Innovative experimental approach for the early detection of rotor winding asymmetry in wound-rotor induction machines

• Develop an innovative electrical residual signal extraction (RESE) method, inspired by mechanical vibration analysis, for the early detection of rotor winding asymmetry faults. • Establish the first direct analogy in the scientific literature between mechanical dynamic signatures and harmonic variations of electrical currents in wound rotor machines. • Surpass conventional techniques (MCSA) by tracking the dynamic evolution of harmonics, thus offering greater diagnostic precision and responsiveness. • Demonstrate the application potential on squirrel-cage generators and high-capacity wind turbines, helping to optimize reliability in renewable energies. • Strengthen asset management in the renewable energies sector through early and reliable fault detection, reducing costs associated with breakdowns and corrective maintenance. • Contribute to applied research on wound rotor induction generators (DFIG), essential in modern high-efficiency wind power plants. The monitoring method proposed in this article is aimed at diagnosing Rotor Winding Asymmetry (RWA) faults in Wound Rotor Induction Machines (WRIM). Inspired by the principles of mechanical vibration analysis, this approach is based on the Residual Electrical Signal Extraction (RESE) strategy, which transposes the concepts of dynamic analysis to the electrical domain, exploiting the cycloergodicity of electrical signals rather than physical vibrations. The main challenge addressed by this method is the detection of faults, even at a very early stage, whether they appear as soon as the machine is commissioned or coexist with other internal conditions, even under light load. The RESE strategy has been validated by simulations and experimental tests on different configurations: healthy state, presence of defects, and several levels of severity. An IDSC (Incipient Defect Severity Criterion) indicator was introduced to quantify defect severity. The results show that the RESE strategy can effectively detect asymmetry defects of <10 % with a good level of reliability. The fact that the RESE strategy can be used off-line, without vibration sensors, simplifies its adoption in an industrial environment.