Methods for the predictive maintenance of offshore wind turbines

For the wind turbine manufacturer Senvion SE, research into comprehensive methods and tools for a predictive maintenance strategy for services was of great interest. Taking various data sources into account, SWMS contributed to a Data Analytics prototype.

Challenges

The aim of the project "preInO - Methods and tools for the predictive maintenance of offshore wind turbines" was the realization of a predictive maintenance strategy for offshore wind turbines using artificial intelligence and automatic self-organization. The challenge was to provide the best possible prediction of the condition of a component or system from different data sources. This knowledge served as a basis for the automatic prioritization of maintenance measures and, in addition, great optimization potentials could be uncovered with regard to the planning of the deployment of personnel, spare parts and means of transport.

Procedures

As part of the project, the maintenance processes were first recorded and the data sources required for automated decision support were identified. Based on this, a basic concept for predicting the system status was developed. The heart of the system is the so-called "Processing Engine", which was developed by SWMS as part of the project. Using this system, different algorithms can be linked to different data sources in order to carry out a precise analysis of the condition of the respective component. It is also possible to incorporate the results of an evaluation into another analysis.

Results

By using the developed tools and methods for a predictive maintenance strategy, it is possible to plan logistic processes better and earlier. Duration and execution can thus be shortened and costs can be saved. In order to prove the economic potential of the system, a simulation was developed within the framework of the project which depicts the real processes with all restrictions and in particular the weather restrictions. The results of the simulation show that, depending on the initial scenario, maintenance costs can be reduced by approx. 50% and at the same time technical availability and thus energy production can be increased.

"The Processing Engine makes it easy to perform complex analyses from different data sources. This made it possible to automatically assign a priority to the detected errors and thus optimize the execution of the maintenance measures", Stephan Oelker, BIBA Bremen.

Project Details

  preInO - sponsored by the Ministry of Economics and Energy
 Oldenburg, Bremen, Hamburg
 http://www.preino.de