Navigation

WiSAbigdata

Title

Wind farm virtual Site Assistant for O&M decision support – advanced methods for big data analysis; Sub-project: Data-based methods for early error detection and method transfer in WiSA demonstrator as well as evaluation

Funding

BMWi (FKZ 03EE3016A)

Period

01.11.2019 - 30.10.2022

Project partners

  • Universität Oldenburg
    AG TWiSt (ForWind / Physics) Coordinator
    AG WESys (ForWind / Physics)
    AG KomplSyst (ForWind / ICBM)
    AG VLBA (Computer Science)
  • Fraunhofer-Institut für Windenergiesysteme (IWES), Hannover
  • Universität Duisburg-Essen - Fakultät für Physik, Duisburg
  • OFFIS e.V., Oldenburg
  • Ramboll Deutschland GmbH, Hamburg
  • Ocean Breeze Energy GmbH & Co. KG, Bremen
  • Deutsche Windtechnik X-Service GmbH, Osnabrück

Project objectives

In modern wind turbines, a vast amount of operational data is acquired with a high temporal resolution, and
developments in measurement engineering and digitalization will even increase the amount of data. This
data is currently only sparsely evaluated, and almost exclusively 10-minute average values are stored and
analyzed. Using high-frequency operational data instead constitutes a very promising approach to gain
further insights about the wind energy system. However, this potential comes with a significant demand for
research in order to explore and subsequently exploit the information contained in high frequency data.
The project “WiSA big data” aims to advance methods of early fault detection based on high-frequency data
to provide decision support for wind turbine maintenance planning and execution. Therefor established
methods used for 10-minute averaged operational data will be extended towards high-frequency
capabilities. Moreover, novel analysis methods, which have been established in other fields of research, will
be transferred and possibly combined to suite the demands of wind energy applications. All advanced and
tested methods will be subject to a quantitative comparison including a practical evaluation. Based on this
evaluation, an automatized selection of the best performing method for a respective use case will be
pursued in close cooperation with industrial users. A soft- and hardware platform set up as a core system
for WiSA is established to provide a common platform for management, analyses and evaluation of high
frequency data. Selected capable methods shall be implemented in a WiSA demonstrator and made
available for industrial testing as a means to enable preventive maintenance and detailed analyses of
operational data. The connection of the WiSA demonstrator with the core system shall allow for the future
integration of further innovative methods for early fault detection and thus enable a long-term usage of the
system.

 

Specific contributions and objectives WESys

  • Development and testing of the neural networks method for early fault detection
  • Development and testing of state identification on the base of Support Vector Machines and Local Outlier Factor
  • Evaluation of methods

Category

applied science

Webmastekqclfr/sn (martvvin.kracmft@uol.dhcue) (Changed: 2020-01-29)