PreCon Project Page


Predictive Maintenance and Conflict Handling

Data Acquisition and Deployment Support

Industrial Communications (Prof. Wollschlaeger)
In order to perform predicitive maintenance, a flexible concept for acquiring data from automation devices is needed. Identification data, configuration information, run-time data, and status information provide valuable sources for maintenance tasks. This research field aims at providing an information model for automation devices from the maintenance point of view:
  • flexible data acquisition methods using diverse base technologies (fieldbus, Industrial Ethernet, OPC, Web Services)
  • automated information retrieval for connected automation devices
  • development of an information model for maintenance information for automation devices

Manufacturing Data Management and Analysis

Database Technology Group (Prof. Lehner)
Condition data of machines is the basis for estimating failures. However, it is characterized by its large volume, contained noise, irregularities and dependence on the highly dynamic production environment, i.e. flexible manufacturing processes. Therefore, the research in this field aims at:
  • developing specialized compression and approximation procedures that retain the predictive accuraccy of the base data
  • extracting knowledge to improve failure estimation by developing data mining algoritms that perform well in the presence of noise and conceptual change induced from the environment as well as from aging

Modeling and Forecasting Machine Behavior

Technical Information Systems (Prof. Kabitzsch)
Data from industrial devices must be modeled in order to forecast their future behavior. Therefore, efficient prognoses strategies are needed including:
  • determination of characteristics in industrial devices that are relevant for state changes
  • modeling of machine behavior regarding future state changes
  • determination, classification and evaluation of machine states

Framework for Modeling Support

Software Technology (Prof. A▀mann)
Modeling and forecasting of data captured from shop floor environments is essential for applying predictive maintenance. For such purposes there are standard approaches as well as a large amount of different vendor specific modeling techniques that have to be integrated in the existing architecture. Research in this field is focused on a generic framework for modeling concerns with the following point of views:
  • definition of a layered framework architecture for different modeling concerns
  • semantic description of different modeling techniques
  • investigation of suitable strategies for framework tests

SOA in Manufacturing

Computer Networks (Prof. Schill)
Manufacturing environments are characterized by a large variety of heterogeneous devices, networks, specific protocols and applications. Lacking standardized interfaces there is still a gap between shopfloor systems and enterprise applications. Therefore, a service-oriented middleware as an integration layer between these two worlds is envisioned addressing:
  • abstraction from heterogenity exposed by enterprise and automation layer
  • definition and distribution of core service
  • flexible mapping to underlying communication technologies