TU Kaiserslautern Lehrstuhl integrierte Sensorsysteme · beim Kunden zu einem Mitarbeiter im Unternehmen. Datenvisualisierung Produktionsanlagen Visualisierung wichtiger Daten einer

TU Kaiserslautern – Lehrstuhl integrierte SensorsystemeProf. Dr.-Ing. König

Kunststofffolien - Industrie 4.0

Dr.-Ing. Michael Kohlert / 12.2017Head of IT & Automation bei Mondi Group

CONCEPT

Mondi Gronau GmbH

2

3

Industry 4.0: Overview

Embedded

Systems

CPS

Smart

Factory

IT

SecurityRobust

Network Cloud

Computing

Robust NetworkMobileBroadband

Cloud ComputingReal-Time DataBig DataAppsIT Security

Data ProtectionInformation Security

Embedded Systems CPS

Machine-2-MachineSensors & ActuatorsIntelligent Products

Smart Factory

VirtualizationHuman-Machine-InterfacePlug & ProduceAutomation

Source: According to Bitkom

Internet of Things

Dr. Michael Kohlert 18/12/2017

4

Industry 4.0: Selection

Internet of Things

Big Data

Virtualization

Artificial Intelligence


5

Industry 4.0: Internet of Things

Image: Google Images

Image: Scott Bedford/ Shutterstock


https://www.google.de/url?sa=i&rct=j&q=&esrc=s&source=images&cd=&cad=rja&uact=8&ved=0ahUKEwjbwcrk6IXSAhXjllQKHRC8CtMQjRwIBw&url=https://computingcccu.wordpress.com/2016/02/23/internet-of-things-iot-how-private-is-your-private-life/&bvm=bv.146496531,d.d24&psig=AFQjCNHRvnnkXXVzOrHmtnPaBq4jUMkS2w&ust=1486826014770828

6

Industry 4.0: Internet of Things

Blockchain Technology

Image: agenda.weforum.org

Cyber Security

Image: IoT Analytics GmbH Image: NCTA

Connected Devices

AI Cloud Services

Image: Bloomberg BETA

Standards in Connectivity

Image: Worldsensing SLImage: Institute for the Future for University

of Phoenix Research Institute

Talent Skills


7

Industry 4.0: Big Data - Scenarios

Image: IBM

Asset-based: Predictive Maintenance for Smart Equipment

Process-based: Process Optimization

for Smart Factory

Image: Forschungsunion

Wirtschaft und Wissenschaft

Image: Harvard Business Review

Product-based: Quality Improvement for Smart Product

Supply-chain-based: Cross-factory collaboration Person-based: Control-on-the-go for Smart Human

Image: Zetes Image: Xively


8

Industry 4.0: Future Trends

Communication

Virtual Reality

1952 1999 2016 2040

1952 1999 2016 2040


AI

9

Industry 4.0: Augmented & Virtual Reality

What is VR ?

Spatial Visualization of Informations in Digital Models

& Interaction (Engineering, Design, Safety, Sales,

Maintenance)

What is AR ?

Overlapping of Virtual and Real Objects

with Geometric Fitting

Image: Robert Bosch GmbH Image: Microsoft Corporation

Image: The Coca Cola Company Image: Mondoworks

Image: Siemens AG Image: Illogic S.r.l.

Image: Virtualware 2007 S.A. Image: RenderSide


10


Use Cases for VR ?

Training, Virtual Roundtrips, Simulation,

Logistics

Use Cases for AR ?

Maintenance & Warehouse

Image: K2 Image: BMW AG

Image: WaveOptics Image: Daimler AG

Image: WordPress Image: Leyland Trucks Ltd.

Image: RWTH Aachen Image: SEP Logistik AG


11Dr. Michael Kohlert 18/12/2017

Safety, Health & Environment

• Schulung durch Visualisierung von Gefahrenstellen im Produktionsbereich

Fernwartung

• Remote Support bei der Wartung von Produktionsanlagen in Kooperation mit dem Hersteller

Besuchertour

• Zusätzliche Datenvisualisierung für den Kunden bei einer Produktionsbesichtigung

Prozessschulungen

• Prozesse werden über eine Datenbrille visualisiert und mit virtuellen Objekten unterstützt.

Kundenservice (Technischer Remote Support)

• Geographisch unabhängige Unterstützung bei der Problemlösung durch Video Live Stream Übertragung aus der Perspektive eines Mitarbeiters beim Kunden zu einem Mitarbeiter im Unternehmen.

Datenvisualisierung Produktionsanlagen

• Visualisierung wichtiger Daten einer Produktionsanlage über eine Datenbrille

Instandhaltung/Wartung

• Visualisierung wichtiger Dokumente wie Schaltplänen, Wartungsplänen usw. über eine Datenbrille

SOP – Rüsten, Reinigen, Einrichten

• Prozessunterstützung bei Rüst-, Reinigungs- und Einrichtungsprozessen von Produktionsanlagen


1218/12/2017

Daten

Dr. Michael Kohlert

Alle benötigten Daten

Good Product

SAP [100 %]

Produkt

Einstufiger Überblick

Bei einer Betrachtung der SCM über

mehrere Prozesspunkte durchläuft ein

Produkt weitere Datenquellen mit dem

gleichen Aufbau.

QS Online

[56,8%]

MES [100%]

QS Offline [4%]

SPS [96%]

Lieferant,

Kunde [0%]

Dr. Michael Kohlert

Übersicht Anbindung

Good ProductStatus Auslesen

Lieferant

Kunde

QS Offline

Dickenmessung

Optische

Kontrollsysteme

MES, SAP

SPS 70/72 96 %

72/72 100 %

26,5 %

87 %

4 %

0 %

0 %

20/ 23

2/ 50 (100)

0/ 50

0/ 20

9/ 34

Dr. Michael Kohlert

1518/12/2017

3 Phasen der Datennutzung

TasksMonitoring von wichtigen KPIs

STEP1-Visualisierung

Dr. Michael Kohlert

16

Industry 4.0: Big Data - Machines

Machine Parameter

Measurement M127

MDE Server with

SQL Database

(PMDE2 Oracle-

DB 11g

Enterprise)

Industrial Ethernet

SQL query triggered

(every 60 seconds)

Result data set returned to

select queryProcess Control Chart

displayed through

Webpage/HTML refreshed

every 60 seconds

Process Control Chart displayed on Mobile

Phone, displaying following information:

• Data captured every minute

• 1h data range (3.600 data points)

• 1h mean (calculated from 3.600 data points)

• Target value (predefined based on results from

data analysis and experience)

M127 Machine

Data Collection

(Beckhoff CX9010)

OPC UA

TCP/IP

Client TCP/IP Server

Image: Mondi Group


Thickness Measuring Optical Control System

1718/12/2017

Fie

ldE

TL

Dashboard

● Visualisierung

● Monitoring

● Speicherung

● Verarbeitung

● Aggregation

● Vorverarbeitung

● Erfassung

Industry 4.0: Informationsverarbeitung

180

bar

200

bar

220

°C

200

bar180

bar

Sensor kSensor 7Sensor 6

200

bar

+ Machine

180

bar

+ Machine+ Timestamp+ Timestamp+ Order+ Order

…

…

Dr. Michael Kohlert

1818/12/2017

Database

Machine

Interface

Processing Data

Facts & Figures: Measurement

Dr. Michael Kohlert

1918/12/2017Dr. Michael Kohlert

2018/12/2017

OsiSoft PI Quality Window Statistica

Dr. Michael Kohlert

2118/12/2017




TasksInformation über

Prozessverbesserungen

STEP2-Empfehlungswesen

Dr. Michael Kohlert

Process Monitoring Algorithms and Software

Sensor Data

(10-100 /plant)

Quality State

Which sensor measurements indicate machine failure?

Quality Data (OCS, Laboratory)MES, SAP

PLC

Dr. Michael Kohlert

Basic Workflow

1.

Process Monitoring Algorithms and Software

Evaluate Model

Fit Model

Choose Algorithm

Preprocess Data

Choose Model

Make Predictions

Dr. Michael Kohlert

Process Monitoring Algorithms and Software Pre-Processing

Sensor data and quality states are aggregated (per time stamp)

Sensor Data

(10-100 /plant)

Quality Stateupdate ~ 60 min.

Dr. Michael Kohlert

Process Monitoring Algorithms and Software -Train a prediction model

Basic Workflow

2.

Evaluate Model

Fit Model

Choose Algorithm

Preprocess Data

Choose Model

Make Predictions

Dr. Michael Kohlert

Process Monitoring Algorithms and Software–Train a prediction model

Possible Classification Methods

Discriminant Analysis

Dr. Michael Kohlert


Basic Workflow

3.

Evaluate Model

Fit Model

Choose Algorithm

Preprocess Data

Choose Model

Make Predictions

Dr. Michael Kohlert

Process Monitoring Algorithms and Software–Train a prediction model

Fit model based on historic data

Training Data

e.g. 60% of

historic data

(3 months)

PredictionModel = fitctree(PARAMETER, STATE)

Dr. Michael Kohlert


Basic Workflow

4. Evaluate Model

Fit Model

Choose Algorithm

Preprocess Data

Choose Model

Make Predictions

Dr. Michael Kohlert

Process Monitoring Algorithms and Software –Train a prediction model

Validation

Data, e.g.

40% of

historic data

(3 months)

predictedState

1

1

1

1

2

2

1

PredictionModel

Misclassification rate 1 of 7: 14.28 %

predictedState = PredictionModel(Parameter)

Dr. Michael Kohlert

Process Monitoring Algorithms and Software –

Basic Workflow

4.

Evaluate Model

Fit Model

Choose Algorithm

Preprocess Data

Choose Model

Make Predictions

Dr. Michael Kohlert

Process Monitoring Algorithms and Software -Application

Sensor Data

(10-100 /plant)

Quality State

Predict current machine states during operation

Recycling of old data for new models

Train Prediction Model

(historic data)Prediction

Model

Sensor data

(now)

Predicted State

(now)

update ~ 60-90

min.

State is: not okState is: ok

Prediction

Model

Dr. Michael Kohlert

Next Step in understandable Visualization

▪ Reduction of information to understandable level (1, 2, 3 dimensions)

▪ Visualization in real-timeVisualization

Up to 200 parameters in one point [temperature, pressure, speed,…]

Acquired per minute

Stored on an Oracle database

Processed for visualization in lower dimensions

Processing

Dr. Michael Kohlert

Good ProductNext Step in using Prediction Methods

Machine

PLC

Support Vector Machine

Acquisition

Loop

▪ Acquisition

▪ Pre-Processing (ETL)

▪ Machine Learning Methods/ Models

▪ Visualization On-Line/ Off-Line

Quality Data (OCS, Laboratory)

MES, SAP

Processing

Dr. Michael Kohlert

State is: not okState is: ok

Process Monitoring Algorithms and Software -Application

Fehlerrate

Abgleichsrate

14.85 %

0 %

Dr. Michael Kohlert

Processing Loop

▪ Open Processing Loop

▪ Recommendation System

36Dr. Michael Kohlert 18/12/2017

Condition Monitoring: Human-Machine-Interface 4.0

Open Loop Monitoring

Good ProductNext Step in Software Development

Acquisition

Loop

▪ Acquisition of more Datasets

▪ Pre-Processing (ETL)

▪ Extended Machine Learning Methods/ Models

▪ Version 3.0 of Visualization On-Line/ Off-Line

Version 1.0 (Dr. Kohlert, Prof. König) Version 3.0 (Mathworks)

Processing

Dr. Michael Kohlert

3818/12/2017

Vorhersage

95%

● Trajektorie

3%

2%

● Schulabschluss

● Region ● Ausbildung

● Firma

Dr. Michael Kohlert

3918/12/2017



TasksInformation über

Prozessverbesserungen

STEP2-Empfehlungswesen

TasksMaschine steuert Problemenentgegen und zeigt Fehler

selbstständig an

STEP3-Selbststeuerung


Dr. Michael Kohlert

Problem Statement & Scope Objectives & Expected Outcome

Problem Statement:

Fehlenden Anbindung von Qualitätsdaten an Datenbank

Roadmap für Anbindung

Objectives and Outcome:

Anbindung aller QS Geräte an lokale Oracle Datenbank

- Phase 1: Automatisierte Reports aus Qualitäts- und Maschinendaten

Phase 2: Entwicklung eines Prototyp einer selbststeuernden Maschine

2018

Setup Timeline

Project Titel: Autonome Maschine

Project No: 17-0001-IT-K

Project

Type:Express

Project x Project Lead: Michael Kohlert

Project Charter Overview

Dr. Michael Kohlert

41

Industry 4.0: Artificial Intelligence

Image: Baxter AI

Image: Baxter AI

Classical Automation & Artificial Intelligence = Advanced Automation

260

155

121

106

91

83

77

65

60

33

28

15

14

0 50 100 150 200 250 300

Machine learning (applications)

Natural language processing

Machine learning (general)

Computer vision (general)

Virtual personal assistants

Computer vision (applications)

Speech recognition

Smart robots

Recommendation engines

Gesture control

Context aware computing

Speech to speech translation

Video content recognition

Number of startup companies

Number of startup companies working in the artificial intelligence (AI)

market worldwide, as of March 2016, by category

Source: Statista estimates;

Medium


Renewable Energy LawEnergy Net Fee

Cost Reduction4

42

2

3

4

1

Big Data1

Machine Learning

Modell

2

Prediction3

2017

kW

kW

2017

Industry 4.0: Big Data - Energy Efficiency


43

2008

1st Industrial

Revolution

Increasing use of Steam

& Water Power.

Factories rising.

2010

3rd Industrial

Revolution

Programmable Logic

Controller, Electronics,

Automation.

2011

4th Industrial

Revolution

Digitalization, Cyber-

Physical Production

Systems.

Industry 5.0…. ?

2040

5th Industrial

Revolution

Self-Programming Units,

Artificial Surrogate

Friends & Managers

2016196918701784

Industry 4.0: Timeline Manufacturing

2nd Industrial

Revolution

Electricity, Assembly

Line, Automobiles, Tele-

communication, global

Transportation Systems.


44

Industry 4.0: Today … ?

Your new Target with Big Data … ?Today, solving daily issues

in production facilities …

Ideas …

Convince others about your

Methods …

Trial & Error …Find the Path …


45

Predictive MaintenanceTechnical Services

Virtualized Storage SystemLogistics

Data Analysis SkillsHR

Mobile Device ManagementIT

Condition MonitoringProduction

IT Strategy: Investment Focus


● ISO 27001

● IT Security Responsible

● Plant Maintenance Software

● KPI App/ OsiSoft PI

● Forklifter System

● Remote Data Glasses

● Six Sigma Training

● Digital Invoices

CONCEPT

46

Dr. Michael Kohlert

Head of IT & Automation

Mondi Gronau GmbH

Jöbkesweg 11, 48599 Gronau, Germany

Tel: +49 (0) 2562 919-665

Email: [email protected]

mailto:[email protected]

TU Kaiserslautern Lehrstuhl integrierte Sensorsysteme · beim Kunden zu einem Mitarbeiter im Unternehmen. Datenvisualisierung Produktionsanlagen Visualisierung wichtiger Daten einer

Documents