Comparative analysis of binning and support vector regression for wind turbine rotor speed based power curve use in condition monitoring

Pandit, Ravi and Infield, David (2018) Comparative analysis of binning and support vector regression for wind turbine rotor speed based power curve use in condition monitoring. In: 2018 53rd International Universities Power Engineering Conference (UPEC). IEEE, Piscataway, NJ. ISBN 9781538629109

[img]
Preview
Text (Pandit-Infield-UPEC2018-Comparative-analysis-of-binning-and-support-vector-regression)
Pandit_Infield_UPEC2018_Comparative_analysis_of_binning_and_support_vector_regression.pdf
Accepted Author Manuscript

Download (726kB)| Preview

    Abstract

    Unscheduled maintenance consumes a lot of time and effort and hence reduces the overall cost-effectiveness of wind turbines. Supervisory control and data acquisition (SCADA) based condition monitoring is a cost-effective approach to carry out diagnosis and prognosis of faults and to provide performance assessment of a wind turbine. The rotor speed based power curve, which describes the nonlinear relationship between wind turbine rotor speed and power output, is useful for performance appraisal of a wind turbine though limited work on this area has been undertaken to date. Support Vector Machine (SVM) is a data-driven, nonparametric approach used for both classification and regression problems developed initially from statistical learning theory (SLT) by Vapnik. SVM is useful in forecasting and prediction applications. This paper deals with the application of support vector regression to estimate the rotor speed based power curve of a wind turbine and its usefulness in identifying potential faults. It is compared with a conventional approach based on a binned rotor speed power curve to identify operational anomalies. The comparative studies summaries the advantages and disadvantages of these techniques. SCADA data obtained from a healthy operational wind turbine is used to train and validate these methods.