• Citer

Power-Electronic Converter and Sensor Fault Diagnosis for Doubly-Fed Induction Generator Wind Turbines

par Yahiaoui, Mohamed-Amine
Président du jury Henneaux, Pierre
Promoteur Gyselinck, Johan
Co-Promoteur Kinnaert, Michel
Publication Non publié, 2024-12-13

Thèse de doctorat

• Accès en ligne
• Détails
• Contenu
• Statistiques

Résumé :

Fossil fuel-based energy generation remains the dominant method worldwide, but it contributes heavily to greenhouse gas emissions and global warming. As a result, there has been significant growth in the integration of environmentally friendly renewable energy sources (RESs), with many countries aiming to raise the share of renewables in their electricity production. Among RESs, wind power has experienced the most significant expansion in both capacity and market share. Unlike conventional power plants, wind turbines (WTs) are typically installed in remote areas and offshore, which presents difficulties related to reliability and accessibility. Consequently, this results in high maintenance costs and more frequent downtimes. To address these challenges, condition monitoring and fault diagnosis are regarded as crucial solutions for improving reliability while lowering operating and maintenance costs.The reliability and availability of WTs having a doubly-fed induction generator (DFIG) are critically dependent on the performance of their power-electronic converter (PEC) and sensors, as these components have an essential share in failure rates and downtimes. Among PEC components, Insulated Gate Bipolar Transistors (IGBTs) are highly susceptible to both open-circuit and short-circuit faults. Open-circuit faults are less destructive but can cause significant long-term issues and shut down the WT system. Similarly, sensor faults can severely degrade WT performance and potentially result in complete system shutdowns.This thesis presents novel fault detection and isolation (FDI) methods that can diagnose IGBT open-circuit faults and sensor faults. The first FDI method is a single model-based approach that can diagnose both IGBT open-circuit faults and sensor faults. This FDI approach is based on an augmented state observer and combines a chi-square (χ^2 ) statistical test, adaptive thresholds, and normalized diagnostic variables, which improves the reliability of fault diagnosis in different WT operating modes, including sub-synchronous and super-synchronous. This single algorithm avoids the complexity and computational cost of applying separate diagnosis methods for different types of faults and thus enhances the overall efficiency and reliability of the WT system. The second FDI method is a robust signal-based FDI algorithm capable of diagnosing multiple IGBT open-circuit faults in both sub-synchronous and super-synchronous operating modes. The proposed FDI approach uses only rotor current measurements and references, thus avoiding the need for additional sensors, and combines a chi-square (χ^2 ) statistical test, normalized diagnostic variables, and adaptive thresholds to reduce the risk of false alarms and ensure accurate fault diagnosis regardless of the operating conditions. It is also able to diagnose and differentiate between IGBT and rotor current sensor faults.Furthermore, this thesis addresses critical challenges related to the grid synchronization of a DFIG, which is the start-up mode for connecting the stator winding of the DFIG to the grid. This synchronization process is essential to prevent excessive mechanical and electrical stress due to stator current transients. Traditionally, this process involves stator and grid voltage measurements, but to improve system reliability and reduce costs, this research investigates the possibility of eliminating redundant stator voltage sensors by developing a sensorless vector control strategy to synchronize the DFIG with the grid. This method relies on an augmented state observer, which estimates the direct and quadrature components of the stator voltage in a synchronous reference frame using only rotor current measurements.