Wind turbine fault estimation using sliding mode observer based on Takagi–Sugeno fuzzy model

Résumé

This paper presents a fault-estimation approach for utility-scale wind turbines that combines Takagi–Sugeno (TS) fuzzy modeling with a sliding-mode observer (SMO). The nonlinear dynamics of the 4.8 MW benchmark turbine are represented by a TS structure, enabling an LMI-based synthesis of a robust TS–SMO. The proposed observer reconstructs both actuator faults affecting generator torque and sensor faults in blade-pitch measurements. MATLAB/Simulink validations under realistic operating conditions (operating-point variations, wind fluctuations, and disturbances) demonstrate accurate tracking and fast, stable fault reconstruction over the complete simulation horizon. Performance is assessed using the Normalized Sum of Squared Errors (NSSE): the reconstructed faults exhibit low NSSE values in the considered fault scenarios, with the blade-pitch sensor fault achieving NSSE =0.087 %. These results indicate reliable fault estimation while maintaining bounded residuals and avoiding drift. The method relies on standard industrial signals and entails modest online computations (matrix operations and a bounded switching term), facilitating integration into existing condition-monitoring and fault-tolerant control architectures. Overall, TS-guided sliding-mode observation is shown to be an effective and robust solution for wind-turbine fault diagnosis under nonlinearities and exogenous perturbations.

Citer ce document

Accès au document

Texte intégral en lecture en ligne, réservé aux abonnés SPHAERO et aux membres de l'institution. Se connecter

Auteur(s)

Statistiques

Consultations : 1

Téléchargements : 0