Optimizing data access for wind farm control over hierarchical communication networks

In this paper we investigate a centralized wind farm
controller which runs periodically. The controller attempts to
reduce the damage a wind turbine sustains during operation
by estimating fatigue based on the wind turbine state. The
investigation focuses on the impact of information access and
communication networks on the controller performance. We
start by investigating the effects of a communication network
that introduces delays in the information access for the central
controller. The control performance as measured by accumulated
fatigue is shown to be significantly impacted by communication
delays and also by the choice of the time instances at
which sensor information is accessed. In order to optimize
the latter, we introduce an information quality metric and a
mathematical model based on Markov chains, which are compared
performance-wise to a heuristic approach for finding this
parameter. This information quality metric is called mismatch
probability, mmPr, and is used to express quantitatively the
information accuracy in a given scenario. Lastly measurements
of different communication technologies have been performed
in order to carry out the analysis in a practically relevant
scenario with respect to the communication network delays.
These measurements are done in regards to packet loss and
communication delays, and the simulations are rerun using either
the traces from the measurements or scenarios constructed from
the delay parameters.