A Gaussian Process Based Approach to Estimate Wind Speed Using SCADA Measurements from a Wind Turbine. Issue 20 (2021)

• Record Type: Journal Article
• Authors: Paiva, Eduardo B.R.F.; Nguyen, Hoai-Nam; Lepreux, Olivier; Bresch-Pietri, Delphine
• Is Part Of: IFAC-PapersOnLine. Volume 54:Issue 20(2021)
• Page Start: 65
• Page End: 71
• Publication Date: 2021
• Subjects: Wind Speed Estimation -- Wind Energy -- Machine Learning -- Gaussian Process Regression -- Real-time; Automatic control -- Periodicals; 629.805
• DOI: 10.1016/j.ifacol.2021.11.154
• Languages: English
• ISSNs: 2405-8963

Abstract

Abstract: In this paper, we propose a method for estimating in real-time the speed of the wind to which a turbine is subjected using its SCADA (Supervisory Control And Data Acquisition) measurements. The approach is fully data-driven. It is based on Gaussian Process Regression. We use real experimental SCADA data from an operating commercial 3-bladed horizontal axis wind turbine. The reference values for the wind speed are obtained from a nacelle LiDAR (Light Distancing and Ranging) sensor. The comparison of the obtained estimation results with the measurements provided by the LiDAR sensor emphasizes the performance of the proposed method and underlines its interests for control purposes. Assessing performance on a day of operation, we obtain median errors of less than 1%. A numerical comparison with a more traditional model-based approach is also provided.