Comparison of anomaly detection techniques for wind turbine gearbox SCADA data

Mckinnon, C. and Carroll, J. and McDonald, A. and Koukoura, S. and Soraghan, C. (2019) Comparison of anomaly detection techniques for wind turbine gearbox SCADA data. In: Wind Energy Science Conference 2019, 2019-06-17 - 2019-06-20, University College Cork.

[thumbnail of McKinnon-etal-WESC-2019-Comparison-of-anomaly-detection-techniques-for-wind-turbine-gearbox-SCADA-data]
Preview
 Text. Filename: McKinnon_etal_WESC_2019_Comparison_of_anomaly_detection_techniques_for_wind_turbine_gearbox_SCADA_data.pdf
Accepted Author Manuscript
Download (436kB)| Preview

Abstract

This analysis looks at the use of anomaly detection to assess the condition of wind turbine gearboxes based on data from a number of operational turbines. A comparison is made between various methods of anomaly detection, these being one class support vector machine (OCSVM), random forests, and nonlinear autoregressive neural networks with exogenous inputs (NARX).

ORCID iDs

Mckinnon, C., Carroll, J. ORCID logoORCID: https://orcid.org/0000-0002-1510-1416, McDonald, A. ORCID logoORCID: https://orcid.org/0000-0002-2238-3589, Koukoura, S. and Soraghan, C.;
CORE (COnnecting REpositories)