power curve eliminating outlier

Hello all,

I have been trying to plot the power curve of wind turbine. And I tried to eliminate the outliers using ignore if the value is greater than mean+standard deviation or less than mean-standard deviation, however this still does not work. Would please suggest me if there is any other method or how to improve my script? please have a look on the power curve attached with this message.

Thank you for your help!

Teklehaimanot

Answer 1

[Star Strider]

I am not certain what outliers you want to eliminate. I would do a non-linear regression on your data (using nlinfit or lsqcurvefit) and fit your data with a Logistic function.

[Teklehaimanot]

Thank you Star, As you might have seen the figure I uploaded, there are some data outside the thick one, which I want to eliminate.

[Star Strider]

My pleasure.

You can safely eliminate the most extreme of the outliers by using the Statistics Toolbox functions nlinfit (fitting the logistic function to your data) and nlpredci to define the prediction error intervals for generated power for each value of wind speed. Delete the values that lie outside the 95% confidence intervals.

[Teklehaimanot]

I appreciate your kind help and thank you again Star.

[Star Strider]

My pleasure, as always.

[Paolo Erazo Huera]

were you able to filter the erroneous data from the power curve?