte-
nance. In: 2018 5th International Conference on Control, Decision and Information
Technologies (CoDIT). pp. 719–724. IEEE (2018)
12. Jain, S., Shao, G., Shin, S.J.: Manufacturing data analytics using a virtual factory
representation. International Journal of Production Research 55(18), 5450–5464
(2017)
13. Khan, S., Yairi, T.: A review on the application of deep learning in system health
management. Mechanical Systems and Signal Processing 107, 241–265 (2018)
14. Kreps, J., Narkhede, N., Rao, J., et al.: Kafka: A distributed messaging system for
log processing (2011)
15. Lee, J., Kao, H.A., Yang, S.: Service innovation and smart analyt-
ics for industry 4.0 and big data environment. Procedia CIRP 16, 3
– 8 (2014). https://doi.org/https://doi.org/10.1016/j.procir.2014.02.001,
http://www.sciencedirect.com/science/article/pii/S2212827114000857, product
Services Systems and Value Creation. Proceedings of the 6th CIRP Conference
on Industrial Product-Service Systems
16. Lüder, A., Schleipen, M., Schmidt, N., Pfrommer, J., Henßen, R.: One step towards
an industry 4.0 component. In: Automation Science and Engineering (CASE), 2017
13th IEEE Conference on. pp. 1268–1273. IEEE (2017)
17. Nandi, S., Toliyat, H.A., Li, X.: Condition monitoring and fault diagnosis of elec-
trical motors a review. IEEE transactions on energy conversion 20(4), 719–729
(2005)
18. Nectoux, P., Gouriveau, R., Medjaher, K., Ramasso, E., Chebel-Morello, B., Zer-
houni, N., Varnier, C.: Pronostia: An experimental platform for bearings accel-
erated degradation tests. In: IEEE International Conference on Prognostics and
12 P. Klein and R. Bergmann
Health Management, PHM’12. pp. 1–8. IEEE Catalog Number: CPF12PHM-CDR
(2012)
19. Saxena, A., Goebel, K., Simon, D., Eklund, N.: Damage propagation modeling for
aircraft engine run-to-failure simulation. In: Prognostics and Health Management,
2008. PHM 2008. International Conference on. pp. 1–9. IEEE (2008)
20. Selcuk, S.: Predictive maintenance, its implementation and latest trends. Proceed-
ings of the Institution of Mechanical Engineers, Part B: Journal of Engineering
Manufacture 231(9), 1670–1679 (2017)
21. Simons, S., Abé, P., Neser, S.: Learning in the autfab–the fully automated indus-
trie 4.0 learning factory of the university of applied sciences darmstadt. Procedia
Manufacturing 9, 81–88 (2017)
22. Virkler, D.A., Hillberry, B., Goel, P.: The statistical nature of fatigue crack prop-
agation. Journal of Engineering Materials and Technology 101(2), 148–153 (1979)
23. Yuan, M., Wu, Y., Lin, L.: Fault diagnosis and remaining useful life estimation of
aero engine using lstm neural network. In: Aircraft Utility Systems (AUS), IEEE
International Conference on. pp. 135–140. IEEE (2016)
24. Zhang, D., Xu, B., Wood, J.: Predict failures in production lines: A two-stage
approach with clustering and supervised learning. In: Big Data (Big Data), 2016
IEEE International Conference on. pp. 2070–2074. IEEE (2016)
25. Zhao, R., Yan, R., Chen, Z., Mao, K., Wang, P., Gao, R.X.: Deep learning
and its applications to machine health monitoring: A survey. arXiv preprint
arXiv:1612.07640 (2016)