linked engineering and manufacturing -...

EPFL/STI/IGM/ICT for Sustainable Manufacturing @ EMS Summit 2016/ 05.10.2016 http://ict4sm.epfl.ch/ 1

Linked Engineering and Manufacturing

Dimitris Kiritsis

EPFL, ICT for Sustainable Manufacturing

[email protected]

http://ict4sm.epfl.ch/

mailto:[email protected]


Some words about DK…

• STI-IGM

– Prof. ICT for Sustainable Manufacturing

• School of Industrial Engineering – Grenoble INP

– Member of the Board (2nd mandate)

• EC – AG LEIT-NMBP

– Individual Member

• IFIP WG5.7 – Advanced Production Management Systems

– Chair

• Member (representing EPFL)

• Member of the advisory board www.cl2m.com


http://www.cl2m.com/


ICT for Sustainable Manufacturing

Optimisationof Lifecycles

of Product Systems

SYSTEMS ENGBusiness Process ModelingFunctional Modeling…

OPERATIONS RESEARCHPetri NetsGraph TheoryStatistical Analysis…

Sustainable Manufacturing involves a complex system of Products, Processes and Ressources need for multidisciplinary research

Main disciplines:•MECHANICAL ENG

•MANUFACTURING ENG

•LIFECYCLE MANAGEMENT

COMPUTER SCIENCEOntology Engineering / Semantic WebInternet of ThingsKnowledge Management /AI

Data miningData AnalysisReasoning…



What must be “sustained” in manufacturing ?

• Performance, Quality and Safety

• Natural resources and Environment

• Lifecycle thinking

• Human Capital and Education

• Innovation



Lifecycles

MOL

Assembly

Part Manufacturing

Procurement

BOL

Detail Design

Life Cycle Planning

EOL

Disposal

Parts/Component reuse

Collection (Reverse logistics)

Life Cycle Conceptual Design

Distribution (Logistics)

Maintenance (Service)

Use

Material recycleDesign Production

Lifecycle DB/KB

PLMPlatform

Lifecycle dataLifecycle data

Lifecycle data

Remanufacturing

(DisassemblyRefurbishment,

Reassembly,Test, etc)



MOL

Process

Resource

Product

DESIGN

Process

Resource

Product

PRODUCTION

Process

Resource

Product

EOL

Process

Resource

Product

Materials

Disposal

Recycling

Re-mfg.

Service Re-use

Closing the information loops



CL2M: Closed Loop Lifecycle Management for Industry 4.0The PROMISE Concept

- Data

Embeddeddevice

- Diagnosis- Decision making

- Data

- Info

- Info

- Info request

- Info- Advice

PEID(CPS enablers)- Sensors- Tags- Memory- Data

Wireless comm.(short distance)

Internet(long distance)- Semantics

- Ubiquitous computing

- PLM agent- Knowledge agent- DFX agent

Producer’sPDKM

© EPFL

- Product/Asset with

- PLM Agent

In the Cloud- PLM system- KB repositories



Applications

PEID

RFID

Middleware

Data Services

EmbeddedSystems

Read/Write

Business

Processes DecommissioningBusiness

Processes

ProductsProducts

Design/Production

Design for X PreventivePredictiveMaintenance

Tracking &Tracing

Adaptive

Production Effective Recycling

Semanticenrichment

Dispatching

NotificationsPEID

Management

AnalyticsKnowledge

ManagementDecisionMaking

A system architecture for Industry 4.0



CL2M system architecture for Industry 4.0

Data analytics and transformations (Middleware)

Data gathering (PEID)

Business applications

BOL MOL EOL

(1) Horizontally closed

Seamless Lifecycle Data-Information-Knowledge Transformations

- Gathering data- Feedback

- Information, Knowledge- Feedback

(2) Vertically closed

Applications

PEID

RFID

Middleware

Data Services

EmbeddedSystems

Read/Write

Business

Processes DecommissioningBusiness

Processes

ProductsProducts

Design/Production

Design for X PreventivePreventiveMaintenance

Tracking &Tracing

…Adaptive

Production …Effective Recycling

Semanticenrichment

Dispatching

NotificationsPEID

Management

AnalyticsKnowledge

ManagementDecisionMaking



PEID & CPS

IoT

Key Enabling IC Technologies for CL2M



PEID



PEID architecture

PEID

Long range communication

Short range communication

Product identification

Data storing

Power unit

Sensor reading unit

Memory

EPC

Radio wave

Data processing/diagnosis

Internet/GSM/GPRS/Mobile phone

Embedded SW

Data gathering

Power management

PassiveRFID

ActiveRFID On-board

Com.

• PEID stands for ‘Product Embedded Information Device’.• Depending on product types and application area, there are various kinds of information devices built in products to gather and manage product information.

Source: Kiritsis, ch. 5.4, Springer Handbook of Automation , 2009



IoT



The new IT pyramid

• Each Enterprise has «his own» pyramid.• All parts not always needed.• PLM, CPS, ERP and IoT play key roles.



Systems of Systems

System?

System?

System?System?

System?

System?

System?

System?

System?



IoT today

• Sensor data pushed to virtual/real servers

• Data collected into vertical silos

• M2M communication limited to local networks



IoT today: challenges

• Huge amounts of data (Big Data) collected that is difficult to use

• Interoperability is hard to achieve, even with REST APIs

• IoT should not be just local M2M

• IoT should not be just machine-to-cloud communication



IoT with Open Group Standards

• Systems rather than sensors

• Horizontal integration as easy as vertical integration

• Connections created without programming as needed

• Establish two-way, time-limited information flows between trusted entities and the physical product(s)



Standards of IoT Work Group

• Published by The Open Group on October 16th, 2014

• Open Messaging Interface (O-MI): communication

• Open Data Format (O-DF): payload

• May be used independently of each other (as HTTP and HTML)

• Publish available information and services

• Discover availableinformation and services

• Read/write of immediate and historical information, alerts, other events, …

• Subscribe to information using Observer Design Pattern

• Specified using XML schema

• Can be transported by ”any” underlying protocol: HTTP, HTTPS, FTP, SMTP, XMPP, file transfer, USB sticks, plain sockets, …

• Usable as Standardized IoT REST API as well as M2M, M2S etc.

Format

XML,XML schema, JSON,HTML

O-DF

OSI ModelLower Layers

Communication

HTTP(TCP/IP, SOAP, SMTP, …)

O-MI

Web

Application

Ap

plic

atio

n L

ayer

IoT

http://www.opengroup.org/getinvolved/workgroups/iot




Open Lifecycle Management

• O-LM: Open Group (Things) Lifecycle Management: Exchange information about ”Smart Things (or smart products)” that are generated across the whole Lifecycle Phases, using a well defined format

• Currently used to exchange data about components & materials, operations and transformations and data acquired by the physical product

• Format defined and approved internally to IoT WG, formal accompanying document in progress





Systems of Systems: Closed-Loop Lifecycle Management (CL2M)

Instance

Manufacturing 1

Manufacturing 2

User 1

User 2

Recycling

InstanceInstance

Instance

Instance

Product informationWhere? In one or many places?Type/version/instance-specific?

Information queries/updates

Designer 1

Designer 2

”Thing”

”Thing”

”Thing”Business

idea/planIoT

• Lifecycle view: IoT is about managing all information about any product/Thing

• Information is Distributed over Systems (devices, servers, applications, ...)

• Information is Distributed over Organizations (companies, individuals, authorities, …)

• Product (and its parts) are unique instances

• How manage identities, access rights, …?

• IoT should provide necessary Capabilities for CL2M

Red arrows: O-MI & O-DF

www.cl2m.com


http://www.cl2m.com/


Applications in current H2020 & FP7 projects


http://images.google.com/imgres?imgurl=http://www.canadiandesignresource.ca/officialgallery/wp-content/uploads/2006/04/bombardier11.jpg&imgrefurl=http://www.canadiandesignresource.ca/officialgallery/?p=412&h=403&w=410&sz=107&hl=en&start=1&tbnid=R1wEQ-b0nBw49M:&tbnh=123&tbnw=125&prev=/images?q=Bombardier+transportation+logo&gbv=2&ndsp=18&svnum=10&hl=en&sa=N

http://images.google.com/imgres?imgurl=http://www.canadiandesignresource.ca/officialgallery/wp-content/uploads/2006/04/bombardier11.jpg&imgrefurl=http://www.canadiandesignresource.ca/officialgallery/?p=412&h=403&w=410&sz=107&hl=en&start=1&tbnid=R1wEQ-b0nBw49M:&tbnh=123&tbnw=125&prev=/images?q=Bombardier+transportation+logo&gbv=2&ndsp=18&svnum=10&hl=en&sa=N


10/11/201623

Linked Engineering and Manufacturing

Data Federation

• Relevant information, across trusted sources in the product lifecycle

• Format, location and time independent

Context-driven access and analysis

• Collaborative charting and simulations

• Integrated information analysis

User collaboration

• Extension of lean engineering principles

• Collaboration workbench for effective internal and external collaboration

Feedback into existing systems

• Tight connection to the federated systems

• Push-back of enriched information



10/11/2016Adrian Mocan, SAP AG24

LinkedDesign Achievements

• Problem and role-specific information presentation

• Modular, reusable, self-contained logical info units

• Collaboration along the full product lifecycle

• Enabling the human factor

• Domain specific ontologies

• Novel use of algorithms for LCC and LCA

• Domain specific techniques

• Data federation by semantic mediation

Data Integration

Data to Information

Information to Presentation

Knowledge Integration



LinkedDesign Ontology

10/11/2016 25

isSpecialization

isSpecialization

isSpecialization

Product DefinitionLifecycle Cost Assessment

Ontology

Product DesignDesign Knowledge Ontology

Product manufacturing Quality Control Ontology

Generic and lightweight Reusable in different

industrial applications



LinkedDesign Upper Ontology - LDO



LinkedDesign Network Ontology



10/11/2016Adrian Mocan, SAP AG28

Use cases

ApplicationPrototypes

Demon-strators

Manufacturing Design based on

Plant Lifecycle Costs

Knowledge Reuse and Collaboration in Automated Design

for Offshore Engineering

Metrology-driven Manufacturing in the Automobile Industry

Automated Design with

Virtual Obeya

Lifecycle Design Optimisation

Manufacturing Quality Control



COMAU

Product LifecycleOptimization Tool

Data source: Customer Specs.Sensor values

Implemented usingSWRL

Base set of 20 rulesdefining calculationof parameters and relevant set of design constraints

- Support LCA of production line

- Monitor the performance of the line in operational phase

- Calculation of MTTR, MTBF, availability, down time...

- Optimal values for parameters: MTTR, MTBF…

Use case requirements

KnowledgeCapture

KnowledgeConceptualization

ImplementationInteraction with

LEAP

Inference Analysis

10/11/201629

VW-TRIMEK

- Identification of risky parts

- Crack detectionin hot stampingprocess

Optimal values for each parameter: Oven temp., Dwelltime, Insertion and extraction temp., Pressing force, etc.

Production Line Real Time Monitor

Data: Sensor values

Base set of 20 rules defining tolerable deviation of real measured values compared to optimal

Implemented usingSWRL

AKER

- Semanticenrichment of engineering standards

- Recommendationbased on the business context

- Industrialstandards handling and application

- K-briefmaintenance

Base set of 10 rules for K-briefs and set of 27 rules as an examplefor standards usage

Implementedusing SWRL

Concept level- Virtual K-Briefs

Data: Design parameters, K-Briefproperties/



Inference Analysis

Instance Importer

Semantic Reasoner

Base set of rules

If|HotStamping.Optimal.OvenTemp[14]-HotStamping.AsMeasured.Heating.Temperature[14]|>20C then ConsideredRisky=1

If HotStamping.AsMeasured.Heating.DwellTime > HotStamping.Optimal.DwellTimethen ConsideredRisky=1

If |HotStamping.Optimal.BlankholderForce -HotStamping.AsMeasured.Heating.BlankholderForce| > 20 then ConsideredRisky=1

If HotStamping.Optimal.InsertionTemp -HotStamping.AsMeasured.Heating.InsertionTemp < -20C then ConsideredRisky=1

If |HotStamping.Optimal.PressingForce -HotStamping.AsMeasured.Heating.PressingForce|>50KN then ConsideredRisky=1

If HotStamping.Optimal.CycleTime - HotStamping.AsMeasured.Heating.CycleTime > 2s then ConsideredRisky=1

If |HotStamping.Optimal.ExtractionTemp-HotStamping.AsMeasured.Heating.ExtractionTemp|>20°C then ConsideredRisky=1

If HotStamping.AsMeasured.TimeStamp > 0.95*Tool.MaintenancePeriod then ConsideredRisky=1

…

Production Line Real Time Monitor

1

2

3

4



Data correlation and pattern discovery

Proposed methodology

Pre-Processing

• Dealing with missing values , categorical attributes , outliers , etc.

• Supported using LDO and inconsistencychecking

Correlationdetection

• Calculation of correlations between attributes

Modeling

• Application of classification algorithms for pattern discovery (DT; K-NN; SVM)

Reporting

• Generating text files that contain relevant findings



Ontology for quality control in hot-stamping



Sources of information & data

Two sources of information about source of defects

Workers experience Heuristics

Data analytic and data mining of existing data bases

Both are integrated to create automatic alert system for increased quality level risk



Toolset for correlation analysis

• Statistical tests are performed on all pairs of attributes

1. Numerical & Numerical : Pearson’s coefficient

2. Numerical & Categorical : Davies-Bouldin index

3. Categorical & Categorical : Chi-squared test

• Threshold of “significance” are predefined as threestandard deviations distant from mean value of allcorrelations for given pair



Architecture of Data modeling module



Decision tree for quality control in hot-stamping

Data analytics is performed directly on instances populating the ontology

High risk ! High risk !Norisk

High risk !Norisk

Yes

Yes

Yes

Yes

No

No

No



The intelligent consumer/user in the loop …in the Product Lifecycle loops

Designer / Producer

Consumer /User

ServiceProvider

D, I

K

D, I

D, I

IoT&

PEID(CPS enablers)



38

The FALCON H2020 FoF project

FALCON will define and deploy a Virtual Open Platform for Product/Service Design, Manufacturing and Lifecycle Management

FALCON will investigate how user experiences and usage data collected via IoT and social media can be deployed for the improvement and design of product-services.

The Virtual Open Platform platform will incorporate new simulation techniques and benefit from a semantic representation of PLM information for cross-sectoral search



• FALCON Virtual Open Platform• Open system-architecture defined • Systems of Systems Approach • Based on semantic technologies

• Initial FALCON Ontology• Initial definition of upper ontology covering PSS

knowledge domain• Initial definitions of domain-specific ontologies for

FALCON business scenarios

• GUI Mockups & Business Stories• Development of detailed Business Stories• GUI Mockups facilitate understanding of functionality

• Software prototypes of Functional Modules• E.g. data federation module for knowledge acquisition

from static PUI sources (PEIDs, social media, etc.)

39

FALCON Achievements (…so far)

VOP PUI Manager

Aut

horiz

atio

n an

dA

cces

s C

ontr

ol

SPARQL EndpointPub/Sub Query Endpoint

JDBC Endpoint

Linking Component JDBC Interface

Triple StoreRDF Stream

View ManagerFALCON Data

Federation Module

FALCON KBE

Mapping Tool

FALCON KCCM

Management Tool

Social Media Wrapper

Legacy System

Wrappers

KCCM

PEID Stream Wrappers

Structured Data Source

Wrapper

Social Media Stream

Wrapper

FALCONIdea Manager

FALCONSimulation and

Forecasting Manager

FALCONPUI Query Builder

FALCONData Visualisation

Module

FALCONData Export

Module Mediator

ConfiguratorStream

Manager

FALCONOntology

VOPLogin

3rd Party LCA Tools

CAx ToolsSimulation

Tools

OtherTools

FALCON PUI Alert Module

PEID Wrappers

VO

P In

fras

truc

ture

S

ervi

ces

FALCON Open API

Virtual Open Platform – Final Draft Architecture (30.06.16)

WP2

WP3

FALCON Collaborative PSS Design Solution 3rd Party Software

WP4

WP1

FALCON Virtual Open Platform Core 3rd Party

FALCON PUI Wrappers

Legend

Interface

Defines/uses

Brown GoodsWhite Goods

Healthcare Products

Clothing TextilesHigh-tech Products

isSpecialization isSpecialization

isSpecialization



40

Business Scenarios

Clothing Textiles

Business Motivation:

Integration of data sources and creating valuable information to help offering customers

A better product-service quality through optimization of planning, development and testing processes.

New innovative services for existing products.White & Brown Goods


Improve collection of feedback from customers from usage phase and even from new collection proposals

Improve exchange of information both from social networks, marketplaces and our e-commerce site.

Collecting better information about required fitting of the garments.



41

Business Scenarios

High-tech Products


Increase technical reliability of the devices,

Enable comprehensive diagnostics, and

Enable user-friendly devices and services

Healthcare Products


Identify potential demands for equipment before the bid is out (customising needs)

Increase the Competitiveness of Metrology Solutions improving their functionalities and ease to use.

Reinforce Client “loyalty“ offering metrology services

more effective.

Improve the Calibration, Maintenance & Verification processes and Enhance Machine Programming & Report Content Definition.



Taking the LEAP …

10/11/201642


linked engineering and manufacturing -...

Documents