Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
Institute of Industrial Automation and Software
Engineering
Research andTeaching at IAS
2021
Prof. Dr.-Ing.
Dr. h. c.
Michael Weyrich
2021University of Stuttgart, IAS, Prof. Dr.-Ing. Dr. h. c. Michael Weyrich 2
Faculty of Computer Science, Electrical Engineering and Information Technology of the
University of Stuttgart
Institute of Industrial Automation and Software Engineering (IAS)
We see ourselves as a bridgehead to Product and Plant
Automation in the research disciplines of Information Technology,
Software Technology and Electronics.
Prof Weyrich was appointed to the University of
Stuttgart in April 2013.
Research and teaching at the Institute focuses on the topic of Software Systems for
Automation Engineering.
2021University of Stuttgart, IAS, Prof. Dr.-Ing. Dr. h. c. Michael Weyrich 3
1935 – 1970
Institute for Electrical
Plants
Professor A. Leonhard
1970 – 1995
Institute for Control
Engineering and Process
Automation
Professor R. Lauber
1995 – 2015
Institute for Automation
and Software
Engineering
Professor P. Göhner
seit 2013
Professor M. Weyrich
Institute of Industrial
Automation and Software
Engineering
85 years of Tradition and Progress
Institute of Industrial Automation and Software Engineering (IAS)
Faculty of Computer Science, Electrical Engineering and Information Technology of the
University of Stuttgart
17x P
hD
stu
dents
3x
Managm
ent
Prof. Weyrich Dr. Jazdi Jun.-Prof. (TT) Morozov
6x
Serv
ice
30 –
40 x
HiW
is
2021University of Stuttgart, IAS, Prof. Dr.-Ing. Dr. h. c. Michael Weyrich 4
The institute conducts about 1000 exams per year.
Teaching
Study programs
- Electrical Engineering department:
• B. Sc. & M. Sc. Elektrotechnik und Informationstechnik
• B. Sc. Erneuerbare Energien
• M. Sc. Nachhaltige Elektrische Energieversorgung,
• M. Sc. Elektromobilität
• M. Sc. Information Technology
- Exports to other departments
• Mechatronik, Technische Kybernetik, Informatik, Medizin-
technik, Technikpädagogik, Verkehrsingenieurwesen
- Interdisciplinary
• M. Sc. Autonome Systeme (Dean of Studies Office)
Lectures
Industrial Automation I & II
Technologies and Methodologies of Software Systems I & II
Software Engineering for Real-Time Systems
Industrial Automation Systems
Basics of Software Systems
Lecture Series: Software and Automation
Lecture Series: Aspects of Autonomous Systems
Reliability of intelligent distributed Automation Systems
Networked automation systems (planned WS21/22)
Seminar Intelligent Cyber-Physical Systems
Laboratory Course Software Engineering
Laboratory Course Industrial Automation
Laboratory Introduction in Microcontroller Programming
2021University of Stuttgart, IAS, Prof. Dr.-Ing. Dr. h. c. Michael Weyrich 5
We focus on automation systems, especially their software in connection with control
systems.
Research at IAS
• 5G based Intelligent Digital Twin
• Multidimensional Synchronization of Digital
Twins
• Co-simulation of software-defined systems
• Automatic reconfiguration management
• Artificial intelligence and dynamic reliability
• Decentralized, cooperative machine learning
in automation
• Dynamic reliability calculation
• Test of distributed components
• Safeguarding of autonomous systems
Complexity
control in
automation
technology
Intelligent
Automation and
Autonomous
Systems
Safeguarding of
automation
systems and
components
Research topics at the IAS
• 5G-based Intelligent Digital Twin
• Co-simulation of software-defined
automated systems
• Autonomous reconfiguration
management of software-defined
systems
• Multidimensional synchronization of
digital twins for different applications
2021University of Stuttgart, IAS, Prof. Dr.-Ing. Dr. h. c. Michael Weyrich 6
How can the complexity of cyber-physical systems be made manageable in engineering
and operation?
Research Area: Complexity control in automation technology
Research topics at the IAS
• Dynamic reliability calculation
• Testing of distributed components
• Verification and validation of
software updates
• Safeguarding of autonomous
systems
• Cognitive sensor networks in safety-
relevant systems
2021University of Stuttgart, IAS, Prof. Dr.-Ing. Dr. h. c. Michael Weyrich 7
How can we rely on the quality of automated systems in terms of reliability, safety and
availability?
Research Area: Safeguarding of automation systems and components
Research topics at the IAS
• Dynamic Intelligent Reliability
• Optimization of automation systems
using machine learning
• Intelligent automation for user-oriented
• Soft sensors for networked automation
architectures
• Decentralized, cooperative machine
learning in automation
• Simulation of autonomy concepts
2021University of Stuttgart, IAS, Prof. Dr.-Ing. Dr. h. c. Michael Weyrich 8
Are technical systems of tomorrow going to automate themselves?
Research Area: Intelligent Automation and Autonomous Systems
2021University of Stuttgart, IAS, Prof. Dr.-Ing. Dr. h. c. Michael Weyrich 9
The Institute follows the mission statement "Intelligent Systems for a Sustainable
Society" and is part of the Excellence Strategy of the University of Stuttgart.
IAS in the Research Environment of Stuttgart
Institute of Industrial Automation
and Software Engineering
We are part of the profile areas and
emerging fields of the excellence strategy:
• Autonomous systems
• Architecture and adaptive building
• Production technologies
• BMWi flagship initiative: SofDCar
• BMBF flagship initiative: H2Mare
Research Factory
Intelligent Systems
Technology transfer
Industry 4.0
Assembly plant
Data Analytics in
Manufacturing
TestIAS with
Virtual Reality
Intelligent Test
of autonomous
systems
Inte
llig
en
t A
uto
ma
tio
n &
Au
ton
om
ou
sS
ys
tem
s
Co
mp
lex
ity c
on
tro
l o
f C
PS
Re
lia
bil
ity o
f A
uto
ma
tio
n S
ys
tem
s
Simulation of
plant
modernization
Digital Truck
Twin
Plug & Simulate Blockchain
Application
scenarios
Intelligent
Warehouse
(ARENA 2036)
2021University of Stuttgart, IAS, Prof. Dr.-Ing. Dr. h. c. Michael Weyrich 10
The model processes are used to represent special automation technology and to
demonstrate the capabilities of software systems.
Model Processes at IAS
Context aware pill
dispenser
Real-Time
Locating System
Scenario-based
testing of
autonomous
driving functions
Digital twin of a
modular
production
system
2021University of Stuttgart, IAS, Prof. Dr.-Ing. Dr. h. c. Michael Weyrich 11
Cooperation with the following Companies
CompWare Medical GmbH
Daimler AG
Diffblue Ltd.
Festo AG & Co. KG
Hirschvogel Umformtechnik GmbH
iss (Innovative Software Services GmbH)
OTTO FUCHS KG
Robert Bosch GmbH
Siemens AG
SMS group GmbH
Vector Consulting GmbH
Vector Informatik GmbH
Infineon Technologies AG
ZF Friedrichshafen AG
ETAS GmbH
Nokia Corporation
Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e. V.
Validation and verification of
highly automated and
autonomous systems
Indoor
Navigation
Systems
Simulation and commissioning
of robots in virtual reality
Create technologies that
combine power generation with
efficient control systems.
2021University of Stuttgart, IAS, Prof. Dr.-Ing. Dr. h. c. Michael Weyrich 12
IAS supports various start-up companies and cooperates in research projects
Maker Space
since 2021 VC (geplant)
Jan. 2017 – Dez. 2017 EXIST
Aug. 2019 – Juli 2022 EUREKA-Projekt
Apr. 2014 – März 2015 EXIST
März 2016 – Feb. 2018 Junge Innovatoren
Juni 2014 – Mai 2015 EXIST
Juni 2015 – Mai 2016 Junge Innovatoren
RoboTest
deaptwin
Machine parts detection
application for higher product
quality
since 2021 EXIST (geplant)
Prof. Weyrich is also the faculty's start-up officer and thus the first point of contact for those
interested in starting a business.
Co-Simulation (Cooperation with Vector Informatik)
Dynamic Co-Simulation of heterogeneous Internet of Things Systems
2020University of Stuttgart, IAS, Prof. Dr.-Ing. Dr. h. c. Michael Weyrich 14
Requirements:
„Plug-and-Simulate“-capable Co-Simulation of
heterogeneous and dynamically changing IoT
systems
Core technologies:
Agent-based Co-Simulation
Component- and Process modeling with MATLAB,
AnyLogic, Unity, OMNet++, …
Approach
Framework for coupling simulations via an agent
system
Connection of the simulations via interface
adapters
Service-oriented modeling of communication and
physical processes
Synchronization of the partial simulations via a
central clock agent
Agent
Agent
Agent
AnyLogic
MATLAB
OMNet++
Unity
How can
different
simulations be
coupled at
runtime?
𝑋
𝑈
( 𝑢1, 𝑥1 , 𝑦1)
( 𝑢2, 𝑥2 , 𝑦2)
Process 1
Process Chain
Process 3
Process 2
𝑢(𝑠)𝐺(𝑠)
𝑞(𝑠)
𝑒(𝑠)
𝐺(𝑠)
Component-related discretization
and storage of all time series
𝑌
Training the
network on
historical data
1 2 Using the
trained
model for
more robust
intervention
Methodology
2020University of Stuttgart, IAS, Prof. Dr.-Ing. Dr. h. c. Michael Weyrich 15
© 2017 IAS, Univ. Stuttgart
Requirements:
Analysis of process data for compliance with
defined quality characteristics
Real-time recommendations to the worker
Core technologies:
PLC-based data acquisition
Feature extraction
Data Analytics (online/offline)
Motivation
Sensor data contains information about the plant
and process status and can be used to improve the
process quality
Approach
Systematic extraction of unknown relationships and
patterns
Data acquisition and integration, dimension
reduction, data analysis, recommendations
Data-driven quality optimization
Adaptable quality controlControl of discrete manufacturing machines based on Long Short-Term Memory networks
How can we
improve the
product quality
by a learning
control?
Approach
Planning of an industrial automation system is
modelled as a dialog-based process and applied
to an agent system
Agents represent resources and try to integrate
them into the planned automation system
Determination of possible constellations for the
automation system to be planned
2020University of Stuttgart, IAS, Prof. Dr.-Ing. Dr. h. c. Michael Weyrich 16
Requirements:
Support of the planner in the rough planning
phase of industrial automation systems
Core technologies:
Agent technology (Self-organisation)
Metaheuristics (layout optimization)
Self-organized reconfiguration managementDecentralized, self-organized planning of automation systems
ID 577
ID 577
How can we
achieve autonomous
reconfiguration
management?
2021University of Stuttgart, IAS, Prof. Dr.-Ing. Dr. h. c. Michael Weyrich 17
© 2017 IAS, Univ. Stuttgart
Requirements:
Analyze decisions in terms of their risk at runtime.
Evaluate outcome and basis for decision-making
(models)
Core technologies:
Digital Twin
Automatic rule extraction
Data Analytics
Safeguarding of autonomous systems in operationinterpretable safety assessment of dynamic decision processes at runtime
How can the risk
of a decision be
quantified at
runtime?
Is my
route
safe?
Motivation
Avoid critical situations proactively
Operational data contains information about the
uncertainties inherent in the models.
Approach
Systematic extraction of unknown influences through
hypothesis generation and their testing.
Trade-off between safety and efficiency based on
estimated uncertainty in planning
Data-driven optimization of the safety margin
New
2021University of Stuttgart, IAS, Prof. Dr.-Ing. Dr. h. c. Michael Weyrich 18
© 2017 IAS, Univ. Stuttgart
Requirements:
Automated increase of test case coverage of
trained models
Generated test cases are relevant and
challenging for the trained model
Core technologies:
Machine Learning
Parameter dependent environment simulations
Numerical optimization methods
Motivation
Human experts can poorly assess which concrete
test cases are challenging for a machine-trained
model
Approach
Parameterized environment simulation of a typical
deployment scenario
Iterative optimization of simulation parameters to
generate challenging test cases
Automated test case generation for machine-trained modelsGeneration of new challenging test cases by optimization methods
How can
machine-trained
models be
effectively
tested?
New
2021University of Stuttgart, IAS, Prof. Dr.-Ing. Dr. h. c. Michael Weyrich 19
Context-aware cyber-physical systems using the example of adaptive user support in the health sector
© 2017 IAS, Univ. Stuttgart
Approach
Acquisition of heterogeneous and dynamic data for
context modeling (environment parameters, user
data, etc.)
Middleware-based architecture for scalable reuse
of the context model by applications.
Use case: development of a tablet dispenser as an
IoT approach with context-adaptive user support
Requirements:
Acquisition, modeling, and analysis of
heterogeneous data around a cyber-physical
system to derive applicable knowledge.
Core technologies:
Context-aware automation systems
Middleware-based architecture
Graph algorithms for consistency checking of
time-based context models
How can pill-taking be
supported in a context-
sensitive way?Mobiler und stationärer Tablettendispenser
Adaptive Assistenz
Heterogene und
dynamische Daten
Kontext Middleware
Intelligente Kontextgenerierung
Kontext-Modellierung
Kontext-Reasoning
Kontextverwaltung
Datenschicht
Anwendungsschicht
Datenerfassung- und vorbereitung
Adaptive AlarmierungEmpfehlungen zur Adhärenz -
Erhöhung
Alarmeinstellung
Umgebungsdaten
Gesundheitsdaten
Profildaten
Einnahmezeiten
Medikationsplan
New
2021University of Stuttgart, IAS, Prof. Dr.-Ing. Dr. h. c. Michael Weyrich 20
Requirements:
Handling of input data (usually time series data) of
different dimensions
Solution of regression and classification problems
Core technologies:
Two-stage deep neural network algorithm
Representation database for storing and
exchanging characteristic feature sets
Client-to-client communication architecture
Motivation
Efficient learning despite
data sets that are often small in everyday
industrial life
dynamic processes that require continuous
updates of the learning model.
Approach
Transfer of knowledge between algorithms
that are able to learn
Deep Industrial Transfer LearningLearning automation systems with dynamic knowledge transfer
New
AI for intelligent testing of autonomous systemsValidation and verification of autonomous systems and their components
2020University of Stuttgart, IAS, Prof. Dr.-Ing. Dr. h. c. Michael Weyrich 21
Requirements:
Manufacturers announce concepts for
autonomous systems, but tools for verification
and validation are needed
Core technologies:
Computational Intelligence
Probabilistic methods (e.g. Bayesian networks)
Approach
Test cases are grouped with AI
Deep Rule learning for transparent rules
Prioritization of test cases
How can
autonomous
systems be
tested
intelligently?
2021University of Stuttgart, IAS, Prof. Dr.-Ing. Dr. h. c. Michael Weyrich 22
© 2017 IAS, Univ. Stuttgart
Requirements:
Complexity control in variant management
Assistance for the system engineer
Core technologies:
Feature Oriented Domain Analysis
150%-Modell
Data Analytics
Motivation
As systems become more complex, the solution
space for system development becomes
unmanageable
Approach
Model repository for known relationships and
patterns
Use of an appropriate methodology for variant
management
Model and data based variant generation
Automated variant managementGeneration of system architectures in systems engineering
New
How to create
approximately all
variants of a
system and select
one?
2020University of Stuttgart, IAS, Prof. Dr.-Ing. Dr. h. c. Michael Weyrich 23
Assistance of modernization in industrial automationHow can we support the modernization of old plants?
© 2017 IAS, Univ. Stuttgart
Approach
Modeling of the machine with little effort, software
agents represent components of the machine
Automated evaluation of production requests
Generation and evaluation of adaptation options
using the model of the machine
Decision support and assistance in the
modernization planning process
Requirements:
Software tool which supports the modernization
process of machines or plants
Generation and evaluation of adaptation options
based on given requirements
Core technologies:
Product-, process-, resource-model of the
machine
Agent-based assistance system
How can we
support the
modernization of
old plants?
2021University of Stuttgart, IAS, Prof. Dr.-Ing. Dr. h. c. Michael Weyrich 24
© 2017 IAS, Univ. Stuttgart
Requirements:
Integration and uniform semantic description of
heterogeneous data
Generation of knowledge for the application from
unstructured and heterogeneous data
Core technologies:
Ontologies
AI methods (ensemble learning, feature extraction)
Graph Analytics
Motivation
Use of existing data diversity from different, dynamic
sources to generate knowledge for applications
Approach
Intelligent data integration from heterogeneous
sources
Connecting heterogeneous data, artificial
intelligence as well as analytical models
AI-based knowledge generation
Robust learning based on heterogeneous dataKnowledge generation in automation technology using data diversity
How can
knowledge be
generated from
unstructured,
heterogeneous
data?
Künstliche
Intelligenz
Anwendungen
Datenbank
Prozessdaten
Unstrukturierte
Dateien
Nutzer
Analytische
Modelle
New
web
phone +49 (0) 711 685-
fax +49 (0) 711 685-
University of Stuttgart
Thank you!
Pfaffenwaldring 47
70550 Stuttgart
Prof. Dr.-Ing. Dr. h. c. Michael Weyrich
67301
67302
Institut für Automatisierungstechnik und Softwaresysteme
Institute of Industrial Automation
and Software Engineering
www.ias.uni-stuttgart.de