251 / 308
251
Automation and Information Systems
From Big Data to Smart Data
To gain value from the enormous amount of data recorded
during engineering and operation of industrial systems,
an increasing part of the research conducted at AIS is
dedicated to the transformation of big data to smart data.
Within this scope, the chair AIS applies a variety of data
analysis algorithms to data gathered from manufacturing
industry (EU project IMPROVE) and process industry (BMWi
Project SIDAP) as well as other important industries such
as semiconductor manufacturing. Within these projects,
the data acquisition, curation and preprocessing often
becomes as important as the machine learning methods.
For data aggregation and integration as challenging parts
of data collection, architectural solutions are elaborated,
which enable a flexible acquisition of data from a variety
of sources within CPPS. In the phase of data preparation,
AIS develops routines for the evaluation of data quality
and standardized procedures for data preparation itself.
Amongst others, methods based on conditional proba-
bility are developed in order to replace missing values in
incomplete datasets. Furthermore automatic outlier detec-
tion methods are introduced using statistical approaches.
Another important preprocessing step is the automatic
classification of process data into operation phases,
taking into account the different behavior of production
processes and plants during these phases. Several
methods e.g. k-means clustering or self-organizing maps
systems (cf. CRC 768 Subproject A6) and corresponding
tool environments are developed, extended, and adapted
(cf. CRC 768 Transfer Project T3). Model transformation
and formal methods for consistency checking (cf. CRC 768
Subproject D1) are applied. In addition, being a member of
the coordination board of the Priority Program (PP) 1593,
research in the field of cyber-physical (production) sys-
tems’ evolution management is conducted together with
German institutes in the field of software engineering. Spe-
cial focus is, among others, put on regression verification
(cf. PP 1593 ImproveAPS) as well as estimation of main-
tainability effort (cf. PP 1593 DoMain). By means of the
foundational research results in the field of model-based
systems engineering (CRC 768) and formal verification
of variant-rich systems (PP 1593), the foundational basis
to improve forever-evolving cyber-physical (production)
systems’ engineering and operation is laid. Aside from
these research projects, one of AIS’ main objectives is to
rapidly apply the fundamental results for industrial settings,
thereby collaborate with German SMEs to improve the
engineering process. Therefore, five novel application and
industry projects were set up in the field of model-based
testing of variant-rich systems, industrial communication
architectures and their modeling to estimate time and
safety aspects in trains, automated configuration and code
generation for process engineering systems, model-based
testing of safety functions, as well as semantic code anal-
ysis (cf. industry project semantic code analysis), building
on the successful results of the application and research
projects that were completed recently.
Projects
DFG Projects
■■
Spokeswoman CRC 768 ‘Managing Cycles in Inno-
vation Processes – Integrated Development of Prod-
uct-Service Systems Based on Technical Products’
■■
Regression Verification in a User-Centered Software
Development Process for Evolving Automated Produc-
tion Systems
■■
Domain-spanning Maintainability Estimation
■■
Diagnosis and Resolution of Inconsistencies in Hetero-
geneous Models
■■
Decision Making Support in Innovation Processes
under Consideration of the Technical Disciplines
■■
Reverse Engineering Design of Software Product Lines
for Automation Technology
AIF Projects
■■
Variabilität und Versionierung bei der anforderungsbasi-
erten Testfallentwicklung und -auswahl für mechatron-
ische Systeme
■■
Modellsynthese aus sequenzbasierten Verhaltensan-
forderungen zur modellbasierten Testfallgenerierung
VDI/VDE Projects
■■
Automatic configuration and generation of control
codes and visualizations for production plants in
process technology
■■
Train-wide availability of local sensor data
■■
Speaker of GMA Fachausschuss 5.15 ‘Agents in
automation’
BFS Project
■■
Effiziente Fehlersuche für sichere variantenreichen
Maschinen- und Anlagenautomatisierung
Industry Project
■■
Semantic Codeanalysis for IEC 61131-3 Applications
and Libraries