Abnormal vibration detection of wind turbine based on temporal convolution network and multivariate coefficient of variation

A working wind turbine generates a large amount of multivariate time-series data, which contain abundant operation state information and can predict impending anomalies. The anomaly detection the nacelle that houses all generating components in have been challenging due to its inherent complexities, systematic oscillations noise. To address these problems, this paper proposes an unsupervised approach, combines deep learning with multi-parameter relative variability detection. normal behavior model (NBM) vibration is firstly built upon training historical data supervisory control acquisition (SCADA) system high-resolution domain. better capture temporal characteristics frequency signals, spectrum vector integrated as inputs spectrum-embedded convolutional network (SETCN) then used extract latent features. anomalies are detected through multi-variate coefficient variation (MCV) based assessment index (AAI) among residuals environment parameters nacelle. approach considers input preserves spatio-temporal correlation between variables. Validations using collected from real-world farms demonstrate effectiveness proposed approach.