QQ plot for assessment of Gaussian Process wind turbine power curve error distribution function

Pandit, Ravi and Infield, David; (2018) QQ plot for assessment of Gaussian Process wind turbine power curve error distribution function. In: 9th European Workshop on Structural Health Monitoring. British Institute of Non-Destructive Testing, GBR. (In Press)

[thumbnail of Pandit-Infield-EWSHM-2018-QQ-plot-for-assessment-of-Gaussian-Process-wind-turbine-power-curve]
Text. Filename: Pandit_Infield_EWSHM_2018_QQ_plot_for_assessment_of_Gaussian_Process_wind_turbine_power_curve.pdf
Final Published Version
License: Creative Commons Attribution-NonCommercial 4.0 logo

Download (546kB)| Preview


Performance monitoring based on available SCADA data is a cost effective approach for condition monitoring of a wind turbine. Performance is conventionally assessed in terms of the wind turbine power curve that represents the relationship between the generated power and hub height wind speed. Power curves also play a vital role in energy assessment, and performance and warranty formulations. It is considered a most important curve for analyzing turbine performance and also helps in fault detection. Conventional power curves as defined in the IEC Standard take considerable time to establish and are far too slow to be used directly for condition monitoring purposes. To help deal with this issue the Gaussian process (GP) concept is introduced. A Gaussian process (GP) is a nonlinear machine learning technique useful in interpolation, forecasting and prediction. The accuracy of fault identification based on a GP model, depends on its error distribution function. A QQ plot is a useful tool to analyze how well given data follows a specific distribution function. The objective of this paper is to apply QQ plots in the assessment of the error distribution function for a GP model. The paper will outline the advantages and limitations of the QQ plot approach.