FP7-Europe

Overview of Achievements

The vision of SelSus is to create a new paradigm for highly effective, self-healing production resources and systems to maximise their performance over longer life times through highly targeted and timely repair, renovation and up-grading.

The SelSus vision will be achieved by the development of a new synergetic diagnostic and prognosis environment which is fully aware of the condition and history of all the machine components within a system or factory and is in constant knowledge enriched dialogue with their human personnel.

 

End user and component provider
ELECTROLUX

Contacts
ELECTROLUX ITALIA SPA
Corso Lino Zanussi,24 33080 Porcia (PN) ITALY
Contact person: Alessandro Mazzon - alessandro.mazzon@electrolux.com

Demo (Product) description
The demo scenario consists in a press line, installed at Porcia Plant (Italy), and composed of five presses in charge of forming the cabinet of washing machines. Due to the complexity of the manufacturing system, the demo has been limited to the first press of the line, the 800tons Schuler press, considered a valuable and representative element of the whole line. The development activities have been focused on the monitoring and analysis of some key modules of the press and their operative parameters with the aim to introduce a new HW/SW architecture and a methodology based on diagnosis and prognosis tools to reinforce the maintenance strategies around the line. The press line is quite aged, but still essential for the whole factory production flow since it is the only system capable of forming the cabinets. Any unpredictable stoppage, severe or easy-resolvable, is seen as a criticality in the production scheduling, that causes waste of resource in different areas (scraps, missing finished products, components repairing and/or substitution, maintenance costs).

Technical details

The demo is based on a HW and SW architecture that connects the press control (PLC) to a modelling environment located in the cloud with the aim to retrieve functional parameters and identify possible criticalities in the press operations. Since the press is mainly based on mechanical components, the selected press elements in the analysis are the oil circuits (lubrication and hydraulic), the flywheel motor and the air balancer cylinders, considered fundamental in the system cinematics movements and strictly correlated each other’s; an anomalous synthon in one element might be indicative for the malfunctioning of the others. The model environment is based on a dynamic Bayesian network, which is a graphical method that provides a probabilistic representation of the marginal and conditional independence relations in the represented elements domain. For each selected press element, an appropriate model has been built up to monitor the behaviour and the degradation of the elements to predict their lifetime. These models are constantly updated by the data retrieved from the press (sensors, human observations). The modelling and the simulation analysis is at the end supported by a DSS (decision support system) aiming to visualize, in a user-friendly UI, alerts on possible malfunctioning or machine stoppages and to perform root causes analysis to support the maintenance personnel in the activities planning in a more profitable way.


Product / service benefits

  • Reinforce the health monitoring and predictive strategy (diagnosis, degradation) on valuable, but aged manufacturing systems to assure a continuous factory productivity
  • Preserve the machines efficiency (OEE) and the finished products quality (no scraps)
  • Optimize the maintenance activities and the spare parts management to reduce costs
  • Make profit of the human expertise and historical knowledge on maintenance operations to reinforce the modelling and analysis strategies
  •  Introduce new tools and methodologies in manufacturing systems to favour the upcoming digitalization transformation process within Electrolux factories (Industry 4.0)

 

 

IEF Werner Harms Wende - Collaborative demonstrator – Linear axis Welding controller

IEF Werner & Harms & Wende - Collaborative demonstrator – Linear axis & Welding controller

Collaborative demonstrator – linear axis & welding controller

Contact

IEF-Werner GmbH
Ulrich Moser
Wendelhofstr. 6,
D-78120 Furtwangen
ulrich.moser(at)ief-werner.de

Harms & Wende Hamburg
Dr. Michael Peschl
Willy-Andreas-Allee 19,
D-76131 Karlsruhe,
michael.peschl(at)harms-wende.de

Demonstrator description
The demonstrator is an example of a welding robot, which welds metallic plates. It is constructed in collaboration of HWH, IEF, and XETIX SelSus partners. HWH and IEF provide hardware components, while XETIX organizes communication mechanisms via SelSus cloud system. Primary goal of this demonstrator is to present usage and advantages of new paradigms developed within the SelSus project, namely: highly effective self-healing and self-aware intelligent production systems/equipment. This functionality is achieved by updating the existing equipment with sensors and integrating them with intelligent diagnostic and decision support software modules, which analyse these sensor data.
In this demonstrator, two completely independent hardware components: linear axis and welding head are mechanically constructed into single demonstrator without any electrical interconnection. The only available connection is network connection to the SelSus cloud. Due to built-in intelligence, these components are capable to automatically identify all involved SelComp within the process, self-organize, and operate according to the process description provided by the SelSus cloud. Moreover, the involved SelComp equipment is capable to identify potentially dangerous situations and prevent failure by switching to limping home mode. Limping home modes are implemented for: (1) current sensor failure, (2) overloaded welding head, and (3) overload of electrode motor controller. In addition, the involved SelComps send to the SelSus cloud sensor and process state information for detailed condition monitoring.


Technical details
Welding head is equipped with linear axis motor driven by standard Beckhoff driver controlled by dedicated IPC (industrial personal computer). Temperature, voltage, current, and force sensors are mounted on the welding head and are permanently measured by a sensor node. The sensor node stores these data to local database and transfers the measured data to SelComp-IPC for analysis. The SelComp-IPC performs main analytic work. It stores history of last 1000 measurements and implements their permanent analysis, in order to detect possible dangerous process development. Based on this process diagnosis, decision module adapts operation of SelComp to prevent failure and notifies the SelSus cloud about possible delays or errors. Other intelligent SelComp equipment involved in the process also obtain this information from the SelSus cloud and can adapt their operation accordingly. 

Demonstrator benefits

  • The intelligent sensor based diagnosis module allows one to predict dangerous process development scenarios and prevent their further evolution. As a result, production failures are prevented
  • The efficient self-healing features contribute significantly to an increasing overall equipment effectiveness (OEE) as the nowadays implemented simple machine stopping is replaced by intelligent imping-home modes.
  • Due to cloud, based integration of SelComps, detailed system-wide process sensor monitoring is possible. This opens broad possibilities for e.g. quality control purposes or system-wide diagnosis.
  • Built-in intelligent self-organization mechanisms allow one to substantially facilitate process organization. Several SelComps can be mechanically mounted in production line without complicated cable interconnection. SelComps are capable of self-organization and operation according to the process description in the cloud.


TRL level: 6
Availability: Yes

Sensor Integration
Inotec - WDT Therm RF Sensor

Inotec - WDT Therm RF Sensor

INOTEC
INOTEC

Contact
Inotec limited & Barry Auty/Dale Read & 31 Ferry Road, Hullbridge, Hockley, Essex, SS5 6DN & barry.auty(at)inotecuk.com & Dale.read(at)inotecuk.com

Product description
The wireless data transmitter thermo-couples (WDTTHERM) v1.0 Product, has been developed for the SelSus project, to enable SelSus systems, to collect health data, without interfering with the existing manufacturing and control systems, providing an independent communications RF network, that is totally independent of clients IT and control infrastructures. We needed a real-time data transmitter that could collect health condition data from additional health sensors installed on to manufacturing systems and critical Asset systems to enable preventative failure and predicative data to be analysed via the SelSus Selcomp Bayesian models, to provide prognostic data for machines and assets, to increase the productivity and uptime of the manufacturing environment.

Service description
Inotec have concepted, designed, prototyped, tested and developed an RF sensor product (TRL9) to enable five analogue input signals and five thermo-couple signals to be monitored in real-time via our WDTTHERM product.
The WDTTHERM is battery powered (three-year minimum life), zero configuration required and communicates via ISM RF frequency band 868MHz, so the RF network can work independent of the client’s IT infrastructure or PLC network, to enable SelSus system to collect machine health data, without interfering with the manufacturing control systems, providing a totally independent real-time health monitoring system that can be connected to legacy or new manufacturing systems.

Technical details
RF product that works on the same RF communications network as current Inotec RF sensors, providing a range of products, that can real-time monitor, workplace occupancy, asset health and environments, the RF sensors work on their own RF network at 868 or 922MHz providing a global license free communication system that works totally independent of customers IT and PLC control infrastructure, the WDTTHERM product can connect to additional health sensors that are put onto the manufacturing machines, providing real-time data to the SelSus Selcomp to provide the analytics and prognostic data (Bayesian models) to predict failure and keep the machine operational during working periods. Inotec have set the sensor data up to provide local data or send to the SelSus cloud.

Product / service benefits
•    Provides real-time machine health data via five analogue inputs & five thermo-couples
•    Works independent of client’s IT and control system infrastructure (no overhead)
•    Provides SelSus system with a connection to legacy manufacturing systems
•    WDTTHERM will connect to the SelSus Selcomp for local processing and Bayesian modelling to enable predictive failures and prognostics to be performed
•    WDTTHERM connects to RF/Ethernet gateway and via 4G modem can send live health data to the SelSus cloud.


Technical data

Power Supply
1. PCB is powered by qty 3 'AA' Alkaline batteries giving an expected lifespan of up to 3 years.
2. PCB can also be powered by an external dc supply of between +5 and +24V applied to connector CON7.
•    CON7 Pin 3 = +5V to +24V dc input
•    CON7 Pin 4 = 0V
3. PCB is powered while programming through the connector CON4.
•    CON4 Pin 9 = 5V dc input
•    CON4 Pin 1 = 0V
4. Battery voltage is monitored through the software.
5. +24V is available via a MOSFET to power up external equipment from CON7 pin 1 up to 2A.

Signal Inputs

1. Up to 5 Analogue signals can be presented to connectors PL1 to PL5.
•    Pin 1 = +24V from external DC supply from CON7 pin 3
•    Pin 2 = Analogue input 0 - 10V or Digital Input 0 - 20V
•    Pin 3 = 0V or Ground

2. Up to 5 K Type Thermocouples can be presented to connectors PL6 to PL10.
•    Pin 1 = TC +
•    Pin 2 = TC -
•    Temperature range +0°C to + 700°C
•    Error range = +/- 3°C

3. The PCB has an onboard humidity sensor with the following specification:
•    Temperature range +5°C to + 50°C (41°F to 122°F)
•    Humidity range 10% RH to 90% RH
•    Error range = +/- 5% RH

4. The PCB has an onboard temperature sensor with the following specification:
•    Temperature range +5°C to + 50°C (41°F to 122°F)
•    Error range = +/- 1°C

Radio Frequency RF
1. The PCB has a 10mW onboard radio transmitting on the following unlicensed frequency bands:
•    868MHz
•    922MHz
•    434MHz (optional)

2. There are Green and Blue LEDs that are used to indicate the status of the PCB or its software.

TRL level: 9
Availability:
The WDTTHERM product is now at technical readiness level nine, we now have the WDTTHERM working on the SelSus demonstrator at Ford Motor Co. in Dunton on the Robotic RTV dispensing machine, which has been in operation since April 2017, providing real-time health data to the SelSus partners.

ISR – Sensor Cloud

ISR – Sensor Cloud

Monitoring Robotic Arm from Sensor Cloud

Contact
ISR - Porto & João Reis & Faculdade de Engenharia da Universidade do Porto
Rua Dr. Roberto Frias, s/n 4200-465 Porto Portugal & jpcreis(at)fe.up.pt

Product description
Sensor Cloud is a central piece that enables data storage, inter-SelComp communication and interaction with the end-users for monitoring, processing and orchestration of the Cyber-Physical Production System (CPPS) with its focus on Wireless Sensor Networks (WSNs). It is mainly an end-user interaction platform and data repository that aims to give historical and up-to-date information about all the sensing devices installed in the Manufacturing System, along with a set of tools to control and manage these devices according to market demands and product requirements. It has all sorts of graphical interfaces to integrate new sensing devices into the system in order to have a fast impression of the generated data. Additionally, Sensor Cloud is capable of processing these generated data by using software reconfiguration tools and ultimately control shop-floor devices based on the collected and processed information, making is a suitable framework for closed feedback-loop applications in manufacturing systems.

Product / service benefits
•    Transparent Shop-floor Representation
•    Run-time monitoring
•    Shop-floor Device Actuation
•    Graphical Definition of Closed feedback-loop

TRL level: 5

ISR - Sensor Integration

ISR - Sensor Integration

User Interface to insert meta-information to decode a ASCII message from Serial Port Com sensor integration

Contact
ISR - Porto & Ricardo Silva & Faculdade de Engenharia da Universidade do Porto
Rua Dr. Roberto Frias, s/n 4200-465 Porto Portugal & rps(at)fe.up.pt

Product description
The technology developed for flexible and easy Sensor Integration in the SelSus project is a cloud-based user interface solution that enables a person without or with little programming knowledge to specify a new Wireless Sensor Network (WSN) message descriptor / parser. The typical architecture for a WSN is composed of sensor nodes and gateways, being the sensor nodes the generators of data, and it needs to reach the gateway where it is then handled by the Sensor SelComp. The construction of this WSN message parser is not only suitable for new nodes to be integrated in the network, but also to extend existing ones to support new sensors in the node itself. This way, a person can graphically integrate new sensors into the network from different manufacturers, and readily start collecting data from its manufacturing process. We see this technology as a key enabler not only for the CPPS for manufacturing processes, but also for Internet of Things applications where sensors lie in its core technology.

Product / service benefits
•    Reduction of the production cost
•    Sensor Integration with reduced technical knowledge
•    Short time to market
•    Systematic approach to system construction

TRL level: 5

ADP - Sensor data processing and fault detection

ADP - Sensor data processing and fault detection

Sensor Hardware integration
KALMAN filtering of raw sensor data
Sensor Data Visualization
Sensor Data associated with Tool Information
Sensor Data associated with Material Tracking Events
 

Contact
advanced data processing GmbH
Breitscheidstr. 78
D-01237 Dresden
Dr.-Ing. Knut Voigtländer
voigtlaender(at)adp-dresden.de

Product description
Adp has developed a toolkit for message oriented sensor integration, sensor signal processing including material data merging, process visualization and multivariate fault detection
Technical details
The developed concept for sensor integration is based on a message oriented middleware and comprises the definition of well defined Documents transferred as JSON frames for universal loosely coupled sensor integration including data processing and storage within a central non-sql database. An APP-oriented set of services, which is pluggable within the middleware, provides KALMAN Filter based sensor signal processing and multivariate fault detection as well as merging of sensor and material data. Using additional message oriented query services, data visualization components are provided.

Product / service benefits
•    universal loosely coupled sensor integration
•    plug and play data storage and visualization
•    sensor signal filtering for noise removal and significance interval estimation
•    multivariate process supervision regardless of number and content of parameters
•    enabling of product quality assessment by merging tool sensor data with material tracking events

TRL level
TRL 7 (Prototype more than one year running)

Data Management and analysis
GAMAX - SelSus FMEA Data Manager

GAMAX - SelSus FMEA Data Manager

SelSus FMEA Data Manager

Contact
[Gamax kft.] & [Tamás Szabó] & [szabo.tamas@gamax.hu]

Product / service description
The programming language is JAVA and we used the VAADIN framework. It was an important requirement that the xls file transformation, has to happen on the server side, the parametering and the searching of the components on the client side. Moreover the user has the opportunity to export the calculated results into an hkb document.

As it is visible on the picture above, the FMEA Data Manager can be divided into two parts. Client side and server side. The whole process starts with the users uploading their data they store in xls file. The xls parser on the server side reads the data after the upload, and illustrates it in a table. Depending on data and demand there is a chance for the transformations to be made. After (or without) these modifications the list of components belonging to the modified data can be made. When the final list is ready the components needed for the calculations can be chosen. There will be a detailed explanation in the technical details (server side) and in the illustration (Client side) part.

Technical details
The engine was developed in JAVA EE8 with the help of the Vaadin framework. The data (xls file) uploaded by the user is being processed and is stored temporarily on the server. We are making the setup of the necessary components taking the parameters received from the user on the server side, and the webpage creates a JSON file form them. Then it is forwarded to the Hugin Bayesian Network server with the help of the REST Web Service. Then we receive back the created hkb file as a reply that we can save.

The operation of the Back-end can be seen on the above picture. The process starts with uploading the user’s data. The xls parser reads it and store it temporarily. Then it displays the stored data in a table.  There is the chance to modify any data that are being sent back to the server, this is what we can see in the “xls parsing” bubble. The process (modifying data) can be repeated until needed. And when all the data is appropriate, the final component list can be generated, from which the necessary elements can be chosen.

Product / service benefits
•    Online FMEA data management
•    Quick search in FMEA data
•    Platform independent

HUGIN - Application Programming Interface on the Android platform

HUGIN - Application Programming Interface on the Android platform

Contact
HUGIN EXPERT A/S
Anders L Madsen
Gasværksvej 5, 9000 Aalborg, Denmark
anders@hugin.com

Product / service description
An Application Programming Interface (API) to the HUGIN Decision Engine running on the Android platform.
Binaries for all 32-bit Android CPUs armeabi, armeabi-v7a, x86 and mips are included. Lowest Android API level supported by the HUGIN API would be '9' as the HUGIN API does not depend on any Android libraries other than a few default classes available in any recent Java JDK.
HUGIN is powerful analytic software for developing and deploying decision support systems for reasoning and decision making under uncertainty. HUGIN software is based on Bayesian networks and influence diagram technology and it is the ideal choice when uncertainty must be represented and handled efficiently.

Technical details
As Android applications are written in Java, the regular HUGIN Java API has been ported from the PC Oracle Java platform to the Android platform. This makes it possible to integrate HUGIN models into applications running on the Android platform in a way equivalent to how HUGIN models can be integrated in application running on the PC platform, i.e., the control software of a piece of equipment.

Product / service benefits
HUGIN Decision Engine running on the Android platform

TRL level: 8
Availability: 2018

HUGIN for Raspberry Pi

HUGIN for Raspberry Pi

Contact
HUGIN EXPERT A/S
Anders L Madsen
Gasværksvej 5, 9000 Aalborg, Denmark
anders@hugin.com

Product / service description
HUGIN for Raspberry Pi including HUGIN Graphical User Interface and Application Programming Interfaces for the HUGIN Decision Engine for C and Java.
The HUGIN Graphical User Interface and Application Programming Interfaces for the HUGIN Decision Engine for C and Java running on the Raspberry pi platform.
HUGIN is powerful analytic software for developing and deploying decision support systems for reasoning and decision making under uncertainty. HUGIN software is based on Bayesian networks and influence diagram technology and it is the ideal choice when uncertainty must be represented and handled efficiently.
 
Technical details
•    HUGIN for Raspberry Pi is available as a 32 bit version.
•    Included API: Java

Product / service benefits
HUGIN Graphical User Interface and HUGIN Decision Engine running on the Raspberry pi platform

TRL level: 8
Availability: 2018

HUGIN - Enhanced HUGIN software with Online EM parameter learning and data analysis capabilities

HUGIN - Enhanced HUGIN software with Online EM parameter learning and data analysis capabilities

Contact
HUGIN EXPERT A/S
Anders L Madsen
Gasværksvej 5, 9000 Aalborg, Denmark
anders@hugin.com

Product / service description
HUGIN software (Graphical User Interface and Decision Engine) enhanced with functionality for Online EM parameter learning and data analysis capabilities.
HUGIN is powerful analytic software for developing and deploying decision support systems for reasoning and decision making under uncertainty. HUGIN software is based on Bayesian networks and influence diagram technology and it is the ideal choice when uncertainty must be represented and handled efficiently.
With HUGIN software, Component Models can be constructed from domain knowledge, from data or from a combination of the two. Once models have been constructed they can be utilized for fault diagnosis, system monitoring or for maintenance/service scheduling. Furthermore, the tool is equipped with methods for tuning an existing model while data is processed in a live production environment for continuous improvement. HUGIN software supports modular model construction meaning that models constructed on the component can be combined into a higher level perspective in a factory level model.

The software is equipped with APIs for some of the most widely used programming languages and a web service API hence allowing development of desktop applications as well as web applications.

Technical details
The HUGIN GUI is available on:
Windows, Linux in 32 and 64 bit versions and on Mac in a 64 bits version.
The available APIs are: C, C# (on Windows), C++, and Java.

The web service API is RESTful and it is distributed along with a Java Script API and a widget library allowing development of web applications.

Product / service benefits
•    Enhanced functionality for improving model performance by adapting the parameters of the models as data cases are being processed.
•    Enhanced functionality for data analysis in the HUGIN Graphical User Interface.
•    Possibility to extract data for analysis from a local database or from the SelSus Cloud.
•    Online probability calculations (propagation) of live data in the SelSus Cloud (calculated as new data appear).

TRL level: 8
Availability: 2018

ISR - Dynamic Modular Software Reconfiguration

ISR - Dynamic Modular Software Reconfiguration

Directed Acyclic Graph represenation of a Sensor SelComp Reconfiguration

Contact
ISR - Porto & Luis Neto & Faculdade de Engenharia da Universidade do Porto
Rua Dr. Roberto Frias, s/n 4200-465 Porto Portugal & lcneto@fe.up.pt

Product / service description
The Dynamic Modular Software Reconfiguration is the proposed solution embedded in the Sensor SelComp to be highly reconfigurable and to quickly adapt to industrial demands. Therefore, it provides consistent mechanisms to configure, deploy and dynamically reconfigure the Sensor SelComp source code responsible for data processing, which traduces in a component-based middleware. It is component-based because each piece of software that can be reconfigured at the Sensor SelComp level is seen as a component that can be easily updated or exchanged.

Product / service benefits
•    Component reuse
•    Reduction of the production cost
•    Reconfiguration in runtime
•    Short time to market
•    Systematic approach to system construction

TRL level: 5

IPA - Decision support system

IPA - Decision support system

Overview of the decision support system

Contact
Fraunhofer Institut
Produktionstechnik und Automatisierung
Nobelstraße 12
70569 Stuttgart
Michael Kempf
Michael.Kempf(at)ipa.fraunhofer.de

Product / service description
With the aid of a decision support system, the best strategy for effective and efficient repair and renovation can be determined. The decision making process is supported by Bayesian models, such as Bayesian networks and influence diagrams, as well as by a discrete event simulation model, which will contribute to finding an optimal remedy strategy to guarantee upmost yield expectations.

Technical details
The decision support system (DSS) consists of several subsystems – namely a diagnostic module (DM), an influence diagram (ID) and a discrete event simulation (DES) – which interact between each other:
The user, namely the machine operator or the maintenance manager, supplies requests to the DSS. He might have a problem with the equipment and does report this to the DSS together with some observations that were collected. The DSS, in turn, triggers the diagnostic system and transfers the relevant observations there. During the root cause analysis process the diagnosis module may require some additional sensor information, which will be delivered from a database or the sensor cloud. Finally it reports the user about the most probable failure causes via the decision support system.

The user might also want to see the health status of the system or one of its components. Thus the DSS will cause the influence diagram (ID) to predict the lifetime of that specific component under a specific repair strategy. The ID is based on a degradation model, which is periodically recording sensor data to monitor the level of deterioration of the component. Depending on the condition of the component maintenance actions must be initiated when necessary. In case a maintenance activity has to be carried out, the influence diagram has to suggest the type of maintenance to be performed as well as the times of these activities. Consequently, the machine operator knows which maintenance activities have to be executed within the next few shifts.

The discrete event simulation (DES) module is capable to evaluate different detailed sequences of maintenance actions (e.g. replace part or component, maintain component, do nothing) and to calculate the impact thereof as a basis for decision making for the machine operator. The impact has to be defined appropriately for example with the help of certain key performance indicators (KPI). That is, the DES can supply the DSS with alternative sequences of repair and maintenance actions and their respective impacts.

Product / service benefits
•    The main benefits of the system are reduced downtime and higher productivity, as well as higher value added from the same facilities.
•    The decision support system is able to find an optimal strategy for maintenance and repair activities related to components and systems in a manufacturing environment.
•    The optimization is accomplished by using several KPIs (key performance indicators) like the expected lifetime or the degradation of specific components, which all have an impact on the Overall Equipment Effectiveness (OEE).

TRL level: 5