Hypermodeling strategies for operational optimization

HYMOP aims to optimize the operation and maintenance of a fleet of industrial machines by means of innovative modeling and data processing/analysis techniques.

Summary

A collective challenge put forward by a number of Belgian companies that were brought together in the user groups of Sirris’ VIS-Elucidata and OWI-Lab projects is optimizing the operation and maintenance of a fleet of industrial machines that are being monitored by multiple sensors. To address this challenge, Sirris and Vrije Universiteit Brussel initiated the HYMOP project in collaboration with iMinds, Universiteit Antwerpen and KU Leuven.

Connected machines, equipped with several sensors, generate several data streams capturing measurements such as operational temperature, vibrations, pressure, etc. depending on the machine. In addition, machines are rarely unique, and a fleet can be defined as a set of comparable, yet potentially distributed machines that should exhibit, to a certain extent, similar behavior in terms of internal operation, application and usage. These assumptions about a fleet can be leveraged to derive insights about normal and anomalous behavior.

To work towards tangible results, an interdisciplinary research consortium with complementary expertise in system identification and monitoring, operation and maintenance, wind energy, data mining, decision support and scalable data processing has been set up. This research team can rely on the generous support of a large Industrial Advisory Committee.

The HYMOP partners will perform research in different relevant areas:

  • SIRRIS investigates hypermodeling as an approach for knowledge-driven and data-driven modeling with the goal of anomaly detection and failure prognosis, and how this approach can be deployed in a scalable analytics architecture.
  • Vrije Universiteit Brussel focuses on the evaluation and validation of the data-driven solutions for anomaly detection, prognosis and decision support.
  • iMinds investigates methods for anomaly detection and failure prognosis, leveraging the hypermodeling methodology. In addition, iMinds also works on semantic decision support.
  • Universiteit Antwerpen focuses on the pre-processing of the data by means of data cleansing algorithms as to get sufficient data quality for later modeling.
  • KU Leuven spearheads the research on data-driven model construction, which will be embedded in the hypermodeling approach, complementing the knowledge-driven models.

Research team

Research Institute Research Group Research Responsible
VUB AVRG Prof. C. Devriendt, Prof. P. Gauillaume, Dr. J. Helsen
SIRRIS Data Innovation Dr. E. Tsiporkova, Dr. T. Tourwé
Sustainability S. Milis
iMinds IBCN Dr. D. Deschrijver
MMLab Prof. S. Van Hoecke
UA AdRem Prof. B. Goethals
KULeuven DTAI Prof. L. De Raedt, Prof. J. Davis
     

Industrial Advisory Committee

Atlas Copco Laborelec H. Essers Fluxys
BASF Storck Trendminer Yazzoom
Maintenance partners Parkwind C-Power Xpert Services
Skyline communications      

Interested companies can still join the Industrial Advisory Committee.

Support

With the kind support of IWT, Strategisch Basisonderzoek

Agentschap voor Innovatie door Wetenschap en Technologie