A Modular Architecture for the Design of Condition Monitoring Processes

A Modular Architecture for the Design of Condition Monitoring Processes Available online at www sciencedirect com 2212 8271 © 2016 The Authors Published by Elsevier B V This is an open access article[.]

2212-8271 © 2016 The Authors Published by Elsevier B.V This is an open access article under the CC BY-NC-ND license

(http://creativecommons.org/licenses/by-nc-nd/4.0/)

Peer-review under responsibility of the scientific committee of the 49th CIRP Conference on Manufacturing Systems

doi: 10.1016/j.procir.2016.11.071

Procedia CIRP 57 ( 2016 ) 410 – 415

ScienceDirect

49th CIRP Conference on Manufacturing Systems (CIRP-CMS 2016)

A Modular Architecture for the Design of Condition Monitoring Processes

a Institute for Factory Automation and Production Systems, Egerlandstr 7-9, D-91058 Erlangen, Germany

* Corresponding author Tel.: +49-9131-85-28783; fax: +49-9131-85-302528 E-mail address: Hans.Fleischmann@faps.fau.de

Abstract

The increasing complexity of production plants in the context of Industry 4.0 poses new challenges with regards to technical maintenance and process control In this context, Cyber-Physical Systems (CPS) for intelligent condition monitoring enable fault-tolerant, predictable production systems Unfortunately, CPS and Condition Monitoring System development is challenging due to the distribution of computational tasks among heterogeneous industrial IT-architectures It is usually started from scratch, which is time-consuming and error-prone In times of CPS the employee as maintenance worker, who ensures the availability of production machinery, is still important In order to address these challenges, this publication we propose a modular architecture for Socio-CPS-based condition and process monitoring as well as fault diagnostics

© 2015 The Authors Published by Elsevier B.V

Peer-review under responsibility of Scientific committee of the 49th CIRP Conference on Manufacturing Systems (CIRP-CMS 2016)

Keywords: Condition Monitoring Systems; Industry 4.0; Technical Diagnosis; Socio-Cyber-Physical Systems

1 Introduction

The ongoing industrial revolution, called Industry 4.0, is

entirely transforming traditional manufacturing industries As

a technical basis, Cyber-Physical Systems (CPS) as well as

the expanding Internet of Things and Services (IoTS) have

immensely affected value creation, business models and the

organization of work in the domain of industrial production

[1] CPS provide a connection between the virtual and the

physical world combined with various self-x capabilities

These allow the realization of a smart factory, characterized

by high mutability and ergonomic working conditions [2]

Along with a high level of automation, the development of

intelligent monitoring and autonomous decision-making

processes is important in order to control and optimize both

industrial companies and entire value-adding networks

efficiently In today’s factories, operation, monitoring,

diagnosing and troubleshooting automated production systems

is already being supported by information technology, such as

Enterprise Resource Planning (ERP) Systems, Manufacturing

Execution Systems (MES), Programmable Logic Controllers

(PLC) and remote maintenance concepts However, functions

of centralized industrial systems are being increasingly shifted

to decentralized CPSs linked in an IoTS As shown in Fig 1,

an expanding IoTS leads to the successive dissolution of a hierarchical system arrangement in the classical automation pyramid to an automation cloud, in which entities such as CPS are used [1]

Enterprise Level (ERP)

Classical Automation Hierarchy

xUtilization of proprietary communication standards, tree networks

xLimited access to devices due to strict layered encapsulation of entities xLimited condition monitoring capabilities

IoTS/CPS-based Smart Factories

xUtilization of interoperable communication standards, mesh networks

xDecentralized, enhanced condition monitoring methods

Components

IoTS

Control Level (MES) Device Level (PLC) Field Level (Sensors, Actors)

Components

Figure 1 From classical maintenance to IoTS-based maintenance [1]

© 2016 The Authors Published by Elsevier B.V This is an open access article under the CC BY-NC-ND license

( http://creativecommons.org/licenses/by-nc-nd/4.0/ ).

Peer-review under responsibility of the scientifi c committee of the 49th CIRP Conference on Manufacturing Systems

Nevertheless, in practice there are still significant gaps in

networking various physical resources and corresponding

information systems - particularly with regard to

maintenance-related functions, processes and data The

aspects and capabilities of CPS have thus far only been

rudimentarily considered, but there is still no integrated

consideration of production resources, control systems or

employees in the shop floor

Maintenance in a smart factory is particularly characterized

by condition monitoring systems (CMS) for critical

components and sophisticated assistance systems, in order to

achieve a predictable and resilient CPS As a maintenance

worker, who ensures the availability of capital-intensive CPS

production resources, humans are even more of a success

factor This is accompanied by Socio-Cyber-Physical Systems

(Socio-CPS), which emphasize the integration and interaction

of and collaboration between humans as well as smart

machinery and lead to greater potential in the field of

maintenance [3] Within the scope of Socio-CPS, various

technological solutions and system architectures are currently

in discussion, but to date, there is a lack for tangible,

domain-specific architectures and best practices

This paper introduces a holistic, modular architecture for

Socio-CPS-based CMS in smart factories Following the

challenges and literature review in the field of CMS and

architectures for CPS, system design as well as a concept of

this architecture lay the foundation for this work’s discussion

We highlight a novel architecture for CPS-based condition

monitoring, fault diagnosis and predictive maintenance

2 State of the art

In smart factories, CMS represent the business logic that

monitor state variables and produce diagnostics of current and

future errors [4] Consequently, robust and accurate condition

monitoring is indispensable for autonomous production

machines to create products successfully However, most

CMS entail several tasks depending on the associated

production machine or process, which results in varying

requirements for cyber-physical computing architectures

Nevertheless, according to Schenk [5], CMS are arranged

according to a general processing scheme: (1) measuring and

storing machine parameters that reflect the current status of

the production machinery or the corresponding production

processes (state detection), (2) comparison of current data

with predetermined nominal data (state comparison) and (3)

specific error diagnostics (diagnosis)

The state detection process is the measurement and the

declaration of machine parameters that reflect the current

status of a production machine and its components For this

purpose, sensors are used for monitoring and quality

assurance This is associated with the fact that CPS have

diverse types of sensors for perceiving their environment as

well as their condition by definition As regards condition

monitoring applications, this results in vast amounts of data

from sensors concerning loads, temperatures, machine

parameters and environmental conditions yet to be

considered After data collection, identified machine

parameters are compared with predetermined, time-dependent

nominal values These nominal values are declared by the examined process parameters, while limits are usually empirically determined by the manufacturer or user of the machine Based on the results from comparing the states, potential errors can be detected earlier and more readily Hereby, the purpose of the diagnosis is to identify the errors’ and inefficiencies’ source of origin Fig 2 summarizes the partial steps of condition monitoring

CMSs have largely been studied from a theoretical to an applicative point of view Windmann et al [6] presents a novel classification schema for systems, models and model learning algorithms for diagnosis of automation systems Wollschlaeger et al [4] describes a reference architecture for condition monitoring systems In case of machine failures, maintenance workers must initially conduct a time-consuming, root-cause analysis To find the root of the problem, production machines predominantly deliver simple and inaccurate status codes, which are the basis for the repair process Complex errors and inefficiencies depend on several parameters and may likely only arise in specific time periods that cannot be detected by conventional CMS Moreover, relevant data is either distributed over several human machine interfaces (HMI) over the entire shop floor or is only available

in highly complex expert systems It is naturally important for the operator to have the data visualized correctly and with an accurate degree of granularity, which is often not the case due

to condition monitoring tasks being distributed amongst various industrial information systems

Moreover, the application of condition monitoring methods

on production facilities has typically been continuously started anew within industry, which has proven time-consuming and error-prone New methodologies and standards for modeling and implementing such applications are needed in order to reduce development time [4] In addition to the high degree of distribution, current CMS are tightly linked to vendors and proprietary systems, such that there is a need for CPS-based, interoperable monitoring and process control with a manual maintenance interface [7]

In comparison to existing work, we highlight a tangible, shopfloor-ready condition monitoring architecture by using web technologies for platform-independent Human Machine Interfaces (HMI) In the next chapter, we will discuss related scientific work in the field of CPS and CMS

1 State detection

2 State comparison

3

Diagnosis

Sensor value

Upper limit Exceedance

time value

Lower limit

CPS

Socio-CPS

Diagnose faults Take countermeasures

Fig 2 Partial steps of condition monitoring and their relationship according

to CPS [5]

2.1 CPS Architectures

CPS are characterized by their ability to adapt and learn:

They analyze their environment and, based on observations,

learn patterns, correlations and predictive models with

artificial intelligence [1] Typical industrial applications are

condition monitoring, predictive maintenance, image

processing and diagnostics In related scientific work,

high-level approaches have been introduced for reference

architectures as well as templates for implementation

strategies, which show the benefits of CPS and semantic

technologies for automation and smart factories [4] For

example, the term Plug-and-Produce refers to CPS’ ability to

react to the presence of new components and reconfigure

themselves accordingly, particularly for monitoring hardware

and supervising control systems [8] Additionally, the concept

of the Manufacturing Service Bus adapts the idea of

Service-Oriented Architectures (SOA) in the domain of industrial

production and CMS [9] Architectural aspects and

standardized communication are the keys to establish the

interaction of entities in a cyber-physical production scenario

In the field of CPS and IoTS, various communication

standards have been brought forth in public discussion

2.2 Human-CPS Interaction

To date, there continues to be an utter lack of standards,

best practices and design patterns as to how the maintenance

worker should interact with CPS on the shop floor [5]

Proponents of CPS argue that maintenance will be further

complemented through interaction with said maintenance

workers Unfortunately, CPS-based coordination and

decision-making mechanisms generally lead to cognitive

overload and particularly cause problems in the interaction

between employees on the shop floor and in sophisticated

CPS [10] It is furthermore difficult to integrate the

maintenance worker’s experiential knowledge into a

Socio-CPS Other untangible qualities, such as work ethic and

performance, are also necessary to react appropriately to

unpredictable situations This leads to a high demand for new

assistance systems

Frazzon [3] introduced the concept of Socio-CPS in

production and reviewed the social aspects of such systems

In doing so, he intends to promote future research towards

Socio-CPS applied to production networks In this

architecture, context-dependent behavioral aspects and

implications related to the human interface are delimited The

results obtained substantiate the dependence of Socio-CPS on

properly considering to the issues of human interaction

together with technology Smirnov [11] introduced

Cyber-Physical-Social Systems (CPSS), which integrate physical,

cyber, and social worlds Similar to Socio-CPS, CPSS rely on

communication, computation and control infrastructures that

consist of several hierarchical levels with heterogeneous

resources comprised of computational resources, IT services,

humans and automation components Therefore, operating

and configuring CPS requires techniques for managing the

variability during design and the dynamics during runtime,

both of which are caused by a multitude of component types and changing application environments

Furthermore, there is a demand for technical assistance systems with innovative, multi-modal, human-machine interfaces, semantic search models and evaluation techniques [1] Additionally, interaction between people and CPS is not limited to modalities such as a keyboard Research in the field

of mobile devices beyond conventional approaches has also been conducted, as with augmented reality data glasses and exoskeletons [10] Access to systems is even possible via touch screens, language or gestures

In addition to specific tasks sought in the field of condition monitoring, maintenance and system architectures, the question of organizing and controlling Socio-CPS is of great importance The organizational science thus significantly contributes to fundamental considerations for the design of application systems to support the work processes in Socio-CPS [3] Organizational and controlling problems occur through the gap between global system contexts that mainly involve coordination mechanisms for a variety of agents, and the local system context, which mainly include the direct interactions of one or fewer agents These enable the simultaneous consideration of both contexts by establishing appropriate organizational forms Here, research remains insufficient in particular in precisely configuring the forms of coordination

2.3 Smart Condition Monitoring

Technical maintenance and CMS represent two additional research fields within the CPS context Otto et al [12] focused on descriptive engineering approaches and distributed architectures for CPS They showed the ability to automatically generate monitoring software based on new requirements to support cyber-physical infrastructures for industrial applications In addition to signal-based technical diagnosis, model-based methods using sensor pattern recognition have become increasingly important [6]

3 Architecture Description

3.1 Requirements

In accordance with the requirements engineering of the research project S-CPS [7], we suggest the following requirements for our CMS architecture:

The architecture must primarily offer detailed condition monitoring functionalities for supporting the maintenance process Depending on the application, the kind of CPS and the industrial domain, different condition monitoring techniques are of merit Therefore, CMS should be appropriately flexible in terms of CPS-based condition monitoring techniques For precise monitoring purposes, a temporal granularity of operational data at the sub-process level is imperative Condition monitoring must perform data processing from various machine components that can adjust

to the sampling frequency of the monitored components

In order to monitor industrial processes with CPS, near

real-time constraint support is mandatory For this,

TCP/IP-based, real-time-enabled Ethernet protocols, such as Process

Field Network (PROFINET), have prevailed [1] To ensure

the efficient and low-latency execution of production

processes, CMS should analyze operational and sensor data

locally while maintaining low network loads Streamlined

communication solutions are also of necessity, whereby single

CPS with intelligent mechanisms process information in

advance and send events only when necessary In addition to

the time behavior within industrial control systems, the use of

interoperable communication standards with semantic

information models and protocols is necessary in order to

import and export input data and calculated results

The system should be more software-defined than

constrained by hardware in addition to being able to adapt to

future requirements To ensure transferability and adaptability

on existing systems as well as heterogeneous industrial

computing platforms and automation components, such IPC

or PLC, CMS must rely on novel, mainly software-based

solutions In industrial manufacturing companies, machinery

from various suppliers with heterogeneous communication

capabilities is used Steps towards integrating such machinery

into CMS should be taken Particularly in medium-sized

companies, acquired machinery stocks often represent a key

corporate value To ensure the applicability of the concepts

for large industrial companies as well as for medium-sized

companies, the Socio-CPS should be runnable by using

reliable, industry-grade components on low-cost embedded

devices

To fulfill social requirements, certain aspects that influence

HMI must be taken into consideration Firstly, in the

maintenance process, there are many stakeholders involved in

the company with varying roles (e.g maintenance staff,

management, external service providers) This fact requires

adaptability of the HMI as well as different kinds of

visualizations, depending on the requirements of the various

roles When developing the Socio-CPS, the interaction

between the user and CPS must be central To map the

interaction between an intelligent system and the maintenance

worker as well as possible, two-way communication

mechanisms and protocols without polling-mechanisms must

also be realized

3.2 Concept

In accordance with the classical methodology of condition

monitoring and the new requirements from Socio-CPS, an

overall architecture emerges, as depicted in Fig 3 By

categorizing the three partial steps of CMS, separating

concerns, transferability on distributed computing platforms

and a high degree of modularity can be reached Between the

modules, unified interfaces for data streaming guarantee

standardized communication and adaptability Usually,

relevant data for CMS are kept in Production Data

Acquisition (PDA) (I) as well as in (CPS-based) sensors (II)

and actors (III) By using decentralized data acquisition on the

machine level, the requirement for precise and rapid data

provision can be achieved After standardized Data

CPS-based Sensors (II)

Cloud (VI)

CPS-based Actors (III) PDA (I)

Data Aquisition (IV)

HMI (VII) Condition Monitoring Engine (V)

Machine Interface (VIII)

Figure 3 Block Diagram of the Condition Monitoring System

Acquisition (IV), the CMS-functionalities take place in the Condition Monitoring Engine (CME) (V) In order to meet the requirements for communication, the state detection and state comparison functionalities must be decentralized We fulfill the requirements as well as the theoretical construct of intelligent CPS machines for structured monitoring and self-diagnostics By applying this concept, the development of generic CMS with minimized communication overhead is possible For classical condition monitoring functionalities, nearly no communication to cloud-based systems is necessary Nevertheless, in an IoTS-production grid, data exchange with cloud-based applications (VI) and centralized services is still useful For example, in times of low traffic, updates, statistical classifiers or nominal values can be transferred between cloud-based services and the CME Finally, the HMI (VII) considers the integration of humans into the Socio-CPS On the one hand, the CME must have the ability to diagnose errors and push messages to a user if self-repair is not possible On the other hand, the maintenance worker should have the ability to trigger CME-specific abilities, such as adapting and learning nominal values for sensor and operational data based on patterns and observations In order to reflect human workers and the CPS interacting, Socio-CPS must consider the possibilities of bi-directional communication technologies and responsive graphical user interfaces (GUI) Additionally, the results of the CME should be exportable into the IoTS (VIII)

3.3 Socio-CPS CME design

In order to prove the applicability of the CMS architecture,

we will describe the design of the proposed Socio-CPS CME Based on the concept described above, Fig 4 shows an advanced block diagram of the CME architecture In the subsequent sections, details of the implementation are described Conforming to the Socio-CPS architecture, the CME is divided into the modules Import, Data Manager, Analysis and Provider Each module encapsulates a specific

task and exchanges data by using a standardized

publish-subscribe mechanism This approach offers maximum

flexibility, because it is possible to skip existing modules or

insert new modules into the layers Scalability is ensured by

transferring individual modules to various industrial

computing platforms (Cloud Computing, PLC, etc.)

In accordance with the postulated requirements, the

integration of existing machinery as well as modern

CPS-components into the CME must be possible For state

detection, a methodology for integrating relevant data from

heterogeneous sources is necessary, such as streaming data

from sensors or actors in communication protocols, or static

operational data in log files or databases In this context,

different data sources have varying integration demands

Whereas data from SOA and standardized protocols are

comparatively easy to integrate, there are proprietary

protocols that have high integration requirements In our

architecture, we take into account a layer model that evaluates

different machine components for data acquisition If the data

is available on a protocol from a higher level, the integration

of protocols in the lower layers may not necessarily be

mandatory Import must be customizable by using a

configuration manager without programming knowledge,

such that the CME can be tailored to each relevant CPS

Multiple connections and Monitored Items per data source

can be set and reconfigured, while each Monitored Item is a

variable stored in internal data structures that contain

references to specific update events The list of objects is also

held in internal data structures and exposed to subsequent

layers for further processing For efficient data processing in

the CME, a separate layer is needed that stores Monitored

Items from various CPS consistent for use with subsequent

layers The Data Manager is responsible for managing and

storing individual Data Sets A Data Set is created for each

Monitored Item upon initialization It holds a certain time

series of Data Points that is created on update events A Data

Point is a set of attributes and values, in this case generated by

the PDA, and is dedicated to a single Monitored Item at a

specific point in time Internally, the Data Manager uses both

a cache for fast data access and a database for persistent

storage A special set of methods provided by the Data

Manager enables straightforward data retrieval using query

methods The Monitored Items data can be fetched from the

Data Manager by given timestamps, time periods or quantities

of data points relative to the last available data Defined

queries facilitate comprehensive preprocessing by calculating

different metrics such as standard deviation and mean

Analysis is the core module of the CME and provides

implementing a configuration-based data linking strategy

Individual Data Sets can dynamically be connected and

changed to conform to the current production environment

and monitoring demands The combination of specific data

sources and different analysis functionalities also occur in this

module The actual Analysis, data manipulation and export

are conducted in dedicated sub-modules called Condition

Monitoring Adapters (CMA) A CMA performs specific

analysis tasks, such as predictive maintenance, condition

monitoring or error diagnostics Each CMA is computed at a

CPS-based Sensors (II)

CPS-based Actors (III) PDA (I)

Establish connections to data sources Parse condition monitoring configuration Install subscribe-mechanisms Trigger update events for Data Manager

Create Data Sets Cache Data Sets Provide API for data access by CMA Parse configuration

Initialize CMA Assemble condition monitoring results Condition Monitoring Adapter 1 Condition Monitoring Adapter n

Provide the results of CME Establish servers for data export

Serve dashboard content for HMI

Generation Process

CPS

Web Browser

IoTS (VIII)

HMI (VII)

Training Data

Nominal Data

Data Sets

Figure 4 Advanced Block Diagram of the Socio-CPS Architecture

certain update interval, which can be time- or event-based Internally, a CMA fetches the required shared data from the Data Manager module and computes the analysis result Afterwards, the return values are merged into the analysis result and optionally passed on to a subsequent CMA This enables lean, single-task data processing and, with leveraging, the chaining of CMA As explained below, artificial intelligence techniques play a crucial role in the execution of the condition monitoring functionalities In essence, these allow CPS to acquire normal operation procedures in order to detect and diagnose errors as well as inefficiencies Through design and implementation of appropriate CMA, various decentralized monitoring mechanisms can be applied in a standardized way The generated results are stored in an extended Data Set, which, in turn, is passed on to further modules In this case, the CMA is divided into a Cloud Service and a decentralized analysis within the CPS The execution of state comparison, which requires few computational resources, is performed locally on the CPS Creating and updating nominal data with a generation process

is intentionally cloud-based for the following reasons: (a) Easy uploading of diagnostic intelligence to similar machines

or systems, increases effectiveness, as CPS can possibly learn from several other CPS and (b) the condition monitoring functionalities also perform without network access

As outlined by the requirements, the results of the Analysis are exported to the IoTS In order to address any Socio-CPS aspects, it is necessary to present the state comparison and diagnostic results in a configurable, role-specific HMI Therefore, the Provider module was established To support the demands from the different stakeholder roles and those of

Monitored Item Monitored Item Monitored Item Monitored Item

Figure 5 Interactivity between Data Categories in the HMI

Socio-CPS, every user is able to individualize the HMI By

using browser-based technologies, this requirement is able to

be addressed The results are exposed in separate namespaces

to minimize the network traffic Data is only sent to clients

that explicitly connect to an Analysis result In addition to the

Analysis data, the user-specific dashboard configuration as

well as the visualization widgets and their connection to the

data sources are delivered in parallel to the CME data by a

web server The web server also delivers static data such as

Content Style Sheets (CSS) and application logic for

visualization Analogous to the other modules, the dashboard

is controlled by a flexible configuration, in which the widgets’

position and options are declared and linked to the Analysis

results Aside from the WS-based export for web browsers,

the CMS should offer relevant data via IoTS-specific

protocols in industrial networks Fig 5 shows the

relationships between Monitored Items, Data Sets, CMA and

the HMI Finally, the CME initializes the previously

explained modules and links their data streams The modules’

loose coupling makes it possible to extend and substitute any

of them A new module specifies and implements the unified

interfaces for exchanging Data Sets between the related

modules By means of a registration process, the orchestration

of the modules is done internally New CMA can be

implemented and activated in the Analysis, analogously to the

integration of new modules

4 CONCLUSION

Industry 4.0 has enabled many new services and

human-machine applications along industrial value chains Currently,

there is a need for sophisticated diagnostic assistance systems

on the shop floor level, due to the fact that diagnostic

assistance systems are primarily expert systems Socio-CPS

facilitate the interaction of maintenance workers and

intelligent production machines in the field of condition

monitoring, error diagnostics and predictive maintenance The

proposed architecture fulfills the outlined requirements of

CPS and shows new interaction mechanisms in collaboration

with CPS and human operators Cognitive overload in the

diagnosis of complex technical systems is mitigated and

manageable for the maintenance staff by using role-optimized

Socio-CPS CMS, according to the proposed architecture, thus

increase maintenance and repair efficiency in smart factories

The basis for the standardized application of various condition

monitoring functionalities was additionally able to be

established The developed architecture delivers significant

contributions to the ongoing design and architectural

development of IoTS-specialized CMS and Industry 4.0 applications The basis for the standardized application of various condition monitoring functionalities was established Through a modularized approach and respective CMA modules, this architecture represents a starting point for further, domain-specific developments in the environment of CPS The focus on the present work lies in the design and implementation of the architecture at a test cell for electronic assemblies due to validation purposes

Acknowledgements

We would like to thank the Federal Ministry of Education and Research as sponsor of the research project S-CPS

References

[1] ten Hompel M, Vogel-Heuser B, Bauernhansl T Industrie 4.0 in Produktion, Automatisierung und Logistik: Anwendung, Technologien, Migration Wiesbaden: Springer Vieweg; 2014

[2] Deutsche Akademie der Technikwissenschaften Cyber-Physical Systems: Driving force for innovation in mobility, health, energy and production Berlin Heidelberg: Springer Vieweg; 2011

[3] Frazzon EM, Hartmann J, Makuschewitz T, Scholz-Reiter B Towards Socio-Cyber-Physical Systems in Production Networks Procedia CIRP (Setúbal, Portugal, May 29 - 30, 2013) CIRP CMS 2013 p 49-54

[4] Wollschlaeger M, Theurich S, Winter A, Lubnau F, Paulitsch C A reference architecture for condition monitoring Proceedings World Conference on Factory Communication Systems Systems (WCFS) Palma

de Mallorca: 2015 p 1-8

DOI=http://dx.doi.org/10.1109/WFCS.2015.7160555 [5] Schenk M Instandhaltung technischer Systeme: Methoden und Werkzeuge zur Gewährleistung eines sicheren und wirtschaftlichen Anlagenbetriebs Berlin Heidelberg: Springer Vieweg; 2010

[6] Windmann S, Niggemann O Efficient fault detection for industrial automation processes with observable process variables Proceedings IEEE International Conference on Industrial Informatics (INDIN) Cambridge: 2015 p 121-126

DOI=http://dx.doi.org/10.1109/INDIN.2015.7281721 [7] Fleischmann H, Kohl J, Schneider M, Franke J Intelligent fault diagnosis

in Industry 4.0 - New opportunities for the sophisticated diagnosis in automated production environments wt Werkstattstechnik online 2015; 105:3 p 90–95

[8] Dürkop L, Jahanzaib I, Trsek H, Wisniewski L, Jasperneite J 2013 Using OPC-UA for the autoconfiguration of Real-time Ethernet Systems In Proceedings of the 11th IEEE International Conference on Industrial Informatics (Piscataway, N.J., July 29 - 31, 2013) p 248-253 DOI=http://dx.doi.org/10.1109/INDIN.2013.6622890

[9] Holtewert P, Wutzke R, Seidelmann J, Bauernhansl T 2013 Virtual Fort Knox Federative, Secure and Cloud-based Platform for Manufacturing In Procedia CIRP (Setúbal, Portugal, May 29 - 30, 2013) CIRP CMS 2013

p 527-532

DOI=http://dx.doi.org/10.1016/j.procir.2013.06.027 [10] Geisberger E, Broy M agendaCPS Berlin Heidelberg: Springer Vieweg;

2012

[11] Smirnov A, Levashova T, Shilov N, Sandkuhl K 2014 Ontology for cyber-physical-social systems self-organisation In Proceedings of the 16th Conference of Open Innovations Association (Oulu, Finland, October 27-31, 2014) FRUCT16 p 101–107 DOI=http://dx.doi.org/10.1109/FRUCT.2014.7000933

[12] Otto J, Henning S, Niggemann O 2014 Why Cyber-physical Production Systems Need a Descriptive Engineering Approach – A Case Study in Plug & Produce In Proceedings of the 2nd International Conference on System-Integrated Intelligence: New Challenges for Product and Production Engineering (Bremen, Germany, July 2-4, 2014) SysInt

2014 p 295–302

DOI=http://dx.doi.org/10.1016/j.protcy.2014.09.08