Polar Diagram Generation

Sailing boats are using the power of the wind to travel through water.
During racing all boats have the same wind conditions and the winning team is the one that uses that power best to reach the finish line.
This competence is also desired aside of competitions during navigation.

The usable part of the wind’s power for driving is dependent on wind speed and its relative angle to the boat.
This relationship can be plot into performance polar diagrams.
Each combination of sails has its own characteristic polar diagram.
There are sails that are fast when sailing downwind but can hardly be used against the wind.

The currently favored method to forecast speed from wind conditions are velocity prediction programs.
They rely on mathematical models that are based on measurements of the boat and all appendages.

This master’s thesis discusses the disadvantages of this method and develops an alternative that allows the generation of polar diagrams on sailing boats without additional effort besides normal sailing.
The required measurements are already taken on most sailing boats and a software was developed that can be linked to the recorded data of the standardized communication on a boat.


It uses data analysis to find a hypothesis that represents the underlying data.
To solve this problem, a regression method that support measurement errors of all features and modeling of non-linear data was researched.
Additional hypothesis were raised to add the possible influence as a time series by weighting each data point by an established quality function.
Out of all raised hypothesis, the best is selected utilizing techniques from machine learning.

The resulting performance polar diagram has similar shape as the corresponding velocity prediction program.
Indeed it indicates slower boat speeds in general.
This traces back that the resulting performance polar diagram represents how the crew applied the potentiality of the boat and reveals deficiencies especially on lower wind speeds.

The algorithm is ready to be validated with more varying sails and types of boats in further research.