An improved diagnostic methodology for rotor electrical unbalance fault and its effects on power converters in DFIGs using RLMD-based signal decomposition

Abstract Accurate detection of rotor electrical unbalance (REU) is essential for ensuring the reliability of doubly-fed induction generators (DFIGs), as REU not only distorts electromechanical performance but also propagates harmful harmonics into the rotor-side and grid-side converters, increasing current stress and accelerating semiconductor ageing. However, REU-induced sidebands are typically weak and difficult to detect under the non-stationary operating conditions of wind turbines (WTs) using conventional spectral tools. This paper proposes an enhanced diagnostic framework based on advanced signal decomposition to identify REU and assess its impact on converter operation. Fast Fourier transform (FFT), empirical mode decomposition (EMD), and robust local mean decomposition (RLMD) are comparatively evaluated using stator current measurements from a 30 kW DFIG experimental platform. The results demonstrate that RLMD provides superior separation of low-energy slip-dependent sidebands and improved visibility of converter-coupled harmonics compared to FFT and EMD. Quantitatively, the fault energy ratio (FER) increases monotonically from 5.6874 in the healthy condition to 7.7873 at 300% REU (≈ +37%), confirming its robustness as a severity indicator. RLMD further enables stable extraction of modulation features and early-stage fault signatures that remain partially masked with conventional approaches. Overall, the findings establish RLMD-based decomposition as a robust, cost-effective, and diagnostically sensitive solution for REU detection and converter health assessment, with strong potential for real-time integration in WT condition monitoring systems.